1 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3 * doc/invoke.texi (ARM Options): Document accepted extension options
6 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
9 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
10 or --with-float=softfp, set the default CPU to arm10e.
12 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
14 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
15 * config/visium/predicates.md (const_shift_operand): Likewise.
16 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
17 (visium_legitimize_reload_address): Likewise.
19 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
24 2017-12-20 Jakub Jelinek <jakub@redhat.com>
27 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
29 2017-12-20 Julia Koval <julia.koval@intel.com>
31 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
32 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
33 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
34 * config.gcc: Include vpclmulqdqintrin.h.
35 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
36 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
38 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
39 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
40 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
41 * config/i386/i386.c (isa2_opts): Add -mcx16.
42 (isa_opts): Add -mpclmulqdq, remove -mcx16.
43 (ix86_option_override_internal): Move mcx16 to flags2.
44 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
45 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
46 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
47 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
48 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
49 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
50 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
51 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
52 * doc/invoke.texi: Add -mvpclmulqdq.
54 2017-12-20 Tom de Vries <tom@codesourcery.com>
57 * config/i386/i386.c (ix86_static_chain): Move DECL_STATIC_CHAIN test ...
58 * calls.c (rtx_for_static_chain): ... here. New function.
59 * calls.h (rtx_for_static_chain): Declare.
60 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
61 instead of targetm.calls.static_chain.
62 * df-scan.c (df_get_entry_block_def_set): Same.
64 2017-12-19 Tom de Vries <tom@codesourcery.com>
66 PR tree-optimization/83493
67 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
70 2017-12-19 Jakub Jelinek <jakub@redhat.com>
72 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
73 inform with hardcoded english plural handling.
75 2017-12-18 Jeff Law <law@redhat.com>
77 PR tree-optimization/83477
78 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
79 a non-virtual PHI, always push a new range.
81 2017-12-19 Martin Sebor <msebor@redhat.com>
84 * builtins.c (compute_objsize): Handle non-constant offsets.
86 2017-12-19 Jakub Jelinek <jakub@redhat.com>
88 PR tree-optimization/83444
89 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
90 character load case, if get_stridx on MEM_REF's operand doesn't
91 look usable, retry with get_addr_stridx.
93 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
96 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
97 (variable_tracking_main_1): Keep markers even when VTA fails.
100 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
101 even if there are markers before it.
102 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
104 2017-12-19 Jakub Jelinek <jakub@redhat.com>
106 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
107 typical order conditions.
108 * sel-sched.c (extract_new_fences_from): Likewise.
109 * config/visium/constraints.md (J, K, L): Likewise.
110 * config/visium/predicates.md (const_shift_operand): Likewise.
111 * config/visium/visium.c (visium_legitimize_address,
112 visium_legitimize_reload_address): Likewise.
113 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
114 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
115 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
116 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
117 SET_DUMP_DETAIL): Likewise.
118 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
119 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
120 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
121 avr_set_core_architecture, avr_set_current_function,
122 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
123 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
124 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
125 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
126 avr_map_decompose, avr_fold_builtin): Likewise.
127 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
128 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
129 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
130 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
131 * config/m32c/m32c.c (m32c_conditional_register_usage,
132 m32c_address_cost): Likewise.
133 * config/m32c/predicates.md (shiftcount_operand,
134 longshiftcount_operand): Likewise.
135 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
136 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
137 can_use_cdx_ldstw): Likewise.
138 * config/nios2/nios2.h (CDX_REG_P): Likewise.
139 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
141 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
142 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
143 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
144 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
145 * config/vax/vax.c (adjacent_operands_p): Likewise.
146 * config/ft32/constraints.md (L, b, KA): Likewise.
147 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
149 * cfgexpand.c (expand_stack_alignment): Likewise.
150 * gcse.c (insert_expr_in_table): Likewise.
151 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
152 * cgraphunit.c (cgraph_node::expand): Likewise.
153 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
154 * emit-rtl.c (add_insn): Likewise.
155 * input.c (dump_location_info): Likewise.
156 * passes.c (NEXT_PASS): Likewise.
157 * read-rtl-function.c (parse_note_insn_name,
158 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
160 * sched-rgn.c (sched_rgn_init): Likewise.
161 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
162 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
163 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
164 * explow.c (eliminate_constant_term): Likewise.
165 * final.c (leaf_renumber_regs_insn): Likewise.
166 * cfgrtl.c (print_rtl_with_bb): Likewise.
167 * genhooks.c (emit_init_macros): Likewise.
168 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
169 * tree-data-ref.c (conflict_fn): Likewise.
170 * selftest.c (assert_streq): Likewise.
171 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
172 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
173 fold_binary_loc, multiple_of_p): Likewise.
174 * reload.c (push_reload, find_equiv_reg): Likewise.
175 * et-forest.c (et_nca, et_below): Likewise.
176 * dbxout.c (dbxout_symbol_location): Likewise.
177 * reorg.c (relax_delay_slots): Likewise.
178 * dojump.c (do_compare_rtx_and_jump): Likewise.
179 * gengtype-parse.c (type): Likewise.
180 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
181 simplify_const_relational_operation): Likewise.
182 * reload1.c (do_output_reload): Likewise.
183 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
184 * gengtype.c (type_for_name): Likewise.
185 * gimple-ssa-sprintf.c (format_directive): Likewise.
187 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
190 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
191 accessing it. Adjust comment.
193 2017-12-19 Jakub Jelinek <jakub@redhat.com>
196 * predict.c (zero_one_minusone): New function.
197 (apply_return_prediction): Avoid return prediction for functions
198 returning only -1, 0 and 1 values, unless they only return -1 and 0
201 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
203 * config/arc/arc.c (legitimate_scaled_address_p): Clean
204 fall-through warning.
205 (arc_compute_frame_size): Remove unused variables.
206 (arc_print_operand): Fix fprintif format.
207 (arc_can_follow_jump): Clean fall-through warning.
209 2017-12-19 Marek Polacek <polacek@redhat.com>
212 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
215 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
217 * config/arc/arc.c (overriderregs): New variable.
218 (arc_override_options): Track fixed/call saved/call options.
219 (arc_conditional_register_usage): Check against overrideregs
220 variable whenever we change register properties.
222 2017-12-19 Nathan Sidwell <nathan@acm.org>
224 * opts.c (finish_options): Don't prefix dump_base_name if it
225 already contains directories.
227 2017-12-19 Martin Liska <mliska@suse.cz>
229 PR rtl-optimization/82675
230 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
231 more element in sbitmap.
233 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
235 * gimplify.c (gimplify_expr): Use error_operand_p.
237 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
240 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
243 2017-12-19 Daniel Cederman <cederman@gaisler.com>
245 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
246 the jump is to a label.
248 2017-12-19 Jakub Jelinek <jakub@redhat.com>
250 PR tree-optimization/83444
251 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
256 * ipa-inline.c (flatten_remove_node_hook): New function.
257 (ipa_inline): Keep only nodes with flatten attribute at the end of
258 the array in the order from ipa_reverse_postorder, only walk that
259 portion of array for flattening, if there is more than one such
260 node, temporarily register a removal hook and ignore removed nodes.
262 PR tree-optimization/80631
263 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
264 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
265 IFN_REDUC_MAX or IFN_REDUC_MIN.
267 2017-12-18 Jeff Law <law@redhat.com>
269 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
270 record symbolic equivalences from backedges in the CFG.
273 2017-11-19 Jeff Law <law@redhat.com>
275 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
276 of degenerates resulting from ignoring an edge.
278 2017-12-18 Martin Sebor <msebor@redhat.com>
281 PR tree-optimization/78450
282 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
283 (handle_builtin_strlen): Call it.
285 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
287 PR rtl-optimization/83424
288 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
290 2017-12-18 Marek Polacek <polacek@redhat.com>
293 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
294 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
297 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
299 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
300 (umaddsidi4, umaddsidi_split): Likewise.
302 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
304 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
307 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
310 * config/s390/s390.c (s390_option_override): Avoid strncpy.
312 2017-12-18 Richard Biener <rguenther@suse.de>
314 PR tree-optimization/81877
315 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
316 (outermost_indep_loop): Adjust.
317 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
318 (can_sm_ref_p): Adjust.
320 2017-12-18 Richard Biener <rguenther@suse.de>
323 * tree.c (array_at_struct_end_p): Return true if the underlying
324 object has space for at least one element in excess of what
325 the array domain specifies.
327 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
329 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
330 example to avoid overfull hbox.
331 * doc/invoke.texi (Option Summary): Add missing @gol.
332 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
335 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
336 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
338 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
339 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
340 (AArch64 Options): Document -mverbose-cost-dump.
341 (ARM Options): Likewise, plus -mflip-thumb.
343 2017-12-17 Martin Sebor <msebor@redhat.com>
346 * gimple-ssa-warn-restrict.c
347 (builtin_memref::offset_out_of_bounds): Correct the handling of
350 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
352 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
354 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
356 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
358 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
360 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
363 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
365 * doc/invoke.texi: Fix some typos.
367 2017-12-16 Martin Sebor <msebor@redhat.com>
369 PR tree-optimization/78918
370 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
371 * builtins.c (check_sizes): Rename...
372 (check_access): ...to this. Rename function arguments for clarity.
373 (check_memop_sizes): Adjust names.
374 (expand_builtin_memchr, expand_builtin_memcpy): Same.
375 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
376 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
377 (check_strncat_sizes, expand_builtin_strncat): Same.
378 (expand_builtin_strncpy, expand_builtin_memset): Same.
379 (expand_builtin_bzero, expand_builtin_memcmp): Same.
380 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
381 (maybe_emit_sprintf_chk_warning): Same.
382 (expand_builtin_strcpy): Adjust.
383 (expand_builtin_stpcpy): Same.
384 (expand_builtin_with_bounds): Detect out-of-bounds accesses
385 in pointer-checking forms of memcpy, memmove, and mempcpy.
386 (gcall_to_tree_minimal, max_object_size): Define new functions.
387 * builtins.h (max_object_size): Declare.
388 * calls.c (alloc_max_size): Call max_object_size instead of
389 hardcoding ssizetype limit.
390 (get_size_range): Handle new argument.
391 * calls.h (get_size_range): Add a new argument.
392 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
393 * doc/invoke.texi (-Wrestrict): Adjust, add example.
394 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
396 (gimple_fold_builtin_memory_chk): Same.
397 (gimple_fold_builtin_stxcpy_chk): New function.
398 * gimple-ssa-warn-restrict.c: New source.
399 * gimple-ssa-warn-restrict.h: New header.
400 * gimple.c (gimple_build_call_from_tree): Propagate location.
401 * passes.def (pass_warn_restrict): Add new pass.
402 * tree-pass.h (make_pass_warn_restrict): Declare.
403 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
405 (handle_builtin_strcat): Same.
406 (strlen_optimize_stmt): Rename...
407 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
408 stpncpy, strncpy, and their checking forms.
410 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
412 PR rtl-optimization/82849
413 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
414 and get_max_loop_iterations_int.
416 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
417 Alan Hayward <alan.hayward@arm.com>
418 David Sherwood <david.sherwood@arm.com>
420 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
421 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
422 size as a poly_uint64.
423 (mode_for_vector, mode_for_int_vector): Take the number of vector
424 elements as a poly_uint64.
425 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
426 size as a poly_uint64.
427 (mode_for_vector, mode_for_int_vector): Take the number of vector
428 elements as a poly_uint64.
430 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
431 Alan Hayward <alan.hayward@arm.com>
432 David Sherwood <david.sherwood@arm.com>
434 * machmode.h (MACRO_MODE): New macro.
435 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
436 * targhooks.c (default_libcall_value, default_secondary_reload)
437 (default_memory_move_cost, default_register_move_cost)
438 (default_class_max_nregs): Likewise.
440 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
441 Alan Hayward <alan.hayward@arm.com>
442 David Sherwood <david.sherwood@arm.com>
444 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
445 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
447 * genautomata.c (main): Likewise.
448 * genconditions.c (write_header): Likewise.
449 * genemit.c (main): Likewise.
450 * genextract.c (print_header): Likewise.
451 * genopinit.c (main): Likewise.
452 * genoutput.c (output_prologue): Likewise.
453 * genpeep.c (main): Likewise.
454 * genpreds.c (write_insn_preds_c): Likewise.
455 * genrecog.c (writer_header): Likewise.
456 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
457 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
458 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
459 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
460 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
461 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
462 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
463 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
464 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
465 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
466 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
467 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
468 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
469 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
470 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
471 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
472 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
473 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
474 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
475 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
476 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
477 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
478 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
479 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
480 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
481 * config/darwin.c (IN_TARGET_CODE): Likewise.
482 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
483 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
484 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
485 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
486 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
487 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
488 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
489 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
490 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
491 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
492 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
493 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
494 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
495 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
496 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
497 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
498 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
499 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
500 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
501 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
502 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
503 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
504 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
505 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
506 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
507 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
508 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
509 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
510 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
511 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
512 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
513 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
514 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
515 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
516 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
517 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
518 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
519 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
520 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
521 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
522 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
523 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
524 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
525 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
526 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
527 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
528 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
529 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
530 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
531 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
532 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
533 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
534 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
535 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
536 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
537 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
538 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
539 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
540 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
541 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
542 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
543 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
544 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
545 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
546 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
547 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
548 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
549 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
550 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
551 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
552 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
553 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
554 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
555 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
556 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
557 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
558 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
559 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
560 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
561 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
562 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
563 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
564 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
565 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
566 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
567 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
568 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
569 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
570 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
571 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
572 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
573 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
574 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
575 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
576 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
577 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
578 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
579 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
580 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
581 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
582 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
583 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
584 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
585 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
586 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
587 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
589 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
590 Alan Hayward <alan.hayward@arm.com>
591 David Sherwood <david.sherwood@arm.com>
593 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
596 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
597 Alan Hayward <alan.hayward@arm.com>
598 David Sherwood <david.sherwood@arm.com>
600 * doc/generic.texi (VEC_SERIES_EXPR): Document.
601 * doc/md.texi (vec_series@var{m}): Document.
602 * tree.def (VEC_SERIES_EXPR): New tree code.
603 * tree.h (build_vec_series): Declare.
604 * tree.c (build_vec_series): New function.
605 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
606 * tree-pretty-print.c (dump_generic_node): Likewise.
607 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
608 * tree-inline.c (estimate_operator_cost): Likewise.
609 * expr.c (expand_expr_real_2): Likewise.
610 * optabs-tree.c (optab_for_tree_code): Likewise.
611 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
612 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
613 * expmed.c (make_tree): Handle VEC_SERIES.
614 * optabs.def (vec_series_optab): New optab.
615 * optabs.h (expand_vec_series_expr): Declare.
616 * optabs.c (expand_vec_series_expr): New function.
617 * tree-vect-generic.c (expand_vector_operations_1): Check that
618 the operands also have vector type.
620 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
621 Alan Hayward <alan.hawyard@arm.com>
622 David Sherwood <david.sherwood@arm.com>
624 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
625 (VEC_COND_EXPR): Add missing @tindex.
626 * doc/md.texi (vec_duplicate@var{m}): Document.
627 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
628 * tree.c (build_vector_from_val): Add stubbed-out handling of
629 variable-length vectors, using VEC_DUPLICATE_EXPR.
630 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
631 * cfgexpand.c (expand_debug_expr): Likewise.
632 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
633 * tree-inline.c (estimate_operator_cost): Likewise.
634 * tree-pretty-print.c (dump_generic_node): Likewise.
635 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
636 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
637 (test_vec_duplicate_folding): New function.
638 (fold_const_c_tests): Call it.
639 * optabs.def (vec_duplicate_optab): New optab.
640 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
641 * optabs.h (expand_vector_broadcast): Declare.
642 * optabs.c (expand_vector_broadcast): Make non-static. Try using
644 * expr.c (store_constructor): Try using vec_duplicate_optab for
646 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
648 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
651 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
652 div/mod latencies a bit.
654 2017-12-15 Jeff Law <law@redhat.com>
656 PR tree-optimization/36550
657 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
658 (mark_threaded_blocks): Rewrite code to avoid block copying when
659 optimizing for size. Don't pessimize blocks which will be
660 copied, but all the statements will be dead.
662 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
664 PR tree-optimization/81165
665 * tree-ssa-threadupdate.c (uses_in_bb): New.
666 (estimate_threading_killed_stmts): New.
667 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
668 * tree-ssa-threadedge.c
669 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
672 2017-12-15 Jeff Law <law@redhat.com>
674 PR tree-optimization/83410
675 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
676 threads when parallelizing loops.
678 2017-12-15 Jakub Jelinek <jakub@redhat.com>
680 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
682 * config/alpha/alpha.c (vms_attribute_table): Swap
683 affects_type_identity and handler fields, adjust comments.
684 * config/mips/mips.c (mips_attribute_table): Likewise.
685 * config/visium/visium.c (visium_attribute_table): Likewise.
686 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
687 * config/microblaze/microblaze.c (microblaze_attribute_table):
689 * config/spu/spu.c (spu_attribute_table): Likewise.
690 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
691 * config/arc/arc.c (arc_attribute_table): Likewise.
692 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
693 * config/v850/v850.c (v850_handle_interrupt_attribute,
694 v850_handle_data_area_attribute): Formatting fixes.
695 (v850_attribute_table): Swap affects_type_identity and handler
696 fields, adjust comments.
697 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
698 * config/arm/arm.c (arm_attribute_table): Likewise.
699 * config/avr/avr.c (avr_attribute_table): Likewise.
700 * config/s390/s390.c (s390_attribute_table): Likewise.
701 * config/sh/sh.c (sh_attribute_table): Likewise.
702 * config/i386/i386.c (ix86_handle_cconv_attribute,
703 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
704 (ix86_attribute_table): Swap affects_type_identity and handler
705 fields, adjust comments.
706 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
707 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
708 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
709 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
710 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
711 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
712 * config/rx/rx.c (rx_attribute_table): Likewise.
713 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
714 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
715 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
716 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
717 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
718 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
719 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
720 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
721 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
722 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
723 * doc/plugins.texi (user_attr): Likewise. Add NULL for
725 * attribs.c (empty_attribute_table): Swap affects_type_identity and
727 (register_scoped_attributes, decl_attributes): Formatting fixes.
729 PR tree-optimization/83269
730 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
731 subtraction in arg0's type if type is signed and arg0 is unsigned.
735 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
736 convert? on @0 instead of convert. Check type of @1, not @0.
737 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
738 convert? on @0 instead of convert. Check type of @1, not @0.
739 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
740 only optimize if either both @1 and @2 types are narrower
741 precision, or both are wider or equal precision, and in the former
742 case only if both have undefined overflow.
744 2017-12-15 Richard Biener <rguenther@suse.de>
747 * internal-fn.def (IFN_NOP): Add.
748 * internal-fn.c (expand_NOP): Do nothing.
749 * lto-streamer-in.c (input_function): Instead of removing
750 sanitizer calls replace them with IFN_NOP calls.
752 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
753 Alan Hayward <alan.hayward@arm.com>
754 David Sherwood <david.sherwood@arm.com>
756 * dse.c (store_info, read_info_type): Replace begin and end with
758 (print_range): New function.
759 (set_all_positions_unneeded, any_positions_needed_p)
760 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
762 (record_store): Likewise. Optimize the case in which all positions
764 (get_stored_val): Replace read_begin and read_end with read_offset
766 (replace_read): Update call accordingly.
768 2017-12-15 Bin Cheng <bin.cheng@arm.com>
770 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
771 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
772 (loop_cand::loop_cand): Initialize above members.
773 (loop_cand::supported_operations): Delete.
774 (loop_cand::can_interchange_p): Inline above function.
775 (loop_cand::classify_simple_reduction): Record number of constant
776 initialized simple reductions.
777 (should_interchange_loops): New parameters. Check stmt cost of loops
779 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
780 Update call to should_interchange_loops.
781 (should_interchange_loop_nest): Update call to
782 should_interchange_loops.
784 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
787 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
788 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
789 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
791 2017-12-15 Julia Koval <julia.koval@intel.com>
793 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
794 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
795 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
796 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
797 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
799 2017-12-15 Julia Koval <julia.koval@intel.com>
801 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
802 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
803 * config/i386/sse.md (vaesenc_<mode>): New pattern.
804 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
805 _mm_aesenc_epi128): New intrinsics.
807 2017-12-15 Julia Koval <julia.koval@intel.com>
809 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
810 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
811 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
812 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
813 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
815 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
817 * gimple-ssa-strength-reduction.c (analyze_increments):
818 Distinguish replacement costs for constant strides from those for
821 2017-12-14 Jakub Jelinek <jakub@redhat.com>
823 * var-tracking.c (variable_tracking_main_1): Formatting fix.
825 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
827 * doc/invoke.texi: Document -Wcast-function-type.
828 * recog.h (stored_funcptr): Change signature.
829 * tree-dump.c (dump_node): Avoid warning.
830 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
832 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
835 * gimple-fold.c (get_range_strlen): Add the handling of non-member
838 2017-12-14 David Malcolm <dmalcolm@redhat.com>
840 PR tree-optimization/83312
841 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
842 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
843 "val" param, and to cope with arbitrary basic blocks.
844 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
845 handle NULL_TREE for "val", dropping "bb" param.
846 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
847 drop "bb" param. Handle NULL_TREE for "val".
848 (find_case_label_for_value): Make "switch_stmt" param const.
849 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
851 (vrp_prop::check_all_array_refs): Reimplement as...
852 (check_array_bounds_dom_walker::before_dom_children): ...this new
853 vfunc. Replace linear search through BB block list, excluding
854 those with non-executable in-edges via dominator walk.
856 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
858 * config/arm/arm.opt (mverbose-cost-dump): New option.
859 * config/arm/arm.c (arm_rtx_costs): Use it.
861 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
864 * reload1.c (emit_input_reload_insns): Skip debug markers.
866 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
868 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
872 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
873 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
874 * config/sh/sh.c: Skip debug insns besides notes.
875 * config/sh/sh.md: Likewise.
876 * config/sh/sh_treg_combine.cc: Likewise.
877 * config/sh/sync.md: Likewise.
879 2017-12-14 Tom de Vries <tom@codesourcery.com>
881 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
882 Add item for weak_undefined.
884 2017-12-14 Richard Biener <rguenther@suse.de>
886 PR tree-optimization/67842
887 * sese.h (bb_in_region): Remove #if 0'ed code.
889 2017-12-14 Richard Biener <rguenther@suse.de>
891 PR tree-optimization/83326
892 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
893 may_be_zero parameter and handle it by not marking the first
894 peeled copy as not exiting the loop.
895 (try_peel_loop): Likewise.
896 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
897 to handle the case of constant or zero iterations and perform
898 loop header copying on-the-fly.
900 2017-12-14 Richard Biener <rguenther@suse.de>
902 PR tree-optimization/83418
903 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
904 Instead of asserting we don't get unfolded comparisons deal with
907 2017-12-14 Jakub Jelinek <jakub@redhat.com>
910 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
911 basic blocks. Assert debug bind insns don't appear outside of bbs,
912 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
915 PR tree-optimization/83198
916 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
917 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
918 value if arg is a REAL_CST with incompatible type.
920 2017-12-14 Sudakshina Das <sudi.das@arm.com>
921 Bin Cheng <bin.cheng@arm.com>
924 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
926 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
929 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
931 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
932 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
934 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
935 Alan Hayward <alan.hayward@arm.com>
936 David Sherwood <david.sherwood@arm.com>
938 * poly-int.h: New file.
939 * poly-int-types.h: Likewise.
940 * coretypes.h: Include them.
941 (POLY_INT_CONVERSION): Define.
942 * target.def (estimated_poly_value): New hook.
943 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
944 * doc/tm.texi: Regenerate.
945 * doc/poly-int.texi: New file.
946 * doc/gccint.texi: Include it.
947 * doc/rtl.texi: Describe restrictions on subreg modes.
948 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
949 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
950 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
951 * targhooks.h (default_estimated_poly_value): Declare.
952 * targhooks.c (default_estimated_poly_value): New function.
953 * target.h (estimated_poly_value): Likewise.
954 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
955 (wi::unary_traits): Delete.
956 (wi::binary_traits::signed_shift_result_type): Define for
957 offset_int << HOST_WIDE_INT, etc.
958 (generic_wide_int::operator <<=): Define for all types and use
959 wi::lshift instead of <<.
960 (wi::hwi_with_prec): Add a default constructor.
961 (wi::ints_for): New class.
962 (operator <<): Define for all wide-int types.
963 (operator /): New function.
964 (operator %): Likewise.
965 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
966 (ASSERT_MAYBE_NE_AT): New macros.
968 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
969 Dominik Vogt <vogt@linux.vnet.ibm.com>
972 * emit-rtl.c (init_emit): Remove ??? comment.
973 * explow.c (get_dynamic_stack_size): Take known alignment of stack
974 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
975 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
976 alignment of 3 virtual registers to BITS_PER_WORD.
978 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
980 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
982 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
983 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
985 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
988 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
990 2017-12-13 Jakub Jelinek <jakub@redhat.com>
992 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
993 with the exception of debug begin stmt markers appear before
997 * final.c (rest_of_handle_final): Call variable_tracking_main only
998 if !flag_var_tracking.
1000 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
1001 Jakub Jelinek <jakub@redhat.com>
1005 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
1006 labels debug stmts that can only appear after labels.
1008 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
1010 PR rtl-optimization/82398
1011 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
1012 EXPR_USEFULNESS in priority comparison.
1014 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
1016 PR rtl-optimization/83393
1017 * combine.c (move_deaths): If reg_stat points to a too new insn in
1018 last_death, do not use it: find the proper insn instead.
1020 2017-12-12 Jeff Law <law@redhat.com>
1022 PR tree-optimization/83298
1023 PR tree-optimization/83362
1024 PR tree-optimization/83383
1025 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
1026 push_value_range a public interface. Add new argument to
1027 record_ranges_from_stmt.
1028 * gimple-ssa-evrp-analyze.c
1029 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
1030 Update comments. Handle recording temporary equivalences.
1031 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
1032 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
1033 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
1034 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
1035 gimple-ssa-evrp-analyze.h.
1036 (record_temporary_equivalences_from_phis): Add new argument. When
1037 the PHI arg is an SSA_NAME, set the result's range to the range
1039 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
1040 from statements too.
1041 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
1042 Pass it down to children as needed.
1043 (thread_outgoing_edges): Likewise.
1044 (thread_across_edge): Likewise. Push/pop range state as needed.
1045 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
1047 2017-12-12 Julia Koval <julia.koval@intel.com>
1049 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
1050 (PTA_CANNONLAKE): Remove PTA_CLWB.
1052 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1055 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
1056 (vcondv2div2df): Likewise.
1057 (vconduv2dfv2di): Likewise.
1059 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1061 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
1062 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
1063 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
1064 ATTR_MATHFN_FPROUNDING_ERRNO.
1066 2017-12-12 Richard Biener <rguenther@suse.de>
1068 PR tree-optimization/83385
1069 * graphite-scop-detection.c (get_order, order): Remove.
1070 (bb_to_rpo): New global.
1072 (build_scops): Sort pbbs in RPO order.
1074 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
1076 * combine.c (simplify_set): Do not transform subregs to zero_extends
1077 if the destination mode is a vector mode.
1079 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1081 PR tree-optimization/83359
1082 * tree-cfg.h (fold_loop_internal_call): Declare.
1083 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
1084 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
1085 (find_loop_dist_alias): New function.
1086 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
1087 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
1090 PR tree-optimization/80631
1091 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
1092 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
1093 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
1094 hardcoding zero as the value if COND_EXPR is never true. For
1095 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
1096 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
1097 hardcoding MAX_EXPR as the reduction operation.
1098 (is_nonwrapping_integer_induction): Allow negative step.
1099 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
1100 vect_create_epilog_for_reduction, if no value is suitable, don't
1101 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
1103 2017-12-12 Richard Biener <rguenther@suse.de>
1105 PR tree-optimization/81889
1106 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
1107 range info from the non-wrapping IV instead of just the range
1110 2017-12-12 Julia Koval <julia.koval@intel.com>
1112 * config.gcc: Add vaesintrin.h.
1113 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1115 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1116 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1118 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1119 * config/i386/immintrin.h: Include vaesintrin.h.
1120 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1121 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1122 _mm_aesdec_epi128): New intrinsics.
1124 2017-12-12 Julia Koval <julia.koval@intel.com>
1126 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
1127 OPTION_MASK_ISA_VAES_UNSET): New.
1128 (ix86_handle_option): Handle -mvaes.
1129 * config/i386/cpuid.h: Define bit_VAES.
1130 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
1131 * config/i386/i386-c.c (__VAES__): New.
1132 * config/i386/i386.c (ix86_target_string): Add -mvaes.
1133 (ix86_valid_target_attribute_inner_p): Ditto.
1134 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
1135 * config/i386/i386.opt: Add -mvaes.
1136 * doc/invoke.texi: Ditto.
1138 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
1140 * debug.h (gcc_debug_hooks): Add inline_entry.
1141 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
1142 * debug.c (do_nothing_debug_hooks): Likewise.
1143 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1144 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
1145 (dwarf2_lineno_debug_hooks): Likewise.
1147 * common.opt (gstatement-frontiers): New, setting
1148 debug_nonbind_markers_p.
1149 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
1150 * toplev.c (process_options): Autodetect value for debug statement
1152 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
1153 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
1155 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
1156 markers. Integrate source bind into debug stmt expand loop.
1157 (pass_expand::execute): Check debug marker limit. Avoid deep
1158 TER and expand debug locations for debug bind insns only.
1159 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
1161 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
1162 * final.c (reemit_insn_block_notes): Take current block from
1163 nonbind markers. Declare note where it's first set.
1164 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
1165 begin stmt markers if enabled.
1166 (notice_source_line): Handle nonbind markers. Fail if their
1167 location is unknown or that of builtins.
1168 (rest_of_handle_final): Convert begin stmt markers to notes if
1169 var-tracking didn't run.
1170 (rest_of_clean_state): Skip begin stmt markers.
1171 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
1173 * function.c (allocate_struct_function): Set begin_stmt_markers.
1174 * function.h (struct function): Add debug_marker_count counter
1175 and debug_nonbind_markers flag.
1176 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
1177 * gimple-low.c (lower_function_body): Adjust
1178 debug_nonbind_markers.
1179 (lower_stmt): Drop or skip gimple debug stmts.
1180 (lower_try_catch): Skip debug stmts.
1181 * gimple.c (gimple_build_debug_begin_stmt): New.
1182 (gimple_copy): Increment debug_marker_count if copying one.
1183 * gimple.h (gimple_build_debug_begin_stmt): Declare.
1184 * gimplify.c (rexpr_location): New.
1185 (rexpr_has_location): New.
1186 (warn_switch_unreachable_r): Handle gimple debug stmts.
1187 (shortcut_cond_r): Call expr_location.
1189 (find_goto_label): New.
1190 (shortcut_cond_expr): Call expr_has_location, expr_location, and
1192 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
1194 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
1195 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
1196 (LANG_HOOKS_INITIALIZER): ... this.
1197 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
1198 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
1200 (update_ebb_live_info): Skip debug insn markers.
1201 * lra.c (debug_insn_static_data): Rename to...
1202 (debug_bind_static_data): ... this.
1203 (debug_marker_static_data): New.
1204 (lra_set_insn_recog_data): Select one of the above depending
1206 (lra_update_isn_regno_info): Don't assume debug insns have
1208 (push_insns): Skip debug insns.
1209 * lto-streamer-in.c (input_function): Drop debug stmts
1210 depending on active options. Adjust debug_nonbind_markers.
1211 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
1212 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
1213 begin stmt marker notes.
1214 (print_insn): Likewise.
1215 * recog.c (extract_insn): Recognize rtl for debug markers.
1216 * rtl.def (DEBUG_MARKER): New.
1217 * tree-inline.c: Include params.h.
1218 (remap_gimple_stmt): Handle nonbind markers.
1219 (maybe_move_debug_stmts_to_successors): Likewise.
1220 (copy_debug_stmt): Likewise.
1221 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
1222 markers regardless of no side effects.
1223 (tsi_link_before): Don't update container's side effects when adding
1224 a begin stmt marker.
1225 (tsi_link_after): Likewise.
1226 (expr_first): Skip begin stmt markers.
1227 (expr_last): Likewise.
1228 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
1229 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1230 Disregard nonbind markers.
1231 * tree.c (make_node_stat): Don't set side effects for begin stmt
1233 (build1_stat): Likewise.
1234 * tree.def (DEBUG_BEGIN_STMT): New.
1235 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
1236 * var-tracking.c (delete_debug_insns): Renamed to...
1237 (delete_vta_debug_insns): ... this.
1238 (reemit_marker_as_note): New.
1239 (vt_initialize): Reemit markers.
1240 (delete_vta_debug_insns): Likewise.
1241 (vt_debug_insns_local): Reemit or delete markers.
1242 (variable_tracking_main_1): Likewise.
1243 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
1244 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
1245 (gimple_debug_nonbind_marker_p): New.
1246 (gimple_build_debug_bind): Adjust.
1247 (gimple_build_debug_begin_stmt): New.
1248 * doc/invoke.texi (max-debug-marker-count): New param.
1249 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
1250 (debug_parameter_ref, debug_marker): New.
1251 (NOTE_INSN_BEGIN_STMT): New.
1252 (DEBUG_INSN): Describe begin stmt markers.
1254 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
1255 without debug insns, we wouldn't, but clean up debug insns
1256 after a control flow insn nevertheless.
1258 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
1259 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
1260 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
1261 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
1262 (rtl_tidy_fallthru_edge): Likewise.
1263 (rtl_verify_fallthru): Likewise.
1264 (rtl_verify_bb_layout): Likewise.
1265 (skip_insns_after_block): Likewise.
1266 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
1267 * dwarf2out.c: Include print-rtl.h.
1268 (dwarf2out_next_real_insn): New.
1269 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
1270 Dump debug binds in asm comments.
1271 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
1272 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
1273 callers to use gsi_start_nondebug_bb instead.
1274 (gsi_after_labels): Skip gimple debug stmts.
1275 (gsi_start_nondebug): New.
1276 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
1277 (proper_loop_form_for_interchange): Adjust.
1278 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
1279 * gimple.h (gimple_seq_last_nondebug_stmt): New.
1280 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
1281 (collect_fallthrough_labels): Likewise.
1282 (should_warn_for_implicit_fallthrough): Likewise.
1283 (warn_implicit_fallthrough_r): Likewise.
1284 (expand_FALLTHROUGH_r): Likewise.
1285 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
1286 (graphite_copy_stmts_from_block): Skip nonbind markers.
1287 * haifa-sched.c (sched_extend_bb): Skip debug insns.
1288 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
1289 * jump.c (clean_barriers): Skip debug insns.
1290 * omp-expand.c (expand_parallel_call): Skip debug insns.
1291 (expand_task_call): Likewise.
1292 (remove_exit_barrier): Likewise.
1293 (expand_omp_taskreg): Likewise.
1294 (expand_omp_for_init_counts): Likewise.
1295 (expand_omp_for_generic): Likewise.
1296 (expand_omp_for_static_nochunk): Likewise.
1297 (expand_omp_for_static_chunk): Likewise.
1298 (expand_omp_simd): Likewise.
1299 (expand_omp_taskloop_for_outer): Likewise.
1300 (expand_omp_taskloop_for_inner): Likewise.
1301 (expand_oacc_for): Likewise.
1302 (expand_omp_sections): Likewise.
1303 (expand_omp_single): Likewise.
1304 (expand_omp_synch): Likewise.
1305 (expand_omp_atomic_load): Likewise.
1306 (expand_omp_atomic_store): Likewise.
1307 (expand_omp_atomic_fetch_op): Likewise.
1308 (expand_omp_atomic_pipeline): Likewise.
1309 (expand_omp_atomic_mutex): Likewise.
1310 (expand_omp_target): Likewise.
1311 (grid_expand_omp_for_loop): Likewise.
1312 (grid_expand_target_grid_body): Likewise.
1313 (build_omp_regions_1): Likewise.
1314 * omp-low.c (check_combined_parallel): Skip debug stmts.
1315 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
1316 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
1317 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
1319 * tree-cfg.c (make_blobs_1): Skip debug stmts.
1320 (make_edges): Likewise.
1321 (cleanup_dead_labels): Likewise.
1322 (gimple_can_merge_blocks_p): Likewise.
1323 (stmt_starts_bb_p): Likewise.
1324 (gimple_block_label): Likewise.
1325 (gimple_redirect_edge_and_branch): Likewise.
1326 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
1328 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
1330 * tree-pretty-print (print_declaration): Omit initializer in slim
1332 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
1334 (eliminate_unnecessary_stmts): Stabilize block removal order.
1335 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
1336 * var-tracking.c (get_first_insn): New.
1337 (vt_emit_notes): Call it.
1338 (vt_initialize): Walk any insns before the first BB.
1339 (delete_debug_insns): Likewise.
1341 * gimple.h (enum gimple_debug_subcode): Add
1342 GIMPLE_DEBUG_BEGIN_STMT.
1343 (gimple_debug_begin_stmt_p): New.
1344 (gimple_debug_nonbind_marker_p): New.
1345 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
1346 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
1347 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
1348 * insn-notes.def (BEGIN_STMT): New.
1349 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
1350 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
1351 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
1352 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
1353 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
1354 (INSN_DEBUG_MARKER_KIND): New.
1355 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
1356 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
1357 (INSN_VAR_LOCATION_PTR): New.
1358 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
1360 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
1361 deps only if debug bind insns are enabled.
1362 (pass_expand::execute): Avoid deep TER and expand
1363 debug locations for debug bind insns only.
1364 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
1365 debug stmts special handling down to debug bind stmts.
1366 * combine.c (try_combine): Narrow debug insns special handling
1367 down to debug bind insns.
1368 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
1369 Narrow debug insns preexisting special handling down to debug
1371 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
1372 handling down to debug bind insns.
1373 * function.c (instantiate_virtual_regs): Skip debug markers,
1374 adjust handling of debug binds.
1375 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
1376 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
1377 * haifa-sched.c (schedule_insn): Narrow special handling of debug
1378 insns to debug bind insns.
1379 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
1380 special handling of debug stmts to debug bind stmts.
1381 * ipa-split.c (split_function): Likewise.
1382 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
1383 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
1385 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
1386 * regrename.c (build_def_use): Likewise.
1387 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
1388 (pass_cprop_hardreg): Narrow special casing of debug insns to
1390 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
1391 * reload1.c (reload): Likewise.
1392 * sese.c (sese_insert_phis_for_liveouts): Narrow special
1393 casing of debug stmts to debug bind stmts.
1394 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
1395 * ssa-iterators.h (num_imm_uses): Likewise.
1396 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
1397 debug stmts to debug bind stmts.
1398 * tree-inline.c (tree_function_versioning): Narrow special casing
1399 of debug stmts to debug bind stmts.
1400 * tree-loop-distribution.c (generate_loops_for_partition):
1401 Narrow special casing of debug stmts to debug bind stmts.
1402 * tree-sra.c (analyze_access_subtree): Narrow special casing
1403 of debug stmts to debug bind stmts.
1404 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
1405 stmts to debug bind stmts.
1406 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
1407 casing of debug stmts to debug bind stmts.
1408 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
1409 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
1410 casing of debug stmts to debug bind stmts.
1411 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1413 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
1414 debug stmts to debug bind stmts.
1415 (gimple_replace_ssa_lhs): Likewise.
1416 (insert_debug_temp_for_var_def): Likewise.
1417 (insert_debug_temps_for_defs): Likewise.
1418 (reset_debug_uses): Likewise.
1419 * tree-ssanames.c (release_ssa_name_fn): Likewise.
1420 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
1421 (adjust_debug_stmts): Likewise.
1422 (adjust_phi_and_debug_stmts): Likewise.
1423 (vect_do_peeling): Likewise.
1424 * tree-vect-loop.c (vect_transform_loop): Likewise.
1425 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
1426 * var-tracking.c (adjust_mems): Narrow special casing of debug
1427 insns to debug bind insns.
1428 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
1429 (compute_bb_dataflow, vt_find_locations): Likewise.
1430 (vt_expand_loc, emit_notes_for_changes): Likewise.
1431 (vt_init_cfa_base): Likewise.
1432 (vt_emit_notes): Likewise.
1433 (vt_initialize): Likewise.
1434 (vt_finalize): Likewise.
1436 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
1437 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
1438 (next_nonnote_nondebug_insn_bb): New.
1439 (prev_nonnote_nondebug_insn_bb): New.
1440 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
1441 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
1442 (prev_nonnote_nondebug_insn_bb): Declare.
1443 (next_nonnote_nondebug_insn_bb): Declare.
1444 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
1445 * cfgrtl.c (get_last_bb_insn): Likewise.
1446 * lra.c (push_insns): Likewise.
1448 2017-12-11 David Malcolm <dmalcolm@redhat.com>
1451 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
1453 * selftest.c (selftest::test_runner::test_runner): New ctor.
1454 (selftest::test_runner::~test_runner): New dtor.
1455 * selftest.h (class selftest::test_runner): New class.
1457 2017-12-11 Carl Love <cel@us.ibm.com>
1459 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
1460 vec_extract_fp32_from_shortl]): Add #defines.
1461 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
1462 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
1463 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
1464 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
1465 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
1466 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
1467 * doc/extend.texi: Add documentation for the added builtins.
1469 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
1471 PR rtl-optimization/80693
1472 PR rtl-optimization/81019
1473 PR rtl-optimization/81020
1474 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
1475 are not mentioned in i3. Place the REG_UNUSED note on i2,
1476 possibly modified to REG_DEAD, if it did not originate in i3.
1478 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1480 * recog.c (store_data_bypass_p_1): New function.
1481 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
1482 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
1484 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
1486 PR rtl-optimization/83361
1487 * ifcvt.c (if_convert): Call fixup_partitions.
1489 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1491 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1492 early folding of splat_u{8,16,32}.
1494 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1496 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
1497 output_operand_lossage first argument with capital letter.
1498 (aarch64_override_options): Don't start error and sorry first argument
1499 with capital letter.
1501 2017-12-11 Andi Kleen <ak@linux.intel.com>
1503 PR gcov-profile/83355
1504 * auto-profile.c (string_table::get_index_by_decl): Don't
1505 recurse when abstract origin points to itself.
1507 2017-12-11 Bin Cheng <bin.cheng@arm.com>
1509 PR tree-optimization/83320
1510 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
1511 (prune_datarefs_not_in_loop): Ditto.
1513 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
1515 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
1517 2017-12-10 Jakub Jelinek <jakub@redhat.com>
1519 PR tree-optimization/83337
1520 * gimple-loop-interchange.cc (compute_access_stride): Handle
1521 bitfield DRs properly.
1523 2017-12-09 Jakub Jelinek <jakub@redhat.com>
1525 PR tree-optimization/83338
1526 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
1527 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
1528 vector of unsigned integers to vector of signed integers.
1530 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
1532 PR rtl-optimization/83317
1533 * lra-constraints.c (process_address_1): Add insn code check.
1535 2017-12-08 Michael Matz <matz@suse.de>
1537 Fix PR tree-optimization/83323
1538 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
1539 head-controlled loops and loop BBs.
1540 * common.opt (funroll-and-jam): Remove, instead ...
1541 (floop-unroll-and-jam): ... reuse this option.
1542 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
1543 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
1544 (-floop-unroll-and-jam): ... this option.
1546 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1548 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
1549 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
1550 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
1552 2017-12-08 Richard Biener <rguenther@suse.de>
1555 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
1556 handle accesses outside of zero-sized vars.
1558 2017-12-08 Martin Jambor <mjambor@suse.cz>
1560 PR tree-optimization/83141
1561 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
1562 test for MEM_REFs implicitely changing types with padding. Remove
1564 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
1566 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1568 * config/arc/arc.c (arc_attribute_table): Add exclusions to
1570 * config/avr/avr.c (avr_attribute_table): Likewise.
1571 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1572 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1573 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1574 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
1575 of struct attribute_spec.
1576 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1578 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
1581 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
1583 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
1586 * profile-count.c (profile_count::from_gcov_type): Move from
1587 profile-count.h; handle overflow.
1588 * profile-count.h (profile_count::from_gcov_type): Move offline.
1590 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
1592 PR rtl-optimization/83304
1593 * combine.c (move_deaths): If we do not know where a register died,
1596 2017-12-08 Richard Biener <rguenther@suse.de>
1598 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
1599 Provide -fopt-info-loop feedback when we interchange in a nest.
1601 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1603 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
1604 for armv6 ARM CPU IDs.
1606 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1608 * common/config/arm/arm-common.c: Include <algorithm>.
1609 (INCLUDE_VECTOR): Define.
1610 (compare_opt_names): New function.
1611 (arm_rewrite_selected_arch): Only strip out extensions that can be
1612 expressed through -mfpu. Sort the remaining extensions
1615 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1617 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
1618 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
1619 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
1620 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
1622 2017-06-08 Tristan Gingold <gindold@adacore.com>
1625 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
1626 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
1628 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1630 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
1631 of struct attribute_spec.
1633 2017-12-08 Julia Koval <julia.koval@intel.com>
1635 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
1636 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
1637 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
1638 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
1639 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
1640 _mm_maskz_dpwssds_epi32): Ditto.
1642 2017-12-08 Richard Biener <rguenther@suse.de>
1644 PR tree-optimization/81303
1645 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
1646 conditions try to create a comparison vector type matching
1647 the data vector type.
1648 (vectorizable_condition): Adjust.
1649 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
1650 Leave invariant conditions alone in case we can vectorize those.
1652 2017-12-08 Julia Koval <julia.koval@intel.com>
1654 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
1655 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
1656 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
1657 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
1658 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
1660 2017-12-08 Julia Koval <julia.koval@intel.com>
1662 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
1663 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
1664 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
1665 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
1666 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
1667 _mm_maskz_dpbusds_epi32): New intrinsics.
1669 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1671 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
1673 (ld<bh>io_signed): Likewise.
1674 (st<bhw>io): Likewise.
1675 * config/nios2/predicates.md (ldstio_memory_operand): Allow
1676 SMALL_INT12 constant integer operand.
1678 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1680 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
1682 (nios2_large_constant_allowed): ...this. Adjust uses.
1683 (nios2_plus_symbolic_constant_p): Rename to...
1684 (nios2_plus_large_constant_p): ...this. Adjust uses.
1685 (nios2_legitimate_address_p): Correct CONST_INT handling.
1686 (nios2_symbolic_memory_operand_p): Rename to...
1687 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
1688 (nios2_large_constant_p): Check for large constant integers too.
1689 (nios2_split_large_constant): Handle constant integers.
1690 (nios2_split_symbolic_memory_operand): Rename to...
1691 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
1692 (nios2_legitimize_constant_address): Handle constant integers.
1693 (r0rel_constant_p): Handle small constant integers.
1694 (nios2_print_operand_address): Handle r0-relative integer addresses.
1695 * config/nios2/nios2-protos.h: Adjust for renamed functions.
1696 * config/nios2/nios2.md: Adjust for renamed functions.
1698 2017-12-07 Andrew Waterman <andrew@sifive.com>
1700 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
1701 (TARGET_HAVE_SRODATA_SECTION): New define.
1702 (riscv_select_section): New function.
1704 2017-12-08 Joseph Myers <joseph@codesourcery.com>
1705 Alexander Monakov <amonakov@ispras.ru>
1706 Jakub Jelinek <jakub@redhat.com>
1709 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
1711 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1713 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
1714 a bool success value. Don't call output_operand_lossage here.
1715 (aarch64_print_ldpstp_address): Return a bool success value.
1716 (aarch64_print_operand_address): Call output_addr_const if
1717 aarch64_print_address_internal fails.
1718 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
1719 'y'; call output_operand_lossage instead. Call output_operand_lossage
1720 if aarch64_print_ldpstp_address fails.
1722 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1724 * tree-vector-builder.h
1725 (tree_vector_builder::binary_encoded_nelts): Declare.
1726 * tree-vector-builder.c
1727 (tree_vector_builder::binary_encoded_nelts): New function.
1728 * fold-const.c (negate_expr_p): Likewise.
1729 (operand_equal_p, fold_checksum_tree): Likewise.
1730 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
1731 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
1732 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
1733 (uniform_vector_p): Likewise.
1734 * varasm.c (const_hash_1, compare_constant): Likewise.
1735 * tree-ssa-ccp.c: Include tree-vector-builder.h.
1736 (valid_lattice_transition): Operate directly on the VECTOR_CST
1738 * ipa-icf.c: Include tree-vector-builder.h.
1739 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
1740 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
1742 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1744 * tree.c (build_vector): Delete.
1745 * tree.h (build_vector): Make static and move into the self-testing
1748 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1750 * vector-builder.h (vector_builder::derived): New const overload.
1751 (vector_builder::elt): New function.
1752 * tree-vector-builder.h (tree_vector_builder::type): New function.
1753 (tree_vector_builder::apply_step): Declare.
1754 * tree-vector-builder.c (tree_vector_builder::apply_step): New
1756 * gimple-fold.h (tree_vector_builder): Declare.
1757 (gimple_build_vector): Take a tree_vector_builder instead of a
1758 type and vector of elements.
1759 * gimple-fold.c (gimple_build_vector): Likewise.
1760 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
1762 (get_initial_defs_for_reduction): Likewise.
1763 (vectorizable_induction): Likewise.
1765 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1767 * tree-vector-builder.h
1768 (tree_vector_builder::new_binary_operation): Declare.
1769 * tree-vector-builder.c
1770 (tree_vector_builder::new_binary_operation): New function.
1771 * fold-const.c (fold_relational_const): Use it.
1772 (const_binop): Likewise. Check that both input vectors have
1773 the same number of elements, thus excluding things like WIDEN_SUM.
1774 Check whether it is possible to operate directly on the encodings
1777 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1779 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
1780 new_unary_operation, operating only on the encoded elements.
1781 (const_unop): Likewise.
1782 (exact_inverse): Likewise.
1783 (distributes_over_addition_p): New function.
1784 (const_binop): Use tree_vector_builder and new_unary_operation
1785 for combinations of VECTOR_CST and INTEGER_CST. Operate only
1786 on the encoded elements unless the encoding is strided and the
1787 operation does not distribute over addition.
1788 (fold_convert_const): Use tree_vector_builder and
1789 new_unary_operation. Operate only on the encoded elements
1790 for truncating integer conversions, or for non-stepped encodings.
1792 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1794 * config/sparc/sparc.c: Include tree-vector-builder.h.
1795 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
1796 * expmed.c: Include tree-vector-builder.h.
1797 (make_tree): Use tree_vector_builder instead of build_vector.
1798 * fold-const.c: Include tree-vector-builder.h.
1799 (const_binop): Use tree_vector_builder instead of build_vector.
1800 (const_unop): Likewise.
1801 (native_interpret_vector): Likewise.
1802 (fold_vec_perm): Likewise.
1803 (fold_ternary_loc): Likewise.
1804 * gimple-fold.c: Include tree-vector-builder.h.
1805 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
1807 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
1808 (simplify_vector_constructor): Use tree_vector_builder instead
1810 * tree-vect-generic.c: Include tree-vector-builder.h.
1811 (add_rshift): Use tree_vector_builder instead of build_vector.
1812 (expand_vector_divmod): Likewise.
1813 (optimize_vector_constructor): Likewise.
1814 * tree-vect-loop.c: Include tree-vector-builder.h.
1815 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
1816 of build_vector. Explicitly use a stepped encoding for
1818 * tree-vect-slp.c: Include tree-vector-builder.h.
1819 (vect_get_constant_vectors): Use tree_vector_builder instead
1821 (vect_transform_slp_perm_load): Likewise.
1822 (vect_schedule_slp_instance): Likewise.
1823 * tree-vect-stmts.c: Include tree-vector-builder.h.
1824 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
1825 (vect_gen_perm_mask_any): Likewise.
1826 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
1827 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
1829 (build_vector_from_val): Likewise. Explicitly use a duplicate
1832 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
1834 * doc/generic.texi (VECTOR_CST): Describe new representation of
1836 * vector-builder.h: New file.
1837 * tree-vector-builder.h: Likewise.
1838 * tree-vector-builder.c: Likewise.
1839 * Makefile.in (OBJS): Add tree-vector-builder.o.
1840 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
1841 * tree-core.h (tree_base): Add a vector_cst field to the u union.
1842 (tree_vector): Change the number of elements to
1843 vector_cst_encoded_nelts.
1844 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
1845 (VECTOR_CST_ELTS): Delete.
1846 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
1847 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
1848 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
1849 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
1850 (VECTOR_CST_ENCODED_ELT): Likewise.
1851 (vector_cst_encoded_nelts): New function.
1852 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1853 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1854 (vector_cst_int_elt, vector_cst_elt): Declare.
1855 * tree.c: Include tree-vector-builder.h.
1856 (tree_code_size): Abort if passed VECTOR_CST.
1857 (tree_size): Update for new VECTOR_CST layout.
1858 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1859 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1860 (build_vector): Use tree_vector_builder.
1861 (vector_cst_int_elt, vector_cst_elt): New functions.
1862 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
1863 encoded elements and then create the vector in the canonical form.
1864 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
1865 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
1866 (tree_c_tests): Call test_vector_cst_patterns.
1867 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
1869 (hash_tree): Likewise.
1870 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
1871 (streamer_write_tree_header): Likewise.
1872 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
1873 (streamer_alloc_tree): Likewise. Update call to make_vector.
1874 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
1876 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1878 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
1879 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
1880 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
1882 2017-12-07 Bin Cheng <bin.cheng@arm.com>
1883 Richard Biener <rguenther@suse.de>
1885 PR tree-optimization/81303
1886 * Makefile.in (gimple-loop-interchange.o): New object file.
1887 * common.opt (floop-interchange): Reuse the option from graphite.
1888 * doc/invoke.texi (-floop-interchange): Ditto. New document for
1889 -floop-interchange and mention it for -O3.
1890 * opts.c (default_options_table): Enable -floop-interchange at -O3.
1891 * gimple-loop-interchange.cc: New file.
1892 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
1893 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
1894 * passes.def (pass_linterchange): New pass.
1895 * timevar.def (TV_LINTERCHANGE): New time var.
1896 * tree-pass.h (make_pass_linterchange): New declaration.
1897 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
1898 interchange. Record IV before/after increment in new parameters.
1899 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
1900 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
1902 (check_reduction_path): ...New function here.
1903 * tree-vectorizer.h (check_reduction_path): New declaration.
1905 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
1908 PR rtl-optimization/80818
1909 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
1910 always early clobbered.
1911 * lra-lives.c (process_bb_lives): Check input hard regs for early
1912 clobbered non-operand hard reg.
1914 2017-12-07 Jakub Jelinek <jakub@redhat.com>
1917 * tree-cfg.c (verify_gimple_assign_binary): Don't require
1918 types_compatible_p, just that TYPE_MODE is the same.
1920 2017-12-07 Martin Sebor <msebor@redhat.com>
1923 * attribs.c (empty_attribute_table): Initialize new member of
1924 struct attribute_spec.
1925 (decl_attributes): Add argument. Handle mutually exclusive
1926 combinations of attributes.
1927 (selftests::test_attribute_exclusions): New function.
1928 (selftests::attribute_c_tests): Ditto.
1929 * attribs.h (decl_attributes): Add default argument.
1930 * selftest.h (attribute_c_tests): Declare.
1931 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1932 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1934 * doc/extend.texi (Common Function Attributes): Update const and pure.
1935 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1936 of struct attribute_spec.
1937 * config/arc/arc.c (arc_attribute_table): Same.
1938 * config/arm/arm.c (arm_attribute_table): Same.
1939 * config/avr/avr.c ( avr_attribute_table): Same.
1940 * config/bfin/bfin.c (bfin_attribute_table): Same.
1941 * config/cr16/cr16.c (cr16_attribute_table): Same.
1942 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1943 * config/h8300/h8300.c (h8300_attribute_table): Same.
1944 * config/i386/i386.c (ix86_attribute_table): Same.
1945 * config/ia64/ia64.c (ia64_attribute_table): Same.
1946 * config/m32c/m32c.c (m32c_attribute_table): Same.
1947 * config/m32r/m32r.c (m32r_attribute_table): Same.
1948 * config/m68k/m68k.c (m68k_attribute_table): Same.
1949 * config/mcore/mcore.c (mcore_attribute_table): Same.
1950 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1951 * config/mips/mips.c (mips_attribute_table): Same.
1952 * config/msp430/msp430.c (msp430_attribute_table): Same.
1953 * config/nds32/nds32.c (nds32_attribute_table): Same.
1954 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1955 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1956 * config/rl78/rl78.c (rl78__attribute_table): Same.
1957 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1958 * onfig/rx/rx.c (rx_attribute_table): Same.
1959 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1960 * config/sh/sh.c (sh_attribute_table): Same.
1961 * config/sparc/sparc.c (sparc_attribute_table): Same.
1962 * config/spu/spu.c (spu_attribute_table): Same.
1963 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1964 * config/v850/v850.c (v850_attribute_table): Same.
1965 * config/visium/visium.c (visium_attribute_table): Same.
1967 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1970 * config/arm/arm.c (INCLUDE_STRING): Define.
1971 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1972 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1975 2017-12-07 Michael Matz <matz@suse.de>
1977 Add unroll and jam pass
1979 * gimple-loop-jam.c: New file.
1980 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1981 * common.opt (funroll-and-jam): New option.
1982 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1983 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1984 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1985 * passes.def: Add pass_loop_jam.
1986 * timevar.def (TV_LOOP_JAM): Add.
1987 * tree-pass.h (make_pass_loop_jam): Declare.
1988 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1989 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1990 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1991 to flow_loop_tree_node_add.
1992 (duplicate_subloops, copy_loops_to): Append to sibling list.
1993 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1994 * doc/invoke.texi (-funroll-and-jam): Document new option.
1995 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1997 2017-12-07 Richard Biener <rguenther@suse.de>
1999 PR tree-optimization/83296
2000 PR tree-optimization/67769
2001 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
2002 flow sensitive info in an unrelated BB.
2003 (value_replacement): Use reset_flow_sensitive_info.
2004 (minmax_replacement): Reset flow sensitive info on the def
2005 we move. Do not reset flow sensitive info in the whole BB
2006 we move the stmt to.
2007 (abs_replacement): Likewise.
2009 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2012 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2013 rs6000_cpu to the given -mcpu=, or to the default processor.
2015 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2017 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
2018 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
2019 instead of rs6000_cpu_attr.
2020 (rs6000_adjust_cost): Ditto.
2021 (is_microcoded_insn): Ditto.
2022 (rs6000_adjust_priority): Ditto.
2023 (rs6000_issue_rate): Ditto.
2024 (rs6000_use_sched_lookahead): Ditto.
2025 (rs6000_use_sched_lookahead_guard): Ditto.
2026 (rs6000_sched_reorder): Ditto.
2027 (force_new_group): Ditto.
2028 * config/rs6000/rs6000.md (cpu attribute): Ditto.
2029 (group_ending_nop): Ditto.
2031 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
2033 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
2034 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
2035 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
2036 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
2038 (rs6000_reassociation_width): Ditto.
2039 (rs6000_emit_epilogue): Ditto.
2040 (rs6000_adjust_cost): Ditto.
2041 (is_microcoded_insn): Ditto.
2042 (is_cracked_insn): Ditto.
2043 (rs6000_adjust_priority): Ditto.
2044 (rs6000_sched_reorder): Ditto.
2045 (rs6000_sched_reorder2): Ditto.
2046 (insn_must_be_first_in_group): Ditto.
2047 (insn_must_be_last_in_group): Ditto.
2048 (rs6000_register_move_cost): Ditto.
2049 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
2052 2017-12-07 Julia Koval <julia.koval@intel.com>
2054 * config.gcc: Add vaesintrin.h.
2055 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
2057 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
2058 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
2060 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
2061 * config/i386/immintrin.h: Include vaesintrin.h.
2062 * config/i386/sse.md (vaesdec_<mode>): New pattern.
2063 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
2064 _mm_aesdec_epi128): New intrinsics.
2066 2017-12-06 David Malcolm <dmalcolm@redhat.com>
2068 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
2069 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
2070 c-family/c-spellcheck.cc.
2071 (best_macro_match::best_macro_match): Likewise.
2073 (struct edit_distance_traits<cpp_hashnode *>): Move to
2074 c-family/c-spellcheck.h.
2075 (class best_macro_match): Likewise.
2077 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2079 PR tree-optimization/83293
2080 * gimple-ssa-strength-reduction.c (insert_initializers): Use
2081 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
2082 might insert into empty bb.
2085 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
2086 simplify for plus with :c added, and pointer_plus without that.
2087 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
2088 with undefined overflow and the conversion is not widening,
2089 perform negation in utype and only convert to type afterwards.
2090 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
2091 simplify for plus with :c added, and pointer_plus without that.
2092 If type is integral with undefined overflow and the conversion is
2093 not widening, perform minus in utype and only convert to type
2094 afterwards. Move the last pointer_diff_expr simplify into the
2097 2017-12-06 Martin Sebor <msebor@redhat.com>
2099 PR tree-optimization/82646
2100 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
2101 strncpy, not maxlen.
2103 2017-12-06 Martin Sebor <msebor@redhat.com>
2105 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
2108 PR tree-optimization/83075
2109 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
2110 strncat/strncpy don't change length of source string.
2112 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
2115 2017-11-29 Martin Aberg <maberg@gaisler.com>
2117 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2118 to prevent b2bst errata sequence.
2119 (sqrtdf2_fix): Likewise.
2121 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2123 PR tree-optimization/81945
2124 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
2125 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
2126 to dest_cfun has orig_loop_num set, either remap it to the new
2127 loop number if the loop got moved too, or clear it.
2129 2017-12-05 Steve Ellcey <sellcey@cavium.com>
2131 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
2133 (thunderx2t99_nothing): New insn reservation.
2134 (thunderx2t99_mrs): New insn reservation.
2135 (thunderx2t99_multiple): New insn reservation.
2136 (thunderx2t99_alu_basi): Add bfx to reservation.
2137 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
2139 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2142 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
2144 2017-12-05 Bin Cheng <bin.cheng@arm.com>
2146 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
2147 tree-ssa-pre.c::remove_dead_inserted_code.
2148 * tree-ssa-dce.h: New file.
2149 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
2150 (remove_dead_inserted_code): Move and rename to function
2151 tree-ssa-dce.c::simple_dce_from_worklist.
2152 (pass_pre::execute): Update use.
2154 2017-12-05 Richard Biener <rguenther@suse.de>
2156 PR tree-optimization/83277
2157 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
2158 to code-gen liveout vars.
2160 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
2162 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
2163 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
2164 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
2167 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
2169 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
2171 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
2172 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
2175 2017-12-05 Richard Biener <rguenther@suse.de>
2177 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
2178 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
2179 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
2180 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
2182 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2184 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
2186 2017-12-05 Martin Liska <mliska@suse.cz>
2187 Jakub Jelinek <jakub@redhat.com>
2189 * doc/invoke.texi: Document the options.
2190 * flag-types.h (enum sanitize_code): Add
2191 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2192 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
2193 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2194 * opts.c: Define new sanitizer options.
2195 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
2196 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
2198 2017-12-05 Julia Koval <julia.koval@intel.com>
2200 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
2201 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
2202 (ix86_handle_option): Handle -mavx512vnni.
2203 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
2204 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2205 * config/i386/i386-c (__AVX512VNNI__): New.
2206 * config/i386/i386.c (ix86_target_string): Handle new option.
2207 (ix86_valid_target_attribute_inner_p): Handle new option.
2208 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
2209 * config/i386/i386.opt (mavx512vnni): New option.
2211 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2214 * athlon.md: Disable for generic.
2215 * haswell.md: Enable for generic.
2216 * i386.c (ix86_sched_init_global): Add core hooks for generic.
2217 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
2219 (ix86_adjust_cost): Move generic to haswell path.
2221 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
2223 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
2224 instead of MEM_P in a couple more places. Fix formatting issues.
2226 2017-12-04 Jim Wilson <jimw@sifive.com>
2228 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
2229 instead of GP_REG_LAST-1.
2230 (riscv_adjust_libcall_cfi_prologue): Likewise.
2231 (riscv_adjust_libcall_cri_epilogue): Likewise.
2232 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
2235 2017-12-04 Luis Machado <luis.machado@linaro.org>
2237 * ipa-pure-const.c (check_decl): Add missing newline.
2238 (state_from_flags): Likewise.
2240 2017-12-04 Jeff Law <law@redhat.com>
2242 PR tree-optimizatin/78496
2243 * gimple-ssa-evrp-analyze.h
2244 (evrp_range_analyzer::get_vr_values): Simplify.
2245 * gimple-ssa-evrp-analyze.c: Corresponding changes.
2246 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
2247 and gimple-ssa-evrp-analyze.h.
2248 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
2249 (simplify_stmt_for_jump_threading): Copy a blob of code from
2250 tree-vrp.c to use ranges to simplify statements.
2251 (dom_opt_dom_walker::before_dom_children): Call
2252 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
2253 (dom_opt_dom_walker::after_dom_children): Similarly for
2254 evrp_range_analyzer::leave.
2255 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
2258 * gimple-ssa-evrp-analyze.c
2259 (evrp_range_analyzer::extract_range_from_stmt): Always use
2260 vr_values::update_value_range so preexisting range info is
2261 medged with new range info, even if the new range is VR_VARYING.
2263 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2265 * combine.c: Adjust comment.
2266 (use_crosses_set_p): Delete.
2267 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
2268 (try_combine): Ditto.
2270 2017-12-04 Richard Biener <rguenther@suse.de>
2272 PR tree-optimization/83255
2273 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
2274 Re-add zero-iteration check.
2276 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2278 PR rtl-optimization/83245
2279 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
2280 hard registers as earlyclobber, also if not in an asm.
2282 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2286 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2289 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2290 rs6000_cpu based on cpu_index, not tune_index.
2292 2017-12-04 Richard Biener <rguenther@suse.de>
2294 PR tree-optimization/83238
2295 * graphite-scop-detection.c (scop_detection::merge_sese): Make
2296 code match comment, rejecting invalid SESE regions.
2298 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
2300 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
2301 require base operand is a REG_POINTER prior to reload on targets
2302 with non-equivalent space registers.
2304 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2306 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
2307 (update_specialized_profile): Fix updating of counts.
2308 (perhaps_add_new_callers): Likewise.
2310 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2313 * x86-tune.def: Remove obsolette FIXMEs.
2314 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
2315 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
2316 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
2318 (X86_TUNE_PAD_RETURNS): Disable for generic.
2320 2017-12-02 Jakub Jelinek <jakub@redhat.com>
2322 PR tree-optimization/83170
2323 PR tree-optimization/83241
2324 * gimple-ssa-store-merging.c
2325 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
2326 gimple_vuse (ins_stmt) in case it has changed.
2327 (imm_store_chain_info::output_merged_store): Likewise.
2329 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
2333 * tree-cfg.c (pass_warn_function_return::execute): Handle
2334 __builtin_ubsan_handle_missing_return like __builtin_unreachable
2335 with BUILTINS_LOCATION.
2339 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
2340 is BLKmode for vector field with vector raw mode, use TYPE_MODE
2341 instead of DECL_MODE.
2343 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
2344 last argument to rtx pointer.
2345 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
2346 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
2347 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
2348 depending on the chosen ISAs.
2349 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
2350 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
2351 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
2353 * config/i386/sse.md (mov<mode>_internal): Likewise.
2354 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
2356 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2358 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
2360 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2362 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
2363 parameters from prototype.
2364 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
2365 parameters. Don't print a comment.
2366 (emit_fusion_gpr_load): Adjust.
2367 (emit_fusion_load_store): Adjust.
2368 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
2369 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
2370 comment on the second line.
2372 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2375 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2376 rs6000_cpu based on cpu_index, not tune_index.
2378 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2380 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
2381 which_alternative instead of which_alternative + 1.
2382 (output_asm_insn): Print an extra tab if the template is short.
2384 2017-12-01 Jim Wilson <jimw@sifive.com>
2386 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
2388 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
2389 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
2391 * doc/tm.texi: Regenerate.
2393 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2396 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
2397 whether we can allocate pseudos before trying to fix an address.
2398 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
2399 memory address is indexed or indirect.
2400 (floatuns_<mode>si2_hw2): Likewise.
2402 2017-12-01 Jason Merrill <jason@redhat.com>
2404 * Makefile.in (TAGS): Add c-family/*.cc.
2406 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
2408 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
2409 (call_value_insn): Likewise.
2410 (sibcall_insn): Likewise.
2411 (sibcall_value_insn): Likewise.
2412 (movsi_aarch64): Likewise.
2413 (movdi_aarch64): Likewise.
2414 (add_losym_): Likewise.
2415 (ldr_got_small_): Likewise.
2416 (ldr_got_small_sidi): Likewise.
2417 (ldr_got_small_28k_): Likewise.
2418 (ldr_got_small_28k_sidi): Likewise.
2419 * config/aarch64/aarch64.c (aarch64_print_address_internal):
2420 Move output_addr_const to symbolic case. Add error check.
2422 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2424 * config/s390/predicates.md (plus16_Q_operand): New predicate.
2425 * config/s390/s390.md: Disable MVC merging peephole if it would
2426 disable operand forwarding.
2427 (new peephole2): Split MVCs if it would turn them into up to 2
2430 2017-12-01 Richard Biener <rguenther@suse.de>
2432 PR tree-optimization/83232
2433 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
2434 detection of same access. Instead of breaking the group here
2435 do not consider the duplicate. Add comment explaining real fix.
2437 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
2439 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
2441 2017-12-01 Sudakshina Das <sudi.das@arm.com>
2443 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
2445 2017-12-01 Jakub Jelinek <jakub@redhat.com>
2447 * function.h (struct function): Remove cilk_frame_decl,
2448 is_cilk_function and calls_cilk_spawn fields.
2449 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
2451 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
2452 cilk_elemental field.
2453 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
2454 * target.def: Adjust comment.
2455 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
2456 Don't test cilk_elemental.
2458 PR tree-optimization/83233
2459 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
2460 bswap_stat name for the struct.
2463 * tree.h (SWITCH_BREAK_LABEL_P): Define.
2464 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
2465 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
2466 SWITCH_BREAK_LABEL_P set on the label.
2467 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
2468 added for default case if it was missing and not all cases covered.
2469 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
2470 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
2472 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
2473 to avoid -Wimplicit-fallthrough warning.
2474 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
2475 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
2477 PR tree-optimization/83221
2478 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
2480 (init_reassoc): Formatting fix.
2483 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
2484 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
2488 * tree-cfg.c: Include asan.h.
2489 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
2491 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2493 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
2495 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2497 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
2498 vpcmpeqd instruction.
2500 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2502 * config/i386/i386.c (standard_sse_constant_opcode): Fix
2503 registers type for 128bit mode.
2505 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2507 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
2509 * gengtype-state.c (read_state_pair): Likewise.
2510 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
2511 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
2512 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
2513 * ifcvt.c (noce_try_store_flag_constants): Likewise.
2514 * tree-ssa-ccp.c (ccp_finalize): Likewise.
2515 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
2516 * builtins.c (fold_builtin_3): Likewise.
2517 * graphite-scop-detection.c
2518 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
2519 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
2522 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
2525 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
2526 parameter. Return true if ignore_decl_rtl_set_p is true and other
2527 conditions are satisfied.
2528 * asan.h (asan_protect_global): Add new parameter.
2529 * varasm.c (categorize_decl_for_section): Pass true as second parameter
2530 to asan_protect_global calls.
2532 2017-11-30 Jim Wilson <jimw@sifive.com>
2534 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
2535 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
2536 -mstrict-align, add info on default value. Delete redundant lines for
2537 -mabi. Add missing -mexplicit-relocs docs.
2539 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2541 * config/arc/arc.md (trap): New pattern.
2543 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2545 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
2546 instruction to end into a delay slot.
2547 * config/arc/arc.md (cond_delay_insn): Check if the instruction
2548 can be placed into a delay slot against reg_note.
2550 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2552 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
2553 labels number of usages.
2555 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2557 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
2559 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
2561 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2564 * internal-fn.c (expand_mul_overflow): Optimize unsigned
2565 multiplication by power of 2 constant into two shifts + comparison.
2567 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
2570 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
2572 2017-11-30 Richard Biener <rguenther@suse.de>
2574 PR tree-optimization/83202
2575 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2576 allow_peel argument and guard peeling.
2577 (canonicalize_loop_induction_variables): Likewise.
2578 (canonicalize_induction_variables): Pass false.
2579 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
2580 peeling from cunrolli.
2582 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2584 * combine.c (try_combine): Print a message to dump file whenever
2585 I0, I1, or I2 cannot be combined into I3.
2587 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2589 PR rtl-optimization/83156
2590 PR rtl-optimization/82621
2591 * combine.c (try_combine): Don't split an I2 if one of the dests is
2592 set again before I3. Allow unused dests.
2594 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2596 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
2598 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
2600 PR rtl-optimization/80818
2601 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
2602 recursively. Use insn code for clobber.
2603 (lra_set_insn_recog_data): Pass the new arg to
2604 collect_non_operand_hard_regs.
2605 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
2607 (lra_update_insn_regno_info): Pass insn to
2608 add_regs_to_insn_regno_info.
2610 2017-11-29 Jim Wilson <jimw@sifive.com>
2611 Andrew Waterman <andrew@sifive.com>
2613 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
2614 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
2615 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
2616 use SINGLE_SHIFT_COST.
2617 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
2618 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
2620 2017-11-29 Julia Koval <julia.koval@intel.com>
2622 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
2623 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
2624 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
2625 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
2626 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
2627 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
2628 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
2629 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
2630 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
2631 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
2632 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
2633 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
2634 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
2635 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
2636 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
2637 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
2638 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
2639 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
2640 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
2641 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
2642 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
2643 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
2644 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
2645 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
2646 __builtin_ia32_vpshldv_v2di_mask,
2647 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
2648 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
2649 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
2651 2017-11-29 Julia Koval <julia.koval@intel.com>
2653 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
2654 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
2655 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
2656 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
2657 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
2658 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
2659 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
2660 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
2661 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
2662 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
2663 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
2664 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
2665 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
2666 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
2667 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
2668 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
2669 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
2670 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
2671 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2672 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
2673 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
2675 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2677 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
2678 movsi_pic_gotdata_op instruction as a load for the UT699 errata
2681 2017-11-29 Martin Aberg <maberg@gaisler.com>
2683 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2684 to prevent b2bst errata sequence.
2685 (sqrtdf2_fix): Likewise.
2687 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2689 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
2690 (div_sqrt_insn_p): New function.
2691 (sparc_do_work_around_errata): Insert NOP instructions to
2692 prevent sequences that could trigger the TN-0013 errata for
2693 certain LEON3 processors.
2694 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
2695 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
2696 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
2697 (in_branch_delay): Prevent div and sqrt in delay slot if
2699 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
2701 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2703 * config/sparc/sparc.c (atomic_insn_p): New function.
2704 (sparc_do_work_around_errata): Insert NOP instructions to
2705 prevent sequences that could trigger the TN-0010 errata for
2707 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
2708 instruction referable in atomic_insns_p.
2710 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2712 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
2713 (atomic_compare_and_swap_leon3_1): Likewise.
2716 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2718 * config/sparc/sparc.c (fpop_insn_p): New function.
2719 (sparc_do_work_around_errata): Insert NOP instructions to
2720 prevent sequences that could trigger the TN-0012 errata for
2722 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
2723 * config/sparc/sparc.md (fix_gr712rc): New attribute.
2724 (in_branch_annul_delay): Prevent floating-point instructions
2725 in delay slot of annulled integer branch.
2727 2017-11-29 Richard Biener <rguenther@suse.de>
2729 PR tree-optimization/83202
2730 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
2732 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
2733 nodes vectorized to the same stmts multiple times.
2734 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
2735 (vect_analyze_slp_instance): Adjust.
2736 (scalar_stmts_to_slp_tree_map_t): New typedef.
2737 (vect_schedule_slp_instance): Add a map recording the SLP node
2738 representing the vectorized stmts for a set of scalar stmts.
2739 Avoid code-generating redundancies.
2740 (vect_schedule_slp): Allocate map and pass it down.
2742 2017-11-29 Nathan Sidwell <nathan@acm.org>
2745 * tree.c (build_complex_type): Fix canonicalization. Only fill in
2748 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
2750 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
2751 cases for printing LDP/STP memory addresses.
2752 (aarch64_print_address_internal): Renamed from
2753 aarch64_print_operand_address, added parameter, add Pmode check.
2754 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
2755 (aarch64_print_operand_address): Indirect to
2756 aarch64_print_address_internal.
2757 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
2758 'y' operand output specifier.
2760 2017-11-29 Jakub Jelinek <jakub@redhat.com>
2763 * tree.c (build_simple_mem_ref_loc): Handle
2764 get_addr_base_and_unit_offset returning a MEM_REF.
2767 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
2770 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
2773 2017-11-28 David Malcolm <dmalcolm@redhat.com>
2775 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
2777 (layout::move_to_column): Likewise.
2779 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2782 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
2783 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
2785 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2786 Martin Jambor <mjambor@suse.cz>
2789 * tree.h (expr_type_first_operand_type_p): Declare
2790 * tree.c (expr_type_first_operand_type_p): New function.
2791 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
2792 (ipa_value_from_jfunc): Adjust declaration.
2793 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
2794 Use it as result type for arithmetics, unless it is NULL in which case
2795 be more conservative.
2796 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
2797 ipa_get_jf_pass_through_result.
2798 (propagate_vals_across_pass_through): Likewise.
2799 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
2800 is to propagate_vals_across_pass_through.
2801 (propagate_constants_across_call): Pass PARM_TYPE to
2802 propagate_scalar_across_jump_function.
2803 (find_more_scalar_values_for_callers_subset): Pass parameter type to
2804 ipa_value_from_jfunc.
2805 (cgraph_edge_brings_all_scalars_for_node): Likewise.
2806 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
2807 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
2808 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
2809 target_type, pass it to ipa_value_from_jfunc.
2810 (update_indirect_edges_after_inlining): Pass parameter type to
2811 try_make_edge_direct_simple_call.
2813 2017-11-28 Jeff Law <law@redhat.com>
2815 * gimple-ssa-evrp-analyze.c
2816 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
2817 refine ranges if scev_initialized_p returns true.
2818 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
2820 2017-11-28 Julia Koval <julia.koval@intel.com>
2822 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
2823 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
2824 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
2825 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
2826 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
2827 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
2828 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
2829 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
2830 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
2831 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
2832 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
2833 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
2834 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
2835 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
2836 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
2837 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
2838 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
2839 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
2840 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
2841 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
2842 __builtin_ia32_vpshrd_v2di_mask): New builtins.
2843 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
2845 2017-11-28 Julia Koval <julia.koval@intel.com>
2847 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
2848 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
2849 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
2850 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
2851 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
2852 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
2853 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
2854 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
2855 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
2856 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
2857 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
2858 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
2859 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
2860 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
2861 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
2862 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
2863 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
2864 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
2865 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
2866 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
2867 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
2868 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
2869 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
2870 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
2871 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
2872 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
2873 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
2874 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
2875 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
2876 __builtin_ia32_vpshld_v2di_mask): New builtins.
2877 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2878 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
2880 2017-11-28 Richard Biener <rguenther@suse.de>
2882 PR tree-optimization/80776
2883 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
2885 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
2887 (evrp_range_analyzer::record_ranges_from_incoming_edges):
2888 If the incoming edge is an effective fallthru because the other
2889 edge only reaches a __builtin_unreachable () then record ranges
2890 derived from the controlling condition in SSA info.
2891 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
2892 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2894 2017-11-28 Olivier Hainque <hainque@adacore.com>
2896 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
2897 on mingw build hosts.
2899 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2902 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
2904 2017-11-28 Richard Biener <rguenther@suse.de>
2907 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
2908 copies generated from memcpy use a character array as reference
2911 2017-11-28 Julia Koval <julia.koval@intel.com>
2912 Sebastian Peryt <sebastian.peryt@intel.com>
2914 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
2915 c-family/c-cilkplus.o, c-family/array-notation-common.o,
2916 cilk-common.o, cilk.h, cilk-common.c): Remove.
2918 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
2919 * builtins.c (is_builtin_name): Remove cilkplus condition.
2920 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
2921 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
2922 cilk-builtins.def, cilkplus.def): Remove.
2923 * cif-code.def (CILK_SPAWN): Remove.
2924 * cilk-builtins.def: Delete.
2925 * cilk-common.c: Ditto.
2927 * cilkplus.def: Ditto.
2928 * config/darwin.h (fcilkplus): Delete.
2929 * cppbuiltin.c: Ditto.
2930 * doc/extend.texi: Remove cilkplus doc.
2931 * doc/generic.texi: Ditto.
2932 * doc/invoke.texi: Ditto.
2933 * doc/passes.texi: Ditto.
2934 * gcc.c (fcilkplus): Remove.
2935 * gengtype.c (cilk.h): Remove.
2936 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2938 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2940 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2942 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2943 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2944 cilkplus conditions.
2945 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2946 inline_read_section): Ditto.
2947 * ipa-inline-analysis.c (cilk.h): Remove.
2948 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2949 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2950 append_linker_options): Remove condition for fcilkplus.
2951 * lto/lto-lang.c (cilk.h): Remove.
2952 (lto_init): Remove condition for fcilkplus.
2953 * omp-expand.c (expand_cilk_for_call): Delete.
2954 (expand_omp_taskreg, expand_omp_for_static_chunk,
2955 expand_omp_for): Remove cilkplus
2957 (expand_cilk_for): Delete.
2958 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2959 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2960 execute_lower_omp, diagnose_sb_0): Ditto.
2961 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2962 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2963 * tree-nested.c: Ditto.
2964 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2965 (dump_generic_node): Ditto.
2966 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2967 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2969 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2971 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2973 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2974 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2975 for vector mode and !TARGET_SIMD.
2977 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2979 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2981 * tree.h (SWITCH_LABELS): Remove.
2982 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2983 assert SWITCH_BODY is non-NULL.
2984 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2986 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2988 PR tree-optimization/80788
2989 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2990 has TREE_OVERFLOW set, call drop_tree_overflow.
2992 2017-11-28 Richard Biener <rguenther@suse.de>
2994 PR tree-optimization/83158
2995 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2997 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
3000 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
3001 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
3003 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
3005 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
3006 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
3008 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3011 * varasm.c (compare_constant): Take the mode of the constants into
3012 account when comparing floating point constants.
3014 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
3016 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
3017 from explicit instantiation of debug_helper.
3018 * vec.h (DEFINE_DEBUG_VEC): Ditto.
3020 2017-11-27 Richard Biener <rguenther@suse.de>
3022 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
3025 2017-11-27 Richard Biener <rguenther@suse.de>
3027 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
3028 (type_hash_canon): Also clear int_cst_hash_table entry for
3030 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
3032 2017-11-27 Tamar Christina <tamar.christina@arm.com>
3034 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3035 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3036 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
3038 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3040 * hash-map.h (gt_cleare_cache): Avoid UB.
3042 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
3044 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
3045 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
3046 (decide_unroll_runtime_iterations): Likewise.
3047 (decide_unroll_stupid): Likewise.
3049 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3052 * config/i386/i386.md: Add a loop with incssp.
3054 2017-11-27 Martin Jambor <mjambor@suse.cz>
3056 PR tree-optimization/81248
3057 * tree-sra.c (splice_param_accesses): Remove size check.
3058 (decide_one_param_reduction): Fix size check.
3059 * gimple-pretty-print.c (dump_profile): Silence warning.
3060 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
3062 2017-11-27 Jakub Jelinek <jakub@redhat.com>
3065 * dbxout.c (lastlineno): New variable.
3066 (dbx_debug_hooks): Use dbxout_switch_text_section as
3067 switch_text_section debug hook.
3068 (dbxout_function_end): Switch to current_function_section
3069 rather than function_section. If crtl->has_bb_partition,
3070 output just one N_FUN, depending on in_cold_section_p.
3071 (dbxout_source_line): Remember last lineno in lastlineno.
3072 (dbxout_switch_text_section): New function.
3073 (dbxout_function_decl): Adjust dbxout_block caller.
3074 (dbx_block_with_cold_children): New function.
3075 (dbxout_block): Return true if any LBRAC/RBRAC have been
3076 emitted. Use dbx_block_with_cold_children at depth == 0
3077 in second partition. Add PARENT_BLOCKNUM argument, pass
3078 it optionally adjusted to children. Output LBRAC/RBRAC
3079 around recursive call only if the block is in the current
3080 partition, if not and anything was output, emit empty
3082 * final.c (final_scan_insn): Compute cold_function_name
3083 before calling switch_text_section debug hook. Call
3084 that hook even if dwarf2out_do_frame if not emitting
3088 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
3089 TREE_READONLY decls.
3091 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3093 PR rtl-optimization/82488
3094 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
3096 2017-11-26 Julia Koval <julia.koval@intel.com>
3098 * config/i386/i386.c (processor_target_table): Add skylake_cost for
3100 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
3103 2017-11-26 Julia Koval <julia.koval@intel.com>
3105 * config/i386/driver-i386.c (host_detect_local_cpu):
3106 Detect skylake-avx512.
3108 2017-11-26 Julia Koval <julia.koval@intel.com>
3110 * config.gcc: Add -march=cannonlake.
3111 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
3112 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
3113 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
3114 (PTA_CANNONLAKE): New.
3115 (processor_target_table): Add cannonlake.
3116 (ix86_option_override_internal): Ditto.
3117 (fold_builtin_cpu): Ditto.
3118 (get_builtin_code_for_version): Handle cannonlake.
3119 (M_INTEL_COREI7_CANNONLAKE): New.
3120 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
3121 * doc/invoke.texi: Add -march=cannonlake.
3123 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
3125 * plugin.c (add_new_plugin): Use platform-specific library extensions.
3126 (try_init_one_plugin): Alternative implementation for MinGW.
3127 * Makefile.in (plugin_implib): New.
3128 (gengtype-lex.c): Fix broken AIX workaround.
3129 * configure: Regenerate.
3130 * doc/plugins.texi: Document support for MinGW.
3132 2017-11-25 Jakub Jelinek <jakub@redhat.com>
3134 PR rtl-optimization/81553
3135 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
3136 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
3137 is a shift where C1 has different mode than the whole shift, use C1's
3138 mode for MULT rather than the shift's mode.
3141 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
3142 builtins not enabled in the currently selected ISA.
3144 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
3146 PR tree-optimization/71026
3147 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
3148 (insert_reciprocals): Change to insert reciprocals before a division
3149 by a square and to insert the square of a reciprocal.
3150 (execute_cse_reciprocals_1): Change to consider division by a square.
3151 (register_division_in): Add importance parameter.
3153 2017-11-24 Richard Biener <rguenther@suse.de>
3155 PR tree-optimization/82402
3156 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
3157 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
3159 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
3161 * match.pd (0-ptr): New transformation.
3163 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
3166 * ipa-inline.c (inline_small_functions): Set current badnes correctly
3167 when skipping checking.
3169 2017-11-24 Richard Biener <rguenther@suse.de>
3171 PR tree-optimization/83128
3172 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
3173 (vn_reference_lookup_3): Likewise.
3175 2017-11-24 Jakub Jelinek <jakub@redhat.com>
3178 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
3179 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
3182 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
3183 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
3184 check for the latter with gimple_call_builtin_p. Do not handle
3185 BUILT_IN_STPNCPY_CHK which is not a pass through call.
3187 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
3189 * config/arm/arm_neon.h: Fix pragma GCC push_options before
3192 2017-11-23 Julia Koval <julia.koval@intel.com>
3194 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
3195 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
3196 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
3197 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
3198 _mm512_maskz_expandloadu_epi16): New intrinsics.
3199 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
3200 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
3201 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
3202 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
3203 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
3204 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
3205 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
3206 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
3207 _mm256_maskz_expandloadu_epi8): New intrinsics.
3208 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
3209 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
3210 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
3211 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
3212 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
3213 * config/i386/sse.md (VI248_VLBW): New iterator.
3214 (expand<mode>_mask, expand<mode>_maskz): New patterns.
3216 2017-11-23 Julia Koval <julia.koval@intel.com>
3218 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
3219 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
3220 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
3221 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
3222 _mm512_mask_compressstoreu_epi16): New.
3223 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
3224 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
3225 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
3226 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
3227 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
3228 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
3229 _mm256_mask_compressstoreu_epi8): New.
3230 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
3231 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
3232 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
3233 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
3234 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
3235 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
3236 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
3237 __builtin_ia32_compresshi128_mask,
3238 __builtin_ia32_compressstoreuqi512_mask,
3239 __builtin_ia32_compressstoreuhi512_mask,
3240 __builtin_ia32_compressstoreuqi256_mask,
3241 __builtin_ia32_compressstoreuqi128_mask,
3242 __builtin_ia32_compressstoreuhi256_mask,
3243 __builtin_ia32_compressstoreuhi128_mask): New builtins.
3244 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
3246 (ix86_expand_special_args_builtin): Handle new types.
3247 (s4fma_expand): Handle new builtin array.
3248 * config/i386/immintrin.h: Include new headers.
3249 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
3250 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
3252 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3254 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
3255 fixes. Declare temp and g variables at the top in order to avoid
3256 {} in most of the cases.
3258 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
3260 * match.pd (ptr-0): New transformation.
3262 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3264 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
3265 (struct addr_mode_cost_table): New.
3266 (struct tune_params): Add field addr_mode_costs.
3267 * config/arm/arm.c (generic_addr_mode_costs): New.
3268 (arm_slowmul_tune): Initialise addr_mode_costs field.
3269 (arm_fastmul_tune): Likewise.
3270 (arm_strongarm_tune): Likewise.
3271 (arm_xscale_tune): Likewise.
3272 (arm_9e_tune): Likewise.
3273 (arm_marvell_pj4_tune): Likewise.
3274 (arm_v6t2_tune): Likewise.
3275 (arm_cortex_tune): Likewise.
3276 (arm_cortex_a8_tune): Likewise.
3277 (arm_cortex_a7_tune): Likewise.
3278 (arm_cortex_a15_tune): Likewise.
3279 (arm_cortex_a35_tune): Likewise.
3280 (arm_cortex_a53_tune): Likewise.
3281 (arm_cortex_a57_tune): Likewise.
3282 (arm_exynosm1_tune): Likewise.
3283 (arm_xgene1_tune): Likewise.
3284 (arm_cortex_a5_tune): Likewise.
3285 (arm_cortex_a9_tune): Likewise.
3286 (arm_cortex_a12_tune): Likewise.
3287 (arm_cortex_a73_tune): Likewise.
3288 (arm_v7m_tune): Likewise.
3289 (arm_cortex_m7_tune): Likewise.
3290 (arm_v6m_tune): Likewise.
3291 (arm_fa726te_tune): Likewise.
3292 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
3294 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3296 * config/arm/arm.c (arm_mem_costs): New function.
3297 (arm_rtx_costs_internal): Use arm_mem_costs.
3299 2017-11-23 Mark Wielaard <mark@klomp.org>
3301 * dwarf2out.c (init_sections_and_labels): Use generation to create
3302 unique ranges_section_label and ranges_base_label. Return generation.
3303 (output_rnglists): Add generation argument. Use generation to create
3304 unique ranges labels.
3305 (dwarf2out_finish): Get generation from init_sections_and_labels
3306 and pass generation to output_rnglists.
3308 2017-11-23 Mike Stump <mikestump@comcast.net>
3309 Eric Botcazou <ebotcazou@adacore.com>
3311 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
3312 * cfgloop.h (struct loop): Add unroll field.
3313 * function.h (struct function): Add has_unroll bitfield.
3314 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
3315 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
3316 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
3317 (pass_rtl_unroll_loops::gate): Likewise.
3318 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
3319 for which loop->unroll==1.
3320 (decide_unroll_constant_iterations): Use note for consistency and
3321 take loop->unroll into account. Return early if loop->unroll is set.
3322 Fix thinko in existing test.
3323 (decide_unroll_runtime_iterations): Use note for consistency and
3324 take loop->unroll into account.
3325 (decide_unroll_stupid): Likewise.
3326 * lto-streamer-in.c (input_cfg): Read loop->unroll.
3327 * lto-streamer-out.c (output_cfg): Write loop->unroll.
3328 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
3330 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
3331 (print_loop): Print loop->unroll if set.
3332 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
3333 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
3334 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
3336 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
3337 loop->unroll is set and smaller than the trip count. Otherwise bypass
3338 entirely the heuristics if loop->unroll is set. Remove dead note.
3339 Fix off-by-one bug in other note.
3340 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
3341 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
3343 (tree_unroll_loops_completely): Make static.
3344 (pass_complete_unroll::execute): Use correct type for variable.
3345 (pass_complete_unrolli::execute): Fix formatting.
3346 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
3348 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
3350 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
3351 enable when TARGET_PREFER_AVX128 is set.
3353 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3355 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
3357 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
3358 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
3360 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3362 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
3363 frequencies. Fix estimation of aggregate parameters.
3365 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3367 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
3370 2017-11-23 Tom de Vries <tom@codesourcery.com>
3372 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
3374 2017-11-23 Tom de Vries <tom@codesourcery.com>
3376 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
3379 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
3382 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
3383 sibcall_value_pcrel_fdpic): Use local variable instead of
3385 (calli_tbr_rel): Add missing operand 2.
3386 (call_valuei_tbr_rel): Add missing operand 3.
3388 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3391 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
3392 bitpos/bitsize covering the whole destination, use store_expr only if
3393 the complex mode is the same. Otherwise, use expand_normal and if
3394 it returns CONCAT, subreg each part separately instead of trying to
3395 subreg the whole result.
3397 2017-11-23 Richard Biener <rguenther@suse.de>
3399 PR tree-optimization/23094
3400 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
3401 come here from walking over backedges in the first iteration.
3402 (vn_reference_lookup_3): Skip clobbers that store the same value.
3404 2017-11-23 Richard Biener <rguenther@suse.de>
3406 PR tree-optimization/81403
3407 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
3408 a block we need a leader relative to.
3409 (phi_translate_1): For nary processing require a leader from
3410 get_representative_for given we run expression simplification
3411 using match-and-simplify. Remove previous fix.
3413 2017-11-22 Jeff Law <law@redhat.com>
3415 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3416 Use new method allocate_value_range rather than accessing the
3417 vrp_value_range_pool data member directly.
3418 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
3419 to use extract_range_from_stmt method to avoid need for
3420 extract_range_from_assignment method.
3421 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
3422 method rather than setting values_propgated data member directly.
3423 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
3424 and values propagated data members and extract_range_from_assignment
3425 method. Reorder private data members to conform to standards.
3426 Add new methods set_lattice_propagation_complete and
3427 allocate_value_range.
3429 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
3431 PR rtl-optimization/83030
3432 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
3433 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
3434 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
3435 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
3436 CROSSING_JUMP_P flag is already set before setting it.
3437 * cfgrtl.c (fixup_partition_crossing): Likewise.
3438 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
3441 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3443 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
3444 Handle the case where both arguments are using gen_const_vec_series.
3446 2017-11-22 David Malcolm <dmalcolm@redhat.com>
3449 * pretty-print.c (pp_format): Move quoting implementation to
3450 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
3451 to pass address of "quote" local.
3452 (pp_begin_quote): New function.
3453 (pp_end_quote): New function.
3454 * pretty-print.h (printer_fn): Convert penultimate param from bool
3456 (pp_begin_quote): New decl.
3457 (pp_end_quote): New decl.
3458 * tree-diagnostic.c (default_tree_printer): Convert penultimate
3459 param from bool to bool *.
3460 * tree-diagnostic.h (default_tree_printer): Likewise.
3462 2017-11-22 Jeff Law <law@redhat.com>
3464 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
3465 Initialize vr_values.
3466 (evrp_range_analyzer::try_find_new_range): Call methods attached to
3467 vr_values via vr_values class instance rather than delegators.
3468 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
3469 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3470 (evrp_range_analyzer::push_value_range): Likewise.
3471 (evrp_range_analyzer::pop_value_range): Likewise.
3472 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
3473 most delegators. Those remaining are exposed as public interfaces.
3474 Make vr_values a pointer and private.
3475 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
3477 (evrp_range_analyzer::get_vr_value): New method.
3478 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
3479 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
3480 (class evrp_dom_walker): Attach evrp_folder class, initialize
3481 it in the ctor. Remove temporary delegators.
3482 (evrp_dom_walker::before_dom_children): Call methods in attached
3483 evrp_range_analyzer class via class instance pointer. Use
3484 free value_range_constant_singleton to remove need for
3485 op_with_constant_singleton_value delegator method. Do not
3486 create a vrp_prop class instance for every call! Narrow
3487 scope of a couple variables.
3488 (evrp_dom_walker::cleanup): Call methods in attached
3489 evrp_range_analyzer class via class instance pointer.
3490 * vr-values.h (class vr_values): Privatize many methods and
3493 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3495 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
3496 * cfgexpand.c (expand_debug_expr): Remove handling for them.
3497 * expr.c (expand_expr_real_2): Likewise.
3498 * fold-const.c (const_unop): Likewise.
3499 * optabs-tree.c (optab_for_tree_code): Likewise.
3500 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3501 * tree-inline.c (estimate_operator_cost): Likewise.
3502 * tree-pretty-print.c (dump_generic_node): Likewise.
3503 (op_code_prio): Likewise.
3504 (op_symbol_code): Likewise.
3505 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
3506 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
3507 * internal-fn.c (direct_internal_fn_optab): New function.
3508 (direct_internal_fn_array, direct_internal_fn_supported_p
3509 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
3510 * fold-const-call.c (fold_const_reduction): New function.
3511 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
3513 * tree-vect-loop.c: Include internal-fn.h.
3514 (reduction_code_for_scalar_code): Rename to...
3515 (reduction_fn_for_scalar_code): ...this and return an internal
3517 (vect_model_reduction_cost): Take an internal_fn rather than
3519 (vect_create_epilog_for_reduction): Likewise. Build calls rather
3521 (vectorizable_reduction): Use internal functions rather than tree
3522 codes for the reduction operation. Update calls to the functions
3524 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
3525 Use calls to internal functions rather than REDUC tree codes.
3526 * config/aarch64/aarch64-simd.md: Update comment accordingly.
3528 2017-11-22 Olivier Hainque <hainque@adacore.com>
3530 * config/vxworks.c (vxworks_override_options): Pick default
3531 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
3532 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
3533 DWARF_GNAT_ENCODINGS_DEFAULT.
3534 * config/vxworksae.h: Likewise.
3536 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
3538 PR tree-optimization/83104
3539 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
3542 2017-11-22 Marek Polacek <polacek@redhat.com>
3543 H.J. Lu <hongjiu.lu@intel.com>
3544 Jason Merrill <jason@redhat.com>
3549 * calls.c (initialize_argument_information): Call
3550 warn_parameter_passing_abi target hook.
3551 (store_one_arg): Use 0 for empty record size. Don't push 0 size
3552 argument onto stack.
3553 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
3554 * common.opt: Update -fabi-version description.
3555 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
3556 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
3558 (ix86_is_empty_record): New function.
3559 (ix86_warn_parameter_passing_abi): New function.
3560 (TARGET_EMPTY_RECORD_P): Redefine.
3561 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
3562 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
3563 * doc/tm.texi: Regenerated.
3564 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
3565 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
3566 * dwarf2out.c (get_ultimate_context): Move to tree.c.
3567 * explow.c (hard_function_value): Call arg_int_size_in_bytes
3568 instead of int_size_in_bytes.
3569 * expr.c (copy_blkmode_to_reg): Likewise.
3570 * function.c (aggregate_value_p): Return 0 for empty types.
3571 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
3573 (locate_and_pad_parm): Call arg size_in_bytes instead
3575 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
3576 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
3577 * target.def (empty_record_p, warn_parameter_passing_abi): New target
3579 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
3580 (std_gimplify_va_arg_expr): Skip empty records. Call
3581 arg_size_in_bytes instead size_in_bytes.
3582 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
3583 * tree-core.h (tree_type_common): Add empty_flag.
3584 (tree_decl_common): Update comments.
3585 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
3587 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3588 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
3590 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3591 * tree.c (default_is_empty_type): New function.
3592 (default_is_empty_record): New function.
3593 (arg_int_size_in_bytes): New function.
3594 (arg_size_in_bytes): New function.
3595 (get_ultimate_context): New function.
3596 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
3597 TRANSLATION_UNIT_WARN_EMPTY_P.
3598 (default_is_empty_record, arg_int_size_in_bytes,
3599 arg_size_in_bytes, get_ultimate_context): Declare.
3601 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3603 * config/arm/arm.c (cmse_clear_registers): New function.
3604 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
3605 code by call to cmse_clear_registers.
3606 (cmse_nonsecure_entry_clear_before_return): Likewise.
3608 2017-11-22 Tamar Christina <tamar.christina@arm.com>
3610 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
3611 (vdot_s32, vdotq_s32): New.
3612 (vdot_lane_u32, vdotq_lane_u32): New.
3613 (vdot_lane_s32, vdotq_lane_s32): New.
3616 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3619 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
3620 for unsigned values with fewer HWIs than the precision.
3621 (test_overflow): New function.
3622 (wide_int_cc_tests): Call it.
3624 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3626 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
3627 has the same mode class as Pmode.
3629 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3631 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
3632 padding_bits_to_clear_ptr.
3633 (cmse_nonsecure_entry_clear_before_return): Likewise.
3635 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3637 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
3638 auto_sbitap instead of integer bitfield to control register needing
3641 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3643 PR tree-optimization/83044
3644 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
3645 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
3646 that need the upper bound. Subtract offset from
3647 get_addr_base_and_unit_offset only if positive and subtract it
3648 before division by eltsize rather than after it.
3651 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
3652 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
3653 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
3657 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
3658 expand_binop with *mul_widen_optab, make sure at least one of the
3659 operands doesn't have VOIDmode.
3662 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
3664 PR rtl-optimization/82044
3665 PR tree-optimization/82042
3666 * dse.c (record_store): Check for overflow.
3667 (check_mem_read_rtx): Properly check for overflow if width == -1, call
3668 add_wild_read instead of clear_rhs_from_active_local_stores on
3669 overflow and log it into dump_file.
3671 2017-11-22 Richard Biener <rguenther@suse.de>
3673 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
3674 fake edges to exit when looking for a place to insert.
3675 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
3677 (insert_into_preds_of_block): Commit edge insertion immediately,
3678 assert that doesn't require new BBs.
3679 (fini_pre): Release expressions.
3680 (pass_pre::execute): Shuffle things around a bit, if the fn
3681 is too large do not compute AVAIL either as this is really the
3684 2017-11-22 Richard Biener <rguenther@suse.de>
3686 PR tree-optimization/83089
3687 * tree-if-conv.c (pass_if_conversion::execute): If anything
3688 changed reset SCEV and free the number of iteration estimates.
3690 2017-11-21 Martin Sebor <msebor@redhat.com>
3692 PR tree-optimization/82945
3693 * calls.h (warn_nonstring_bound): Remove unused function.
3695 2017-11-21 Martin Sebor <msebor@redhat.com>
3697 PR tree-optimization/82945
3698 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
3699 * calls.h (maybe_warn_nonstring_arg): Declare new function.
3700 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
3702 (initialize_argument_information): Call maybe_warn_nonstring_arg.
3703 * calls.h (get_attr_nonstring_decl): Declare new function.
3704 * doc/extend.texi (attribute nonstring): Update.
3705 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
3706 get_attr_nonstring_decl and handle it.
3707 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
3708 detection of nul-termination.
3709 (strlen_to_stridx): Change to a pointer.
3710 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
3711 (pass_strlen::execute): Same.
3713 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
3715 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
3716 for the new option -mprefer-vector-width=[none|128|256|512].
3717 * config/i386/i386.c (ix86_target_string): remove old style options
3718 -mprefer-avx256 and make -mprefer-avx128 as alias.
3719 (ix86_option_override_internal): Apply defaults for the
3720 -mprefer-vector-width=[128|256] option.
3721 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
3722 Implement macros to work with -mprefer-vector-width=.
3723 * config/i386/i386.opt: Implemented option
3724 -mprefer-vector-width=[none|128|256|512].
3725 * doc/invoke.texi: Documentation for
3726 -mprefer-vector-width=[none|128|256|512].
3728 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
3730 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
3732 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
3734 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
3736 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
3738 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
3740 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
3741 * expr.c (expand_expr_real_2): Likewise.
3742 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
3743 fold_binary_loc): Likewise.
3744 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
3745 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
3746 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
3747 MINUS_EXPR transformations.
3748 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
3749 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
3750 * tree-inline.c (estimate_operator_cost): Likewise.
3751 * tree-pretty-print.c (dump_generic_node, op_code_prio,
3752 op_symbol_code): Likewise.
3753 * tree-vect-stmts.c (vectorizable_operation): Likewise.
3754 * vr-values.c (extract_range_from_binary_expr): Likewise.
3755 * varasm.c (initializer_constant_valid_p_1): Likewise.
3756 * tree.def: New tree code POINTER_DIFF_EXPR.
3758 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
3760 * config/i386/i386.md (*bswap<mode>2_movbe): Add
3761 integer suffix to movbe mnemonic.
3762 (*bswaphi2_movbe): Ditto.
3763 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
3765 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3768 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
3769 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
3770 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
3772 2017-11-21 Martin Liska <mliska@suse.cz>
3774 * tree-inline.c (expand_call_inline): Remove not needed
3777 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
3778 Jakub Jelinek <jakub@redhat.com>
3781 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
3782 Remove static keyword from f variable.
3784 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3786 PR tree-optimization/83086
3787 * gimple-ssa-store-merging.c
3788 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
3789 rather than n.base_addr.
3791 2017-11-21 Martin Liska <mliska@suse.cz>
3793 PR rtl-optimization/82044
3794 PR tree-optimization/82042
3795 * dse.c (check_mem_read_rtx): Check for overflow.
3797 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3799 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
3802 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3804 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
3805 and its corresponding call arg location note.
3807 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3808 Andrew Burgess <andrew.burgess@embecosm.com>
3810 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
3812 (arc_return_slot_offset): Likewise.
3813 (arc_eh_return_address_location): New declaration.
3814 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
3815 (MUST_SAVE_REGISTER): Add exception handler case.
3816 (MUST_SAVE_RETURN_ADDR): Likewise.
3817 (arc_frame_pointer_required): Likewise.
3818 (arc_frame_pointer_needed): New function.
3819 (arc_compute_frame_size): Changed.
3820 (arc_expand_prologue): Likewise.
3821 (arc_expand_epilogue): Likewise.
3822 (arc_initial_elimination_offset): Likewise.
3823 (arc_return_slot_offset): Delete.
3824 (arc_eh_return_address_location): New function.
3825 (arc_builtin_setjmp_frame_value): Likewise.
3826 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
3827 (EH_RETURN_STACKADJ_RTX): Define.
3828 (EH_RETURN_HANDLER_RTX): Likewise.
3829 * config/arc/arc.md (eh_return): Delete.
3831 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3833 * print-tree.h (debug_vec_tree): Remove prototype.
3834 * gdbinit.in (pvt): Remove macro.
3836 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3838 PR tree-optimization/83047
3839 * gimple-ssa-store-merging.c
3840 (imm_store_chain_info::output_merged_store): If the loads with the
3841 same vuse are in different basic blocks, for load_gsi pick a load
3842 location that is dominated by the other loads.
3845 * config/i386/i386.c (ix86_memmodel_check): Start
3846 -Winvalid-memory-model diagnostics with lowercase letter.
3849 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
3850 set in_cold_section_p to the partition containing loc_list->first.
3851 When seeing loc_list->last_before_switch node, update secname and
3852 perform range_across_switch second partition handling only after that.
3855 * run-rtl-passes.c: Include debug.h.
3856 (run_rtl_passes): Call debug_hooks->assembly_start.
3857 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
3861 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
3862 OPTAB_DIRECT in calls to expand_simple_binop.
3864 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3867 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
3869 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3872 * spellcheck.h (best_match::blithely_get_best_candidate): New
3875 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3877 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
3878 with lower case letter.
3880 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
3882 * config/i386/i386.md (bswaphi2): New expander.
3883 (*bswaphi2_movbe): New insn pattern.
3884 (bswaphi -> rorhi pepehole2): New peephole pattern.
3886 2017-11-20 Jeff Law <law@redhat.com>
3888 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
3889 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
3890 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
3891 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
3892 gimple-ssa-evrp-analyze.h.
3894 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
3895 set BB_VISITED here.
3896 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
3898 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3900 * tree-predcom.c: Add general comment on Store-Store chains.
3901 (split_data_refs_to_components): Postpone clearing eliminate_store_p
3903 (get_chain_last_ref_at): Rename into...
3904 (get_chain_last_write_at): ...this.
3905 (get_chain_last_write_before_load): New function.
3906 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
3907 CT_STORE_STORE when write reference is added.
3908 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
3909 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
3910 (initialize_root_vars_store_elim_1): Ditto.
3911 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
3912 definition is created.
3913 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
3914 chain by replacing it with dominant stored value.
3916 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3918 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
3920 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
3922 * vr-values.c (extract_range_from_binary_expr): Use a full range
3925 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
3927 * config/arm/arm.md (R4_REGNUM): Define constant.
3928 (nonsecure_call_internal): Remove r4 clobber.
3929 (nonsecure_call_value_internal): Likewise.
3930 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3931 clobber and resequence match_operands.
3932 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3933 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3934 (nonsecure_call_value_reg_thumb2): Likewise.
3936 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3938 PR tree-optimization/78821
3939 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3940 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3941 to the address of the base rather than the base itself.
3942 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3943 (find_bswap_or_nop_finalize): New function.
3944 (find_bswap_or_nop): Use it.
3945 (bswap_replace): Return a tree rather than bool, change first
3946 argument from gimple * to gimple_stmt_iterator, allow inserting
3947 into an empty sequence, allow ins_stmt to be NULL - then emit
3948 all stmts into gsi. Fix up MEM_REF address gimplification.
3949 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3950 (struct store_immediate_info): Add N and INS_STMT non-static
3952 (store_immediate_info::store_immediate_info): Initialize them
3953 from newly added ctor args.
3954 (merged_store_group::apply_stores): Formatting fixes. Sort by
3956 (stmts_may_clobber_ref_p): For stores call also
3957 refs_anti_dependent_p.
3958 (gather_bswap_load_refs): New function.
3959 (imm_store_chain_info::try_coalesce_bswap): New method.
3960 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3961 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3962 (imm_store_chain_info::output_merged_store): Fail if number of
3963 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3964 and NOP_EXPR rhs_code.
3965 (pass_store_merging::process_store): Compute n and ins_stmt, if
3966 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3967 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3969 (pass_store_merging::execute): Calculate dominators.
3971 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3972 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3973 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3974 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3975 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3976 class pass_optimize_bswap, bswap_replace,
3977 pass_optimize_bswap::execute): Moved to ...
3978 * gimple-ssa-store-merging.c: ... this file.
3979 Include optabs-tree.h.
3980 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3981 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3982 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3983 anonymous namespace, remove static keywords.
3984 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3985 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3987 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3990 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3992 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3994 * vec.h (debug_helper): New function.
3995 (DEFINE_DEBUG_VEC): New macro.
3996 * hash-set.h (debug_helper): New function.
3997 (DEFINE_DEBUG_HASH_SET): New macro.
3998 * cfg.c (debug_slim (edge)): New function.
3999 Call DEFINE_DEBUG_VEC for edges.
4000 Call DEFINE_DEBUG_HASH_SET for edges.
4001 * cfghooks.c (debug_slim (basic_block)): New function.
4002 Call DEFINE_DEBUG_VEC for basic blocks.
4003 Call DEFINE_DEBUG_HASH_SET for basic blocks.
4004 * print-tree.c (debug_slim): New function to handle trees.
4005 Call DEFINE_DEBUG_VEC for trees.
4006 Call DEFINE_DEBUG_HASH_SET for trees.
4007 (debug (vec<tree, va_gc>) &): Remove.
4008 (debug (<vec<tree, va_gc>) *): Remove.
4009 * print-rtl.c (debug_slim): New function to handle const_rtx.
4010 Call DEFINE_DEBUG_VEC for rtx_def.
4011 Call DEFINE_DEBUG_VEC for rtx_insn.
4012 Call DEFINE_DEBUG_HASH_SET for rtx_def.
4013 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
4014 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
4015 (debug (vec<rtx_insn *> *ptr): Remove.
4016 (debug_insn_vector): Remove.
4017 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
4019 2017-11-20 Tom de Vries <tom@codesourcery.com>
4021 PR rtl-optimization/82020
4022 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
4023 IF_THEN_ELSE condition.
4025 2017-11-19 Jeff Law <law@redhat.com>
4027 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
4028 of degenerates resulting from ignoring an edge.
4030 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4033 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
4035 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4038 * profile-count.c (profile_count::to_sreal_scale): Fix return value
4039 for uninitialied counts.
4041 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4044 * tree-inline.c (estimate_num_insns): Set to 1 at least.
4046 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
4049 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
4051 2017-11-19 Tom de Vries <tom@codesourcery.com>
4053 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
4056 2017-11-19 Tom de Vries <tom@codesourcery.com>
4058 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
4061 2017-11-19 Tom de Vries <tom@codesourcery.com>
4063 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
4066 2017-11-19 Tom de Vries <tom@codesourcery.com>
4068 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
4070 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
4072 2017-11-19 Tom de Vries <tom@codesourcery.com>
4074 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
4076 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
4077 semicolon after MCORE_EXPORT_NAME call.
4079 2017-11-19 Tom de Vries <tom@codesourcery.com>
4082 * vmsdbgout.c (vmsdbgout_early_finish): New function.
4083 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
4085 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
4087 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
4088 (cgraph_edge::clone): Cleanup updating of profile.
4089 * ipa-cp.c (update_profiling_info): Likewise.
4090 * ipa-inline-transform.c (inline_transform): Likewise.
4091 * ipa-inline.c (inline_small_functions): Add missing space to dump.
4092 * ipa-split.c (execute_split_functions): Do not split when function
4094 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
4095 * profile-count.c (profile_count::dump): Add global0.
4096 (profile_count::to_cgraph_frequency): Do not ICE when entry is
4098 (profile_count::to_sreal_scale): Likewise.
4099 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
4100 (profile_count::combine_with_ipa_count): New function.
4101 * profile-count.h (profile_guessed_global0adjusted): New.
4102 (profile_count::adjusted_zero): New.
4103 (profile_count::global0adjusted): New.
4104 (profile_count::combine_with_ipa_count): New.
4105 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
4106 correct profile of return block of split functions.
4107 (copy_cfg_body): Remove unused profile_count.
4108 (copy_body): Likewise.
4109 (expand_call_inline): Update.
4110 (tree_function_versioning): Update.
4112 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
4114 * hash-set.h (hash_set::empty): New.
4115 * tree-ssa-threadbackward.h: Delete.
4116 * tree-ssa-threadbackward.c (class thread_jumps): New.
4117 Move max_threaded_paths into class.
4118 (fsm_find_thread_path): Remove arguments that are now in class.
4119 (profitable_jump_thread_path): Rename to...
4120 (thread_jumps::profitable_jump_thread_path): ...this.
4121 (convert_and_register_jump_thread_path): Rename to...
4122 (thread_jumps::convert_and_register_current_path): ...this.
4123 (check_subpath_and_update_thread_path): Rename to...
4124 (thread_jumps::check_subpath_and_update_thread_path): ...this.
4125 (register_jump_thread_path_if_profitable): Rename to...
4126 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
4127 (handle_phi): Rename to...
4128 (thread_jumps::handle_phi): ...this.
4129 (handle_assignment): Rename to...
4130 (thread_jumps::handle_assignment): ...this.
4131 (fsm_find_control_statement_thread_paths): Rename to...
4132 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
4133 (find_jump_threads_backwards): Rename to...
4134 (thread_jumps::find_jump_threads_backwards): ...this.
4135 Initialize path local data.
4136 (pass_thread_jumps::execute): Call find_jump_threads_backwards
4137 from within thread_jumps class.
4138 (pass_early_thread_jumps::execute): Same.
4140 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4142 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
4144 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4146 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
4147 to canonical location.
4149 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4152 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
4154 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
4156 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
4158 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
4159 rcrt1.o%s/grcrt1.o%s for -static-pie.
4161 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4163 * i386.c (ix86_multiplication_cost, ix86_division_cost,
4164 ix86_shift_rotate_cost): Break out from ...
4165 (ix86_rtx_costs): ... here.
4166 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
4169 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4171 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
4172 when entry block was promoted unlikely.
4173 (estimate_bb_frequencies): Increase frequency scale.
4174 * profile-count.h (profile_count): Export precision info.
4176 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4178 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
4179 disturbin profile of entry block.
4181 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4183 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
4186 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4188 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
4189 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
4191 2017-11-17 Jeff Law <law@redhat.com>
4193 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
4194 from evrp_dom_walker class. Various methods moved into new class.
4195 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
4196 (evrp_range_analyzer::enter): New method.
4197 (evrp_range_analyzer::leave): New method.
4198 (evrp_dom_walker): Remove delegators no longer needed by this class.
4199 Replace vr_values data member with evrp_range_analyzer
4201 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
4202 method extracted from evrp_dom_walker::before_dom_children.
4203 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
4204 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
4206 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
4207 Add private copy constructor and move assignment operators.
4208 Privatize methods and class data where trivially possible.
4209 (evrp_dom_walker::cleanup): New function, extracted from
4210 execute_early_vrp. Simplify access to class data.
4212 * vr-values.h (get_output_for_vrp): Prototype.
4213 * vr-values.c (get_output_for_vrp): New function extracted from
4214 vrp_visit_assignment_or_call and extract_range_from_stmt.
4215 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
4217 2017-11-17 Luis Machado <luis.machado@linaro.org>
4219 * config/aarch64/aarch64.c
4220 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
4221 (qdf24xx_tunings) <autoprefetcher_model>: Set to
4222 tune_params::AUTOPREFETCHER_WEAK.
4224 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4227 * config/arm/arm.c (arm_valid_target_attribute_rec):
4228 Parse "arch=" and "+<ext>".
4229 (arm_valid_target_attribute_tree): Re-init global options.
4230 (arm_option_override): Make non-static.
4231 (arm_options_perform_arch_sanity_checks): Make errors fatal.
4232 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
4233 (__ARM_FEATURE_CRC32): Support undef.
4234 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
4235 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
4237 2017-11-17 David Malcolm <dmalcolm@redhat.com>
4239 * gdbinit.in (break-on-diagnostic): New command.
4241 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4243 * config/i386/i386.c (ix86_expand_epilogue): Change simple
4244 return to indirect jump for EH return if control-flow
4245 protection is enabled. Change explicit 'false' argument in
4246 pro_epilogue_adjust_stack with a value of
4248 * config/i386/i386.md (simple_return_indirect_internal):
4249 Remove SImode restriction to support 64-bit.
4251 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
4253 * combine.c (added_notes_insn): New.
4254 (try_combine): Handle added_notes_insn like added_links_insn.
4255 Rewrite return value code.
4256 (distribute_notes): Set added_notes_insn to the earliest insn we added
4259 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
4261 PR rtl-optimization/82621
4262 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
4263 dest of one of those SETs is unused.
4265 2017-11-17 Richard Biener <rguenther@suse.de>
4268 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
4269 * tree-pretty-print.c (dump_generic_node): Handle
4270 annot_expr_parallel_kind.
4271 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
4272 * gimplify.c (gimple_boolify): Likewise.
4274 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4276 * config.gcc (extra_headers): Add cet.h for x86 targets.
4277 * config/i386/cet.h: New file.
4278 * doc/install.texi: Add --enable-cet/--disable-cet.
4280 2017-11-17 Richard Biener <rguenther@suse.de>
4282 PR tree-optimization/83017
4283 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
4284 (gen_parallel_loop): Properly count iterations.
4285 (parallelize_loops): Handle loop->can_be_parallel independent
4286 of flag_loop_parallelize_all. Make static profitability test match
4288 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
4289 * invoke.texi (parloops-min-per-thread): Document.
4291 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
4293 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
4294 upstreaming review comments.
4296 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4298 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
4299 with fast unaligned access.
4300 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
4302 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
4304 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
4305 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
4307 2017-11-17 Richard Biener <rguenther@suse.de>
4309 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
4310 folding of references.
4312 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
4315 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
4316 of replacing call to strncmp with corresponding call to strcmp when
4319 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
4321 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
4322 option prefer-avx256 for skylake-avx512 configuration.
4323 * config/i386/i386.c (ix86_option_override_internal): Ditto.
4324 (get_builtin_code_for_version): Ditto.
4326 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4327 Monk Chiang <sh.chiang04@gmail.com>
4329 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
4330 (FIXED_REGISTERS): Reserve more register numbers.
4331 (CALL_USED_REGISTERS): Likewise.
4332 (REG_ALLOC_ORDER): Likewise.
4333 (REG_CLASS_CONTENTS): Likewise.
4334 (REGISTER_NAMES): Likewise.
4336 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4337 Kito Cheng <kito.cheng@gmail.com>
4339 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
4341 * config/nds32/iterators.md: Add vector mode iterators and attributes.
4343 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
4345 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
4346 parameter list for vec_splats.
4348 2017-11-16 Joseph Myers <joseph@codesourcery.com>
4350 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
4352 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
4354 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
4357 * doc/install.texi: Document incompatibility of Perl >=5.6.26
4358 with the required version of automake 1.11.6.
4360 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
4362 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
4364 (DU_C2_3_power9): Correct reservation combinations.
4365 (FP_DIV_power9, VEC_DIV_power9): New.
4366 (power9-alu): Split out rotate/shift...
4367 (power9-rot): ...to here, correct dispatch resource.
4368 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
4370 (power9-fp): Correct latency.
4371 (power9-sdiv): Add div/sqrt resource.
4372 (power9-ddiv): Correct latency, add div/sqrt resource.
4373 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
4374 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
4376 (power9-qpdiv, power9-qpmul): Adjust resource usage.
4378 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
4380 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
4381 switch statement mapping KF built-ins to TF built-ins if we don't
4382 have the proper ISA 3.0 assembler support.
4384 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4386 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
4387 (lower_emutls_function_body): Do not compute it.
4389 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4391 * ipa-split.c (split_bb_info): Turn time to sreal.
4392 (split_point): Likewise.
4393 (dump_split_point): Likewise.
4394 (fine_split_points): Likewise.
4395 (execute_split_functions): Only zero split_bbs; turn time to sreals.
4397 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4399 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
4402 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4404 * predict.c (combine_predictions_for_bb): Preserve zero predicted
4406 (expensive_function_p): Remove useless assert.
4407 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
4409 2017-11-16 Martin Sebor <msebor@redhat.com>
4411 PR tree-optimization/82588
4412 PR tree-optimization/82583
4413 * tree-vrp.c (check_array_ref): Handle flexible array members,
4414 string literals, and inner indices.
4415 (search_for_addr_array): Add detail to diagnostics.
4417 2017-11-16 Nathan Sidwell <nathan@acm.org>
4421 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
4422 (SET_DECL_ASSEMBLER_NAME): Forward to
4423 overwrite_decl_assembler_name.
4424 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
4425 (overwrite_decl_assembler_name): Declare.
4426 * tree.c (overwrite_decl_assembler_name): New.
4427 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
4428 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
4429 (LANG_HOOKS_INITIALIZER): Add it.
4430 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
4431 * langhooks.c (lhd_set_decl_assembler_name): Use
4432 SET_DECL_ASSEMBLER_NAME.
4433 (lhd_overwrite_decl_assembler_name): Default implementation.
4435 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4436 Jackson Woodruff <jackson.woodruff@arm.com>
4438 PR tree-optimization/71026
4439 * match.pd: Canonicalize constant multiplies in division.
4441 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4443 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
4444 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
4445 Remove OPT_fomit_frame_pointer entry.
4446 * common/config/arc/arc-common.c: Likewise.
4447 * common/config/arm/arm-common.c: Likewise.
4448 * common/config/avr/avr-common.c: Likewise.
4449 * common/config/c6x/c6x-common.c: Likewise.
4450 * common/config/cr16/cr16-common.c: Likewise.
4451 * common/config/cris/cris-common.c: Likewise.
4452 * common/config/epiphany/epiphany-common.c: Likewise.
4453 * common/config/fr30/fr30-common.c: Likewise.
4454 * common/config/frv/frv-common.c: Likewise.
4455 * common/config/ia64/ia64-common.c: Likewise.
4456 * common/config/iq2000/iq2000-common.c: Likewise.
4457 * common/config/lm32/lm32-common.c: Likewise.
4458 * common/config/m32r/m32r-common.c: Likewise.
4459 * common/config/mcore/mcore-common.c: Likewise.
4460 * common/config/microblaze/microblaze-common.c: Likewise.
4461 * common/config/mips/mips-common.c: Likewise.
4462 * common/config/mmix/mmix-common.c: Likewise.
4463 * common/config/mn10300/mn10300-common.c: Likewise.
4464 * common/config/nios2/nios2-common.c: Likewise.
4465 * common/config/pa/pa-common.c: Likewise.
4466 * common/config/pdp11/pdp11-common.c: Likewise.
4467 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
4468 * common/config/riscv/riscv-common.c: Likewise.
4469 * common/config/rs6000/rs6000-common.c: Likewise.
4470 * common/config/rx/rx-common.c: Likewise.
4471 * common/config/s390/s390-common.c: Likewise.
4472 * common/config/sh/sh-common.c: Likewise.
4473 * common/config/sparc/sparc-common.c: Likewise.
4474 * common/config/tilegx/tilegx-common.c: Likewise.
4475 * common/config/tilepro/tilepro-common.c: Likewise.
4476 * common/config/v850/v850-common.c: Likewise.
4477 * common/config/visium/visium-common.c: Likewise.
4478 * common/config/xstormy16/xstormy16-common.c: Likewise.
4479 * common/config/xtensa/xtensa-common.c: Likewise.
4480 * invoke.texi (-fomit-frame-pointer): Update documentation.
4482 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4484 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
4486 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4488 * cfg.c (scale_bbs_frequencies_int,
4489 cale_bbs_frequencies_gcov_type): Remove.
4490 * cfg.h (scale_bbs_frequencies_int,
4491 cale_bbs_frequencies_gcov_type): Remove.
4493 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4495 * tree-ssa-loop-manip.c
4496 (scale_dominated_blocks_in_loop): Update to profile counts.
4497 (tree_transform_and_unroll_loop): Likewise.
4499 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4501 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
4502 scale_bbs_frequencies_int.
4504 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4506 * final.c (compute_alignments): Use counts rather than frequencies.
4508 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4510 * cfgloopanal.c: Include sreal.h
4511 (average_num_loop_insns): Use counts and sreal for accounting.
4513 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4515 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
4518 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4519 Kito Cheng <kito.cheng@gmail.com>
4521 * config/nds32/constraints.md: Provide more constraints.
4522 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
4523 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
4524 support constraints usage.
4526 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4528 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
4530 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4531 Kito Cheng <kito.cheng@gmail.com>
4533 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
4534 * config/nds32/nds32.opt: Refine the layout.
4535 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
4536 TARGET_EXT_STRING): Support new options.
4537 * config/nds32/nds32.h: Likewise.
4538 * config/nds32/nds32.md: Likewise.
4539 * config/nds32/nds32-predicates.c: Likewise.
4540 * config/nds32/constraints.md: Likewise.
4541 * common/config/nds32/nds32-common.c: Likewise.
4543 2017-11-16 Julia Koval <julia.koval@intel.com>
4546 * config/i386/gfniintrin.h: Add sse check.
4547 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
4549 2017-11-16 Julia Koval <julia.koval@intel.com>
4551 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
4552 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
4553 (ix86_handle_option): Handle -mavx512vbmi2.
4554 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
4555 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4556 * config/i386/i386-c.c (__AVX512VBMI2__): New.
4557 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
4558 (ix86_valid_target_attribute_inner_p): Ditto.
4559 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
4560 * config/i386/i386.opt (mavx512vbmi2): New option.
4561 * doc/invoke.texi: Add new option.
4563 2017-11-16 Julia Koval <julia.koval@intel.com>
4565 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
4566 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
4567 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
4568 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
4569 _mm512_gf2p8mul_epi8): New intrinsics.
4570 * config/i386/i386-builtin-types.def
4571 (V64QI_FTYPE_V64QI_V64QI): New type.
4572 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
4573 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
4574 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
4575 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
4576 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
4577 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4579 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
4581 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
4582 explicitly as a stream of bytes.
4584 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4586 * config/rs6000/altivec.h (vec_xst_be): New #define.
4587 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
4588 and externalize from *altivec_vperm_<mode>_internal.
4589 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
4591 (XL_BE_V8HI): Likewise.
4592 (XL_BE_V4SI): Likewise.
4593 (XL_BE_V4SI): Likewise.
4594 (XL_BE_V2DI): Likewise.
4595 (XL_BE_V4SF): Likewise.
4596 (XL_BE_V2DF): Likewise.
4597 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
4598 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
4599 all array entries with these keys: VSX_BUILTIN_VEC_XL,
4600 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
4601 VSX_BUILTIN_VEC_XST_BE.
4602 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
4603 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
4605 (altivec_init_builtins): Replace conditional calls to def_builtin
4606 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
4607 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
4608 with unconditional calls. Remove calls to def_builtin for
4609 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
4610 __builtin_vec_xst_be.
4611 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
4612 to define_expand, and add alternate RTL generation for P8.
4613 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
4614 vsx_ld_elemrev_v8hi.
4615 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
4616 add alternate RTL generation for P8.
4617 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
4618 vsx_ld_elemrev_v16qi.
4619 (vsx_st_elemrev_v8hi): Convert define_insn
4620 to define_expand, and add alternate RTL generation for P8.
4621 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
4622 vsx_st_elemrev_v8hi.
4623 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
4624 add alternate RTL generation for P8.
4625 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
4626 vsx_st_elemrev_v16qi.
4628 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
4631 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
4632 TARGET_AVX512ER check.
4633 (ix86_option_override_internal): Set MASK_VZEROUPPER if
4634 neither -mzeroupper nor -mno-zeroupper is used and
4635 TARGET_EMIT_VZEROUPPER is set.
4636 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
4637 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
4639 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
4641 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4642 folding of vector compares.
4643 (fold_build_vec_cmp): New helper function.
4644 (fold_compare_helper): New helper function.
4645 (builtin_function_type): Add compare builtins to the list of functions
4646 having unsigned arguments. Cosmetic updates to comment indentation.
4647 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
4648 the not+eq combination.
4650 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4652 PR tree-optimization/82726
4653 PR tree-optimization/70754
4654 * tree-predcom.c (order_drefs_by_pos): New function.
4655 (combine_chains): Move code setting has_max_use_after to...
4656 (try_combine_chains): ...here. New parameter. Sort combined chains
4657 according to position information.
4658 (tree_predictive_commoning_loop): Update call to above function.
4659 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
4661 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4663 PR tree-optimization/82726
4665 2017-01-23 Bin Cheng <bin.cheng@arm.com>
4667 PR tree-optimization/70754
4668 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
4669 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
4670 combined stmt before it if not NULL.
4671 (combine_chains): Process refs reversely and compute dominance point
4675 2017-02-23 Bin Cheng <bin.cheng@arm.com>
4677 PR tree-optimization/79663
4678 * tree-predcom.c (combine_chains): Process refs in reverse order
4679 only for ZERO length chains, and add explaining comment.
4681 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4683 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
4684 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
4685 * doc/invoke.texi (ARM Options): Add armv8.3-a.
4687 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4689 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
4691 2017-11-15 Martin Liska <mliska@suse.cz>
4693 * tree-cfg.c (pass_warn_function_return::execute):
4694 Compare warn_return_type for greater than zero.
4696 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
4700 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
4701 to return true on Xeon and not on Xeon Phi.
4702 (ix86_check_avx256_register): Changed to ...
4703 (ix86_check_avx_upper_register): ... this. Add extra check for
4704 VALID_AVX512F_REG_OR_XI_MODE.
4705 (ix86_avx_u128_mode_needed): Changed
4706 ix86_check_avx256_register to ix86_check_avx_upper_register.
4707 (ix86_check_avx256_stores): Changed to ...
4708 (ix86_check_avx_upper_stores): ... this. Changed
4709 ix86_check_avx256_register to ix86_check_avx_upper_register.
4710 (ix86_avx_u128_mode_after): Changed
4711 avx_reg256_found to avx_upper_reg_found. Changed
4712 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
4713 (ix86_avx_u128_mode_entry): Changed
4714 ix86_check_avx256_register to ix86_check_avx_upper_register.
4715 (ix86_avx_u128_mode_exit): Ditto.
4716 * config/i386/i386.h: (host_detect_local_cpu): New define.
4718 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4720 * config/arm/xgene1.md (xgene1): Split into automatons
4721 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
4722 (xgene1_f_load): Adjust reservations and/or types.
4723 (xgene1_f_store): Likewise.
4724 (xgene1_load_pair): Likewise.
4725 (xgene1_store_pair): Likewise.
4726 (xgene1_fp_load1): Likewise.
4727 (xgene1_load1): Likewise.
4728 (xgene1_store1): Likewise.
4729 (xgene1_move): Likewise.
4730 (xgene1_alu): Likewise.
4731 (xgene1_simd): Likewise.
4732 (xgene1_bfm): Likewise.
4733 (xgene1_neon_load1): Likewise.
4734 (xgene1_neon_store1): Likewise.
4735 (xgene1_neon_logic): Likewise.
4736 (xgene1_neon_st1): Likewise.
4737 (xgene1_neon_ld1r): Likewise.
4738 (xgene1_alu_cond): Added.
4739 (xgene1_shift_reg): Likwise.
4740 (xgene1_bfx): Likewise.
4741 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
4743 2017-11-15 Jakub Jelinek <jakub@redhat.com>
4746 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
4748 (can_widen_mult_without_libcall): New function.
4749 (expand_mul_overflow): If only checking unsigned mul overflow,
4750 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
4751 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
4752 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
4753 (expand_DIVMOD): Formatting fix.
4754 * expmed.h (expand_mult): Add NO_LIBCALL argument.
4755 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
4756 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
4758 PR tree-optimization/82977
4759 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
4760 constructed temporary to strlen_to_stridx.put.
4762 2017-11-15 Martin Liska <mliska@suse.cz>
4764 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
4765 * configure: Regenerate.
4767 2017-11-15 Martin Liska <mliska@suse.cz>
4770 * config/sh/sh-mem.cc: Use proper probability for
4773 2017-11-14 Jeff Law <law@redhat.com>
4775 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
4776 the red zone for stack_clash_protection_final_dynamic_probe targets
4777 when the total dynamic stack size is zero bytes.
4779 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
4780 blocks is post order.
4782 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
4784 * dumpfile.h (TDF_COMPARE_DEBUG): New.
4785 * final.c (rest_of_clean_state): Set it for the
4786 -fcompare-debug dump.
4787 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
4788 class when TDF_COMPARE_DEBUG is set.
4790 * dwarf2out.c (gen_producer_string): Discard
4793 2017-11-15 Joseph Myers <joseph@codesourcery.com>
4796 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
4797 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
4798 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
4799 (__TGMATH_REAL_2_3): Remove macros.
4800 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
4801 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
4803 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
4805 (remquo): Define using __TGMATH_REAL_3.
4807 2017-11-14 Jeff Law <law@redhat.com>
4809 * vr-values.c: New file with contents extracted from tree-vrp.c.
4810 * Makefile.in (OBJS): Add vr-values.o
4811 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
4812 (set_value_range, set_and_canonicalize_value_range): Likewise.
4813 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
4814 (value_range_constant_singleton, symbolic_range_p): Likewise.
4815 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
4816 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
4817 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
4818 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
4819 (find_case_label_range, find_case_label_index): Likewise.
4820 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
4821 (range_int_cst_singleton_p, value_inside_range): Likewise.
4822 (get_single_symbol): Likewise.
4823 (switch_update): Move structure definition here.
4824 (to_remove_edges, to_update_switch_stmts): Provide externs.
4825 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
4826 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
4827 (vrp_val_is_min, set_value_range): Likewise.
4828 (set_and_canonicalize_value_range, copy_value_range): Likewise.
4829 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
4830 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
4831 (range_is_nonnull, range_int_cst_p): Likewwise.
4832 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
4833 (get_single_symbol, operand_less_p): Likewise
4834 (compare_values_warnv, compare_values): Likewise.
4835 (value_inside_range, value_range_constant_singleton): Likewise.
4836 (zero_nonzero_bitgs_from_vr): Likewise.
4837 (extract_range_from_binary_expr_1): Likewise.
4838 (overflow_comparison_p): Likewise.
4839 (to_remove_edges, to_update_switch_stmts): Likewise.
4840 (find_case_label-index, find_case_label_range): Likewise.
4841 (switch_update, set_value_range_to_nonnegative): Remove.
4842 (set_value_range_to_truthvalue): Likewise.
4843 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
4844 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
4845 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
4846 (find_case_label_ranges, test_for_singularity): Likewise.
4847 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
4848 (x_vr_values): Move to its remaining use site.
4850 2017-11-10 Jeff Law <law@redhat.com>
4852 * vr-values.h (VR_INITIALIZER): Move #define here.
4853 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
4854 * Makefile.in (OBJS): Add tree-evrp.o
4855 * tree-vrp.h (assert_info): Move structure definition here.
4856 (set_value_range_to_varying): Prototype.
4857 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4858 (infer_value_range, register_edge_assert_for): Likewise.
4859 (stmt_interesting_for_vrp): Likewise.
4860 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
4861 (set_value_range_to_varying): No longer static.
4862 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4863 (infer_value_range, register_edge_assert_for): Likewise.
4865 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4867 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
4868 generate the XXBRD instruction.
4870 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
4871 (rs6000_builtin_type_compatible): Treat _Float128 and long double
4872 as being compatible if -mabi=ieeelongdouble.
4873 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
4874 to setup float128 built-ins with hardware support.
4875 (BU_FLOAT128_HW_2): Likewise.
4876 (BU_FLOAT128_HW_3): Likewise.
4877 (BU_FLOAT128_HW_VSX_1): Likewise.
4878 (BU_FLOAT128_HW_VSX_2): Likewise.
4879 (scalar_extract_expq): Change float128 built-in functions to
4880 accommodate having both KFmode and TFmode functions. Use the
4881 KFmode variant as the default.
4882 (scalar_extract_sigq): Likewise.
4883 (scalar_test_neg_qp): Likewise.
4884 (scalar_insert_exp_q): Likewise.
4885 (scalar_insert_exp_qp): Likewise.
4886 (scalar_test_data_class_qp): Likewise.
4887 (sqrtf128_round_to_odd): Delete processing the round to odd
4888 built-in functions as special built-in functions, and define them
4889 as float128 built-ins. Use the KFmode variant as the default.
4890 (truncf128_round_to_odd): Likewise.
4891 (addf128_round_to_odd): Likewise.
4892 (subf128_round_to_odd): Likewise.
4893 (mulf128_round_to_odd): Likewise.
4894 (divf128_round_to_odd): Likewise.
4895 (fmaf128_round_to_odd): Likewise.
4896 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
4897 support for KFmode and TFmode xststdcqp calls.
4898 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
4899 point, switch the built-in handlers for the get/set float128
4900 exponent, get float128 mantissa, float128 test built-ins, and the
4901 float128 round to odd built-in functions. Eliminate creating the
4902 float128 round to odd built-in functions as special built-ins.
4903 (rs6000_init_builtins): Eliminate special creation of the float128
4904 round to odd built-in functions.
4905 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
4906 function insns to support both TFmode and KFmode variants.
4907 (xsxsigqp_<mode>): Likewise.
4908 (xsiexpqpf_<mode>): Likewise.
4909 (xsiexpqp_<mode>): Likewise.
4910 (xststdcqp_<mode>): Likewise.
4911 (xststdcnegqp_<mode>): Likewise.
4912 (xststdcqp_<mode>): Likewise.
4914 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4916 * tree-ssa-threadupdate.c (compute_path_counts): Remove
4917 unused path_in_freq_ptr parameter.
4918 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
4920 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4922 * ipa-inline.c (edge_badness): Dump sreal frequency.
4923 (compute_inlined_call_time): Match natural implementaiton ...
4924 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
4925 forgotten division by CGRAPH_FREQ_BASE.
4927 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4929 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4930 Solaris 11. Update comment.
4931 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4933 * config/sol2.h (STARTFILE_SPEC): Likewise.
4934 * configure: Regenerate.
4936 2017-11-14 Carl Love <cel@us.ibm.com>
4938 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4939 le_ and be_ prefixes to swap* variables. Remove
4940 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4943 2017-11-14 Jason Merrill <jason@redhat.com>
4945 Support GTY((cache)) on hash_map.
4946 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4947 (ggc_cache_remove): Override it instead of ggc_mx.
4948 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4949 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4950 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4951 (simple_cache_map_traits): Override maybe_mx.
4952 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4953 (hash_map): Friend gt_cleare_cache.
4954 (gt_cleare_cache): New.
4955 * tree.h (tree_cache_traits): New hash_map traits class.
4956 (tree_cache_map): New typedef.
4958 2017-11-14 Richard Biener <rguenther@suse.de>
4960 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4961 paramter and handling.
4962 (cleanup_control_flow_bb): Likewise.
4963 (cleanup_control_flow_pre): New helper performing a DFS walk
4964 to call cleanup_control_flow_bb in PRE order.
4965 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4966 via cleanup_control_flow_pre.
4968 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4970 * config/aarch64/aarch64-simd.md
4971 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4972 (*aarch64_simd_bsl<mode>_alt): Likewise.
4973 (aarch64_simd_bsldi_internal): New.
4974 (aarch64_simd_bsldi_alt): Likewise.
4976 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4978 * tracer.c (better_p): Do not compare frequencies.
4979 * reg-stack.c (better_edge): Likewise.
4980 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4983 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4985 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4986 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4987 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4988 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4989 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4990 * predict.c (maybe_hot_frequency_p): Inline to ...
4991 (maybe_hot_count_p): ... here; rewrite to counts.
4992 (counts_to_freqs): Rename to ...
4993 (update_max_bb_count): ... this one.
4994 (expensive_function_p): Use counts.
4995 (estimate_bb_frequencies): Update.
4996 (rebuild_frequencies): Update.
4997 * predict.h (counts_to_freqs): Rename to ...
4998 (update_max_bb_count): ... this one.
4999 * profile.c (compute_branch_probabilities): Add debug info
5000 * tree-inline.c (expand_call_inline): Update debug info.
5001 (optimize_inline_calls): Use update_max_bb_count..
5002 (tree_function_versioning): Use update_max_bb_count..
5003 * value-prof.c (gimple_value_profile_transformations):
5004 Do not use update_max_bb_count.
5006 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5008 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
5009 always use frequencies.
5011 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5013 * bb-reorder.c: Remove frequencies from comments.
5014 (better_edge_p): Use profile counts.
5015 (find_traces): Dump profile counts.
5016 (rotate_loop): Use profile counts.
5017 (find_traces_1_round): Likewise.
5018 (connect_better_edge_p): Use counts instead of probabilities for
5020 (copy_bb_p): Drop early check for non-0 frequency.
5021 (sanitize_hot_paths): Update comments.
5023 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5025 * ipa-split.c (struct split_point): Add count.
5026 (consider_split): Do not compute incoming frequency; compute incoming
5027 count and store it to split_point.
5028 (split_function): Set count of the call to split part correctly.
5030 2017-11-13 Carl Love <cel@us.ibm.com>
5032 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
5034 2017-11-13 Tom Tromey <tom@tromey.com>
5036 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
5038 2017-11-13 Carl Love <cel@us.ibm.com>
5040 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
5041 Add support for builtins:
5042 unsigned int vec_first_{,miss}_match_{,or_eos}index,
5043 vector {un,}signed {char,int,short},
5044 vector {un,}signed {char,int,short}) arguments.
5045 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
5046 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
5047 Add BU_P9V_AV_2 expansions for the builtins.
5048 * config/rs6000/altivec.h (vec_first_match_index,
5049 vec_first_mismatch_index, vec_first_match_or_eos_index,
5050 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
5051 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
5052 new extern declaration.
5053 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
5054 * config/rs6000/vsx.md (first_match_index_<mode>,
5055 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
5056 first_mismatch_or_eos_index_<mode>): Add define expand.
5057 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
5058 * doc/extend.texi: Update the built-in documenation file for the new
5061 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
5063 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
5064 into the min/max operations for _Float<N> and _Float<N>X types.
5066 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5069 * dwarf2out.c (do_eh_frame): New static variable.
5070 (dwarf2out_begin_prologue): Set it.
5071 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
5073 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5075 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
5077 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
5079 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5082 * debug.h (dwarf2out_do_eh_frame): Declare.
5083 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
5084 (dwarf2out_do_frame): Use it.
5085 (dwarf2out_do_cfi_asm): Likewise.
5086 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
5087 (dwarf2out_assembly_start): Likewise.
5088 (dwarf2out_begin_prologue): Fix comment.
5089 * toplev.c (compile_file): Always call dwarf2out_frame_finish
5090 if the target needs either debug or unwind DWARF2 info.
5091 * lto-opts.c (lto_write_options): Do not save -fexceptions,
5092 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
5093 -ftrapping-math, -ftrapv and -fwrapv.
5095 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5097 * cgraph.c (cgraph_edge::sreal_frequency): New function.
5098 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
5099 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
5100 (estimate_edge_size_and_time): Likewise.
5101 (ipa_merge_fn_summary_after_inlining): Likewise.
5102 * ipa-inline.c (cgraph_freq_base_rec): Remove.
5103 (compute_uninlined_call_time): Use sreal_frequency.
5104 (compute_inlined_call_time): Likewise.
5105 (ipa_inline): Do not initialize cgraph_freq_base_rec.
5106 * profile-count.c: Include sreal.h.
5107 (profile_count::to_sreal_scale): New.
5108 * profile-count.h: Forward declare sreal.
5109 (profile_count::to_sreal_scale): Declare.
5111 2017-11-13 Nathan Sidwell <nathan@acm.org>
5113 * diagnostic.c (maybe_line_and_column): New.
5114 (diagnostic_get_location_text): Use it.
5115 (diagnostic_report_current_module): Likewise.
5116 (test_diagnostic_get_location_text): Add tests.
5118 2017-11-13 Luis Machado <luis.machado@linaro.org>
5120 * doc/md.texi (Specifying processor pipeline description): Fix
5121 incorrect latency for the div instruction example.
5123 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5125 PR tree-optimization/78821
5126 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
5127 that bit_not_p is the same.
5128 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
5129 (split_group): Count precisely bit_not_p bits in each statement.
5130 (invert_op): New function.
5131 (imm_store_chain_info::output_merged_store): Use invert_op to
5132 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
5133 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
5135 2017-11-13 Martin Liska <mliska@suse.cz>
5137 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
5138 (struct source_info): Likewise.
5139 (add_branch_counts): Likewise.
5140 (add_line_counts): Likewise.
5141 (function_summary): Likewise.
5142 (output_intermediate_line): Likewise.
5143 (generate_results): Likewise.
5145 2017-11-13 Martin Liska <mliska@suse.cz>
5147 * gcov.c (struct block_info): Remove typedef for block_t.
5148 (struct line_info): Likewise.
5149 (line_info::has_block): Likewise.
5150 (EXIT_BLOCK): Likewise.
5151 (unblock): Likewise.
5152 (circuit): Likewise.
5153 (get_cycles_count): Likewise.
5154 (process_file): Likewise.
5155 (read_graph_file): Likewise.
5156 (solve_flow_graph): Likewise.
5157 (find_exception_blocks): Likewise.
5158 (add_line_counts): Likewise.
5159 (accumulate_line_info): Likewise.
5160 (output_line_details): Likewise.
5162 2017-11-13 Martin Liska <mliska@suse.cz>
5164 * gcov.c (struct arc_info): Remove typedef for arc_t.
5165 (struct line_info): Likewise.
5166 (add_branch_counts): Likewise.
5167 (output_branch_count): Likewise.
5168 (function_info::~function_info): Likewise.
5169 (circuit): Likewise.
5170 (output_intermediate_line): Likewise.
5171 (read_graph_file): Likewise.
5172 (solve_flow_graph): Likewise.
5173 (find_exception_blocks): Likewise.
5174 (add_line_counts): Likewise.
5175 (accumulate_line_info): Likewise.
5176 (output_line_details): Likewise.
5177 (output_function_details): Likewise.
5179 2017-11-13 Martin Liska <mliska@suse.cz>
5181 * gcov.c (struct function_info): Remove typedef for function_t.
5182 (struct source_info): Likewise.
5183 (source_info::get_functions_at_location): Likewise.
5184 (solve_flow_graph): Likewise.
5185 (find_exception_blocks): Likewise.
5186 (add_line_counts): Likewise.
5187 (output_intermediate_file): Likewise.
5188 (process_file): Likewise.
5189 (generate_results): Likewise.
5190 (release_structures): Likewise.
5191 (read_graph_file): Likewise.
5192 (read_count_file): Likewise.
5193 (accumulate_line_counts): Likewise.
5194 (output_lines): Likewise.
5196 2017-11-13 Martin Liska <mliska@suse.cz>
5198 * gcov.c (function_info::function_info): Remove num_counts
5199 and add vector<gcov_type>.
5200 (function_info::~function_info): Use the vector.
5201 (process_file): Likewise.
5202 (read_graph_file): Likewise.
5203 (read_count_file): Likewise.
5204 (solve_flow_graph): Likewise.
5206 2017-11-13 Martin Liska <mliska@suse.cz>
5208 * gcov.c (function_info::is_artificial): New function.
5209 (process_file): Erase all artificial early.
5210 (generate_results): Skip as all artificial are already
5213 2017-11-13 Martin Liska <mliska@suse.cz>
5215 * gcov.c (read_graph_file): Store to global vector of functions.
5216 (read_count_file): Iterate the vector.
5217 (process_file): Likewise.
5218 (generate_results): Likewise.
5219 (release_structures): Likewise.
5221 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5223 PR tree-optimization/82954
5224 * gimple-ssa-store-merging.c
5225 (imm_store_chain_info::coalesce_immediate_stores): If
5226 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
5228 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
5230 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
5231 Upddate call to ENDIAN_LANE_N.
5232 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
5233 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
5234 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
5235 and use aarch64_endian_lane_rtx.
5236 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
5238 2017-11-12 Tom de Vries <tom@codesourcery.com>
5240 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
5242 2017-11-12 Tom de Vries <tom@codesourcery.com>
5244 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
5246 2017-11-12 Tom de Vries <tom@codesourcery.com>
5248 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
5250 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
5251 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
5252 * defaults.h (ASM_OUTPUT_LABELREF): Same.
5254 2017-11-11 Martin Sebor <msebor@redhat.com>
5257 * doc/extend.texi (attribute nonstring): Remove spurious argument.
5260 * prefic.c (translate_name): Replace strncpy with memcpy to
5261 avoid -Wstringop-truncation.
5263 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5265 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
5267 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5269 * predict.c (maybe_hot_frequency_p): Do not use cfun.
5271 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5273 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
5276 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
5278 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
5279 enable generating XXBRH if the value is in a vector register.
5280 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
5281 value is in a vector register.
5282 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
5283 register to register bswap64's instead of doing the GPR sequence
5284 used on previous machines.
5285 (bswapdi2_xxbrd): New insn.
5286 (bswapdi2_reg): Disallow on ISA 3.0.
5287 (register to register bswap64 splitter): Do not split the insn on
5288 ISA 3.0 systems that use XXBRD.
5290 2017-11-10 Martin Sebor <msebor@redhat.com>
5293 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
5295 (find_subframework_file): Same.
5297 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5299 * auto-profile.c (afdo_indirect_call): Drop frequency.
5300 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
5301 (cgraph_node::create_edge): Drop frequency argument.
5302 (cgraph_node::create_indirect_edge): Drop frequency argument.
5303 (cgraph_edge::make_speculative): Drop frequency arguments.
5304 (cgraph_edge::resolve_speculation): Do not update frequencies
5305 (cgraph_edge::dump_edge_flags): Do not dump frequency.
5306 (cgraph_node::dump): Check consistency in IPA mode.
5307 (cgraph_edge::maybe_hot_p): Use IPA counter.
5308 (cgraph_edge::verify_count_and_frequency): Rename to ...
5309 (cgraph_edge::verify_count): ... this one; drop frequency checking.
5310 (cgraph_node::verify_node): Update.
5311 * cgraph.h (struct cgraph_edge): Drop frequency.
5312 (cgraph_edge::frequency): New function.
5313 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
5315 (cgraph_edge::rebuild_edges): Likewise.
5316 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
5317 (duplicate_thunk_for_node): Do not pass frequency.
5318 (cgraph_node::create_clone): Scale only counts.
5319 (cgraph_node::create_virtual_clone): Do not pass frequency.
5320 (cgraph_node::create_edge_including_clones): Do not pass frequency.
5321 (cgraph_node::create_version_clone): Do not pass frequency.
5322 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
5323 (cgraph_node::expand_thunk): Do not pass frequency.
5324 (cgraph_node::create_wrapper): Do not pass frequency.
5325 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
5327 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
5328 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
5329 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
5330 (gather_caller_stats): Use frequency function.
5331 (ipcp_cloning_candidate_p): Use frequency function.
5332 (ipcp_propagate_stage): Use frequency function.
5333 (get_info_about_necessary_edges): Use frequency function.
5334 (update_profiling_info): Update only IPA profile.
5335 (update_specialized_profile): Use frequency functoin.
5336 (perhaps_add_new_callers): Update only IPA profile.
5337 * ipa-devirt.c (ipa_devirt): Use IPA profile.
5338 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
5339 (dump_ipa_call_summary): Use frequency function.
5340 (estimate_edge_size_and_time): Use frequency function.
5341 (ipa_merge_fn_summary_after_inlining): Use frequency function.
5342 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
5343 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
5344 (update_noncloned_counts): ... ths one; scale counts only.
5345 (clone_inlined_nodes): Do not scale frequency.
5346 (inline_call): Do not pass frequency.
5347 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
5348 (compute_inlined_call_time): Use IPA profile.
5349 (want_inline_small_function_p): Use IPA profile.
5350 (want_inline_self_recursive_call_p): Use IPA profile.
5351 (edge_badness): Use IPA profile.
5352 (lookup_recursive_calls): Use IPA profile.
5353 (recursive_inlining): Do not pass frequency.
5354 (resolve_noninline_speculation): Do not update frequency.
5355 (inline_small_functions): Collect max of IPA profile.
5356 (dump_overall_stats): Dump IPA porfile.
5357 (dump_inline_stats): Dump IPA porfile.
5358 (ipa_inline): Collect IPA stats.
5359 * ipa-inline.h (clone_inlined_nodes): Update prototype.
5360 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
5361 (ipa_propagate_frequency): Use frequency function.
5362 (ipa_profile): Cleanup.
5363 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
5364 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
5365 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
5366 (input_node): Do not stream frequency.
5367 (input_edge): Do not stream frequency.
5368 (merge_profile_summaries): Scale only IPA profiles.
5369 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
5370 * predict.c (drop_profile): Do not recompute frequency.
5371 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
5372 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
5373 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
5374 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
5375 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
5376 * tree-inline.c (copy_bb): Do not scale frequency.
5377 (expand_call_inline): Do not scale frequency.
5378 (tree_function_versioning): Do not scale frequency.
5379 * ubsan.c (ubsan_create_edge): Do not pass frequency.
5381 2017-11-10 Julia Koval <julia.koval@intel.com>
5383 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
5384 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
5385 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
5386 (_mm256_maskz_gf2p8affine_epi64_epi8)
5387 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
5388 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
5389 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
5390 (__builtin_ia32_vgf2p8affineqb_v32qi)
5391 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
5392 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
5394 2017-11-10 Tamar Christina <tamar.christina@arm.com>
5398 (arm_option_override): Refactor.
5399 (arm_option_reconfigure_globals): New.
5400 (arm_options_perform_arch_sanity_checks): New.
5401 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
5403 (arm_options_perform_arch_sanity_checks): Likewise
5405 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
5407 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
5408 (power9-qpmul): New.
5409 * rs6000/rs6000.md ("type" attr): Add qmul.
5410 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
5411 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
5412 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
5414 2017-11-10 Martin Sebor <msebor@redhat.com>
5417 * builtins.c (compute_objsize): Handle arrays that
5418 compute_builtin_object_size likes to fail for. Make extern.
5419 * builtins.h (compute_objsize): Declare.
5420 (check_strncpy_sizes): New function.
5421 (expand_builtin_strncpy): Call check_strncpy_sizes.
5422 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
5423 -Wstringop-truncation.
5424 (gimple_fold_builtin_strncat): Same.
5425 * gimple.c (gimple_build_call_from_tree): Set call location.
5426 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
5427 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
5428 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
5429 (handle_builtin_strlen): Use strlen_to_stridx.
5430 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
5432 Use strlen_to_stridx.
5433 (pass_strlen::execute): Release strlen_to_stridx.
5434 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
5435 (-Wstringop-truncation): Document new option.
5437 2017-11-10 Martin Liska <mliska@suse.cz>
5439 PR gcov-profile/82702
5440 * gcov.c (main): Handle intermediate files in a different
5442 (get_gcov_intermediate_filename): New function.
5443 (output_gcov_file): Remove support of intermediate files.
5444 (generate_results): Allocate intermediate file.
5445 (release_structures): Clean-up properly fn_end.
5446 (output_intermediate_file): Start iterating with line 1.
5448 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5450 PR tree-optimization/82929
5451 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5452 ops_swapped_p non-static data member.
5453 (store_immediate_info::store_immediate_info): Clear it.
5454 (imm_store_chain_info::coalesce_immediate_stores): If swapping
5455 ops set ops_swapped_p.
5456 (count_multiple_uses): Handle ops_swapped_p.
5458 2017-11-10 Martin Liska <mliska@suse.cz>
5460 * coverage.c (coverage_init): Stream information about
5461 support of has_unexecuted_blocks.
5462 * doc/gcov.texi: Document that.
5463 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
5464 * gcov.c (read_graph_file): Likewise.
5465 (output_line_beginning): Fix a small issue with
5468 2017-11-10 Bin Cheng <bin.cheng@arm.com>
5470 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
5471 reference of trivial component.
5473 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5476 * gimple-ssa-store-merging.c
5477 (pass_store_merging::terminate_all_aliasing_chains): For
5478 gimple_store_p stmts also call refs_output_dependent_p.
5480 PR rtl-optimization/82913
5481 * compare-elim.c (try_merge_compare): Punt if def_insn is not
5484 2017-11-09 Jeff Law <law@redhat.com>
5486 * vr-values.h: New file with vr_values class.
5487 * tree-vrp.c: Include vr-values.h
5488 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
5489 data objects into the vr_values class.
5490 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
5491 (get_value_range): Make it a member function within vr_values class.
5492 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
5493 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
5494 (op_with_constant_singleton_value_range): Likewise.
5495 (extract_range_for_var_from_comparison_expr): Likewise.
5496 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
5497 (extract_range_from_binary_expr): Likewise.
5498 (extract_range_from_unary_expr): Likewise.
5499 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
5500 (check_for_binary_op_overflow, extract_range_basic): Likewise.
5501 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
5502 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
5503 (compare_name_with_value, compare_names): Likewise.
5504 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
5505 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
5506 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
5507 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
5508 (extract_range_from_phi_node): Likewise.
5509 (simplify_truth_ops_using_ranges): Likewise.
5510 (simplify_div_or_mod_using_ranges): Likewise.
5511 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
5512 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
5513 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
5514 (simplify_float_conversion_using_ranges): Likewise.
5515 (simplify_internal_call_using_ranges): Likewise.
5516 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
5517 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
5518 poitner via x_vr_values for calls into gimple folder.
5519 (vrp_initialize_lattice): Make this the vr_values ctor.
5520 (vrp_free_lattice): Make this the vr_values dtor.
5521 (set_vr_value): New function.
5522 (class vrp_prop): Add vr_values data member. Add various member
5523 functions as well as member functions that delegate to vr_values.
5524 (check_array_ref): Make a member function within vrp_prop class.
5525 (search_for_addr_array, vrp_initialize): Likewise.
5526 (vrp_finalize): Likewise. Revamp to avoid direct access to
5527 vr_value, values_propagated, etc.
5528 (check_array_bounds): Extract vrp_prop class instance pointer from
5529 walk info structure. Use it to call member functions.
5530 (check_all_array_refs): Make a member function within vrp_prop class.
5531 Smuggle class instance pointer via walk info structure.
5532 (x_vr_values): New local static.
5533 (vrp_valueize): Use x_vr_values to get class instance.
5534 (vr_valueize_1): Likewise.
5535 (class vrp_folder): Add vr_values data member. Add various member
5536 functions as well as member functions that delegate to vr_values.
5537 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
5538 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
5539 class instance from vr_values. Use it to call member functions.
5540 (vrp_dom_walker): Add vr_values data member.
5541 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
5542 instance via x_vr_values.
5543 (identify_jump_threads): Accept vr_values as argument. Store
5544 it into the walker structure.
5545 (evrp_dom_walker): Add vr_values class data member. Add various
5547 (evrp_dom_walker::try_find_new_range): Use vr_values data
5548 member to access the memory allocator.
5549 (evrp_dom_walker::before_dom_children): Store vr_values class
5550 instance into the vrp_folder class.
5551 (evrp_dom_walker::push_value_range): Rework to avoid direct
5552 access to num_vr_values and vr_value.
5553 (evrp_dom_walker::pop_value_range): Likewise.
5554 (execute_early_vrp): Remove call to vrp_initialize_lattice.
5555 Use vr_values to get to dump_all_value_ranges member function.
5556 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
5557 and simplify_cond_using_ranges_2 via vrp_prop class instance.
5558 Pass vr_values class instance down to identify_jump_threads.
5559 Remove call to vrp_free_lattice.
5560 (debug_all_value_ranges): Remove.
5562 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
5563 (vrp_folder): Likewise.
5565 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
5566 Get it from the existing bitmap instead.
5567 (vrp_intersect_ranges_1): Likewise.
5569 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5571 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5573 (store_immediate_info::store_immediate_info): Add bitnotp argument,
5574 set bit_not_p to it.
5575 (imm_store_chain_info::coalesce_immediate_stores): Break group
5576 if bit_not_p is different.
5577 (count_multiple_uses, split_group,
5578 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
5579 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
5580 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
5581 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
5582 to store_immediate_info ctor.
5584 2017-11-09 Jim Wilson <jimw@sifive.com>
5586 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
5587 (scan_prog_file): Likewise.
5589 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5591 * bb-reorder.c (max_entry_frequency): Remove.
5592 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
5593 connect_traces, push_to_next_round_p): Remove prototypes.
5594 (find_traces_1_round): Use counts only.
5595 (push_to_next_round_p): Likewise.
5596 (find_traces): Likewise.
5597 (rotate_loop): Likewise.
5598 (find_traces_1_round): Likewise.
5599 (connect_traces): Likewise.
5600 (edge_order): Likewise.
5602 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
5604 * config/arm/arm.c (output_return_instruction): Add comments to
5605 indicate requirement for cmse_nonsecure_entry return to account
5606 for the size of clearing instruction output here.
5607 (thumb_exit): Likewise.
5608 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
5609 return in hardfloat mode.
5611 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
5613 * config/rs6000/rs6000.c (machine_function): Add a bool,
5614 "toc_is_wrapped_separately".
5615 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
5616 if it wasn't explicitly set or unset, we are optimizing for speed, and
5617 doing separate shrink-wrapping.
5618 (rs6000_get_separate_components): Enable the TOC component if
5619 saving the TOC register in the prologue.
5620 (rs6000_components_for_bb): Handle the TOC component.
5621 (rs6000_emit_prologue_components): Store the TOC register where needed.
5622 (rs6000_set_handled_components): Mark TOC as handled, if handled.
5623 (rs6000_emit_prologue): Don't save the TOC if that is already done.
5625 2017-11-09 Martin Jambor <mjambor@suse.cz>
5627 * ipa-param-manipulation.c: New file.
5628 * ipa-param-manipulation.h: Likewise.
5629 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
5630 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
5631 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
5632 (ipa_parm_adjustment): Likewise.
5633 (ipa_parm_adjustment_vec): Likewise.
5634 (ipa_get_vector_of_formal_parms): Moved declaration to
5635 ipa-param-manipulation.h.
5636 (ipa_get_vector_of_formal_parm_types): Likewise.
5637 (ipa_modify_formal_parameters): Likewise.
5638 (ipa_modify_call_arguments): Likewise.
5639 (ipa_combine_adjustments): Likewise.
5640 (ipa_dump_param_adjustments): Likewise.
5641 (ipa_modify_expr): Likewise.
5642 (ipa_get_adjustment_candidate): Likewise.
5643 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
5644 ipa-param-manipulation.c.
5645 (ipa_get_vector_of_formal_parm_types): Likewise.
5646 (ipa_modify_formal_parameters): Likewise.
5647 (ipa_modify_call_arguments): Likewise.
5648 (ipa_modify_expr): Likewise.
5649 (get_ssa_base_param): Likewise.
5650 (ipa_get_adjustment_candidate): Likewise.
5651 (index_in_adjustments_multiple_times_p): Likewise.
5652 (ipa_combine_adjustments): Likewise.
5653 (ipa_dump_param_adjustments): Likewise.
5654 * tree-sra.c: Also include ipa-param-manipulation.h
5655 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
5658 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5659 Alan Hayward <alan.hayward@arm.com>
5660 David Sherwood <david.sherwood@arm.com>
5662 * doc/sourcebuild.texi (vect_masked_store): Document.
5664 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5665 Alan Hayward <alan.hayward@arm.com>
5666 David Sherwood <david.sherwood@arm.com>
5668 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
5670 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5671 Alan Hayward <alan.hayward@arm.com>
5672 David Sherwood <david.sherwood@arm.com>
5674 * doc/sourcebuild.texi (vect_variable_length): Document.
5676 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5677 Alan Hayward <alan.hayward@arm.com>
5678 David Sherwood <david.sherwood@arm.com>
5680 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
5682 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5683 Alan Hayward <alan.hayward@arm.com>
5684 David Sherwood <david.sherwood@arm.com>
5686 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
5688 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5689 Alan Hayward <alan.hayward@arm.com>
5690 David Sherwood <david.sherwood@arm.com>
5692 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
5693 previously undocumented selectors.
5694 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
5696 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5698 * doc/rtl.texi (const_vector): Say that elements can be
5699 const_wide_ints too.
5700 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
5701 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
5702 (gen_vec_duplicate): Use it instead of CONSTANT_P.
5703 * optabs.c (expand_vector_broadcast): Likewise.
5705 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5706 Alan Hayward <alan.hayward@arm.com>
5707 David Sherwood <david.sherwood@arm.com>
5709 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
5710 scaled index even if the unscaled address was invalid.
5711 Don't increase the complexity of using a scale in that case.
5713 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5714 Alan Hayward <alan.hayward@arm.com>
5715 David Sherwood <david.sherwood@arm.com>
5717 * doc/rtl.texi: Rewrite the subreg rules so that they partition
5718 the inner register into REGMODE_NATURAL_SIZE bytes rather than
5719 UNITS_PER_WORD bytes.
5720 * emit-rtl.c (validate_subreg): Divide subregs into blocks
5721 based on REGMODE_NATURAL_SIZE of the inner mode.
5722 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
5723 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
5724 * expmed.c (lowpart_bit_field_p): Divide the value up into
5725 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
5726 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
5727 whether something is likely to occupy more than one register.
5729 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5732 * ipa-inline-transform.c (update_noncloned_frequencies): Use
5733 profile_count::adjust_for_ipa_scaling.
5734 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
5735 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
5737 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
5739 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5741 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
5742 (split_group): Add total_orig and total_new arguments, estimate the
5743 number of statements related to the store group without store merging
5744 and with store merging.
5745 (imm_store_chain_info::output_merged_store): Adjust split_group
5746 callers, punt if estimated number of statements with store merging
5747 is not smaller than estimated number of statements without it.
5749 (handled_load): Remove has_single_use checks.
5750 (pass_store_merging::process_store): Likewise.
5752 2017-11-09 Richard Biener <rguenther@suse.de>
5754 PR tree-optimization/82902
5755 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
5757 2017-11-09 Martin Liska <mliska@suse.cz>
5760 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
5763 2017-11-09 Martin Liska <mliska@suse.cz>
5765 PR tree-optimization/82669
5766 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
5768 2017-11-09 Martin Liska <mliska@suse.cz>
5770 PR gcov-profile/48463
5771 * coverage.c (coverage_begin_function): Output also end locus
5772 of a function and information whether the function is
5774 * gcov-dump.c (tag_function): Parse and print the information.
5775 * gcov.c (INCLUDE_MAP): Add include.
5776 (INCLUDE_SET): Likewise.
5777 (struct line_info): Move earlier in the source file because
5778 of vector<line_info> in function_info structure.
5779 (line_info::line_info): Likewise.
5780 (line_info::has_block): Likewise.
5781 (struct source_info): Add new member index.
5782 (source_info::get_functions_at_location): New function.
5783 (function_info::group_line_p): New function.
5784 (output_intermediate_line): New function.
5785 (output_intermediate_file): Use the mentioned function.
5786 (struct function_start): New.
5787 (struct function_start_pair_hash): Likewise.
5788 (process_file): Add code that identifies group functions.
5789 Assign lines either to global or function scope.
5790 (generate_results): Skip artificial functions.
5791 (find_source): Assign index for each source file.
5792 (read_graph_file): Read new flag artificial and end_line.
5793 (add_line_counts): Assign it either to global of function scope.
5794 (accumulate_line_counts): Isolate core of the function to
5795 accumulate_line_info and call it for both function and global
5797 (accumulate_line_info): New function.
5798 (output_line_beginning): Fix GNU coding style.
5799 (print_source_line): New function.
5800 (output_line_details): Likewise.
5801 (output_function_details): Likewise.
5802 (output_lines): Iterate both source (global) scope and function
5804 (struct function_line_start_cmp): New class.
5805 * doc/gcov.texi: Reflect changes in documentation.
5807 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5810 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
5811 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
5812 and similarly for not instead of neg.
5814 2017-11-08 Andi Kleen <ak@linux.intel.com>
5816 * config/i386/i386.opt: Add -mforce-indirect-call.
5817 * config/i386/predicates.md: Check for flag_force_indirect_call.
5818 * doc/invoke.texi: Document -mforce-indirect-call
5820 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
5822 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
5824 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
5825 config/riscv/riscv.c (predict.h): New include.
5826 (riscv_slow_unaligned_access_p): No longer static.
5827 (riscv_block_move_straight): Add require.
5828 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5830 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5833 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
5834 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
5835 nonimmediate_operand predicate for operand 1 instead of
5838 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5840 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
5842 * config/aarch64/constraints.md (Uml): New constraint.
5843 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
5846 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5848 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
5849 of two vec_duplicates into a vec_concat.
5851 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5853 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5854 Simplify vec_merge of vec_duplicate and vec_concat.
5855 * config/aarch64/constraints.md (Utq): New constraint.
5856 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
5859 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5861 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5862 Simplify vec_merge of vec_duplicate and const_vector.
5863 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
5865 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
5866 mode iterator. Update predicate on operand 1 to
5867 handle non-const_vec constants. Delete constraints.
5868 (*aarch64_combinez_be<mode>): Likewise for operand 2.
5870 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5872 PR tree-optimization/78821
5873 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
5875 (store_operand_info::store_operand_info): Initialize it to false.
5876 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
5877 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
5878 store in the group, and if chain_info is non-NULL, to ignore altogether
5880 (compatible_load_p): Fail if bit_not_p does not match.
5881 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
5882 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
5883 (pass_store_merging::process_store): Adjust
5884 terminate_all_aliasing_chains calls to pass NULL in all current spots,
5885 call terminate_all_aliasing_chains newly when adding a store into
5886 a chain with non-NULL chain_info.
5888 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5890 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
5892 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5894 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5896 (aarch64_layout_frame): Initialise emit_frame_chain.
5897 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
5898 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
5900 2017-11-08 Martin Liska <mliska@suse.cz>
5902 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
5903 of gimple_call_internal_p.
5905 2017-11-07 Tom de Vries <tom@codesourcery.com>
5907 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
5909 2017-11-07 Tom de Vries <tom@codesourcery.com>
5911 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5914 2017-11-08 Martin Liska <mliska@suse.cz>
5917 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
5919 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5921 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5923 2017-11-07 Jakub Jelinek <jakub@redhat.com>
5926 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
5927 (*cmp<mode>_ccz_1): New insn with $k alternative.
5930 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5931 RTX_COMM_COMPARE as commutative as well.
5932 (ix86_binary_operator_ok): Formatting fix.
5933 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5934 *<code><mode>3<mask_name><round_saeonly_name>,
5935 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5936 *<s>mul<mode>3_highpart<mask_name>,
5937 *vec_widen_umult_even_v16si<mask_name>,
5938 *vec_widen_umult_even_v8si<mask_name>,
5939 *vec_widen_umult_even_v4si<mask_name>,
5940 *vec_widen_smult_even_v16si<mask_name>,
5941 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5942 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5943 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5944 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5945 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5946 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5947 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5948 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5949 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5950 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5951 ix86_binary_operator_ok. Formatting fixes.
5952 (*<plusminus_insn><mode>3<mask_name><round_name>,
5953 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5956 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5958 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5959 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5960 eq and ne if TARGET_ISEL.
5961 (cmp): New code_iterator.
5962 (UNS, UNSU_, UNSIK): New code_attrs.
5963 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5964 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5966 ("eq<mode>3"): ... this.
5967 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5969 ("ne<mode>3"): ... this.
5971 2017-11-07 Julia Koval <julia.koval@intel.com>
5974 * common/config/i386/i386-common.c
5975 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5976 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5977 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5978 * config/i386/i386.opt: Ditto.
5979 * config/i386/i386.c (ix86_target_string): Ditto.
5980 (ix86_option_override_internal): Ditto.
5981 (ix86_init_mpx_builtins): Move MPX to args2.
5982 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5983 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5984 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5985 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5986 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5987 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5988 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5990 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5993 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5994 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5995 (zero-extendsidi peephole2): Remove peephole.
5997 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
6000 * stor-layout.c: Include attribs.h.
6001 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
6002 explicit lookup of "aligned" attribute.
6004 2017-11-07 Andrew Waterman <andrew@sifive.com>
6006 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
6007 (riscv_expand_block_move): Likewise.
6008 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
6010 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
6011 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
6012 * config/riscv/riscv.c (riscv_block_move_straight): New function.
6013 (riscv_adjust_block_mem): Likewise.
6014 (riscv_block_move_loop): Likewise.
6015 (riscv_expand_block_move): Likewise.
6016 * config/riscv/riscv.md (movmemsi): New pattern.
6018 2017-11-07 Michael Clark <michaeljclark@mac.com>
6020 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
6021 (MUSL_DYNAMIC_LINKER): Likewise.
6023 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
6025 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
6026 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
6029 2017-11-07 Richard Biener <rguenther@suse.de>
6031 * match.pd: Fix build.
6033 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
6034 Jackson Woodruff <jackson.woodruff@arm.com>
6036 PR tree-optimization/71026
6037 * match.pd: Canonicalize negate in division.
6039 2017-11-07 Sudakshina Das <sudi.das@arm.com>
6042 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
6044 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6046 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
6047 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
6049 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
6051 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
6052 non-scalar integral types.
6053 * match.pd (negate_expr_p): Handle MINUS_EXPR.
6054 (-(A-B), -(~A)): New transformations.
6056 2017-11-07 Tom de Vries <tom@codesourcery.com>
6058 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
6059 semicolon after "do {} while (0)".
6060 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6061 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6062 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6063 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6064 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6066 2017-11-07 Tom de Vries <tom@codesourcery.com>
6068 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
6069 after "do {} while (0)".
6070 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6071 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6072 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6073 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6074 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6076 2017-11-07 Tom de Vries <tom@codesourcery.com>
6079 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
6081 (arm_rtx_costs_internal): Add missing semicolon after
6082 HANDLE_NARROW_SHIFT_ARITH call.
6084 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6086 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6087 disable isel if it was not set explicitly.
6089 2017-11-06 James Bowman <james.bowman@ftdichip.com>
6091 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
6092 (add_type_attribute) likewise.
6094 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
6096 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
6097 of struct ix86_frame.
6098 (ix86_initial_elimination_offset): Likewise.
6099 (ix86_expand_split_stack_prologue): Likewise.
6101 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
6103 * tree-vrp.h (enum value_range_type): Update stale comment.
6105 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6106 Alan Hayward <alan.hayward@arm.com>
6107 David Sherwood <david.sherwood@arm.com>
6109 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
6110 (aarch64_expand_vec_perm_const): Take the number of units too.
6111 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
6112 (aarch64_expand_vec_perm_const): Likewise.
6113 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
6114 (vec_perm<mode>): Update accordingly.
6116 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6117 Alan Hayward <alan.hayward@arm.com>
6118 David Sherwood <david.sherwood@arm.com>
6120 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
6121 Take the number of units too.
6122 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
6123 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
6124 but check for a vector mode before rather than after the call.
6125 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
6126 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
6127 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
6128 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
6129 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
6130 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
6131 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
6132 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
6133 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
6134 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
6135 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
6136 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
6137 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
6138 (aarch64_sqdmull2_n<mode>): Update accordingly.
6140 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6141 Alan Hayward <alan.hayward@arm.com>
6142 David Sherwood <david.sherwood@arm.com>
6144 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
6145 the number of units too.
6146 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
6147 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
6148 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
6149 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
6150 (vec_store_lanesxi<mode>): Update accordingly.
6152 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6153 Alan Hayward <alan.hayward@arm.com>
6154 David Sherwood <david.sherwood@arm.com>
6156 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
6157 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
6158 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
6159 of units rather than the mode.
6160 * config/aarch64/iterators.md (nunits): New mode attribute.
6161 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
6162 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
6163 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
6164 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
6165 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
6166 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
6167 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
6168 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
6169 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
6170 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
6171 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6172 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
6173 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
6174 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
6175 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
6176 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
6177 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
6178 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
6179 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
6180 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
6181 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
6182 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
6183 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
6184 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6185 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
6186 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
6187 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6188 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
6189 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
6190 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
6191 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
6192 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
6193 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
6194 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
6195 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
6196 (aarch64_simd_vec_setv2di): Likewise.
6198 2017-11-06 Carl Love <cel@us.ibm.com>
6200 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
6202 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
6203 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
6205 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
6207 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
6208 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
6209 Add power 8 macro expansions.
6210 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
6211 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
6212 power 8 instructions. (VSX_XXBR): Add iterator.
6214 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
6216 * config/arm/arm.md (predicable_short_it): Change default to "no",
6217 improve documentation, remove uses that are identical to the default.
6218 (enabled_for_depr_it): Rename to enabled_for_short_it.
6219 * gcc/config/arm/arm-fixed.md (predicable_short_it):
6220 Remove default uses.
6221 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
6222 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
6223 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
6224 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
6226 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
6229 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
6230 float128 helper macros, which are no longer used after deleting
6231 the old 'q' built-in functions, and moving the round to odd
6232 built-in functions to being special built-in functions.
6233 (BU_FLOAT128_2): Likewise.
6234 (BU_FLOAT128_1_HW): Likewise.
6235 (BU_FLOAT128_2_HW): Likewise.
6236 (BU_FLOAT128_3_HW): Likewise.
6237 (FABSQ): Delete old 'q' built-in functions.
6238 (COPYSIGNQ): Likewise.
6239 (SQRTF128_ODD): Move round to odd built-in functions to be
6240 special built-in functions, so that we can handle
6241 -mabi=ieeelongdouble.
6242 (TRUNCF128_ODD): Likewise.
6243 (ADDF128_ODD): Likewise.
6244 (SUBF128_ODD): Likewise.
6245 (MULF128_ODD): Likewise.
6246 (DIVF128_ODD): Likewise.
6247 (FMAF128_ODD): Likewise.
6248 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
6249 built-in names to 'f128'.
6250 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
6251 old 'q' built-in functions, as the machine independent code for
6252 'f128' built-in functions handles this.
6253 (rs6000_expand_builtin): Add expansion for float128 round to odd
6254 functions, keying off on -mabi=ieeelongdouble of whether to use
6255 the KFmode or TFmode variant.
6256 (rs6000_init_builtins): Initialize the _Float128 round to odd
6258 * doc/extend.texi (PowerPC Built-in Functions): Document the old
6259 _Float128 'q' built-in functions are now mapped into the new
6260 'f128' built-in functions.
6262 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
6264 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
6265 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
6267 2017-11-06 David Malcolm <dmalcolm@redhat.com>
6270 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
6271 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
6272 * toplev.c: Include "ipa-fnsummary.h".
6273 (toplev::finalize): Call ipa_fnsummary_c_finalize.
6275 2017-11-06 Jakub Jelinek <jakub@redhat.com>
6277 PR tree-optimization/82838
6278 * gimple-ssa-store-merging.c
6279 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
6280 on a separate gimple_seq which is then appended to seq.
6282 2017-11-06 Jeff Law <law@redhat.com>
6285 * config/i386/i386.c (PROBE_INTERVAL): Remove.
6286 (get_probe_interval): New functions.
6287 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
6288 (ix86_adjust_stack_and_probe): Likewise.
6289 (output_adjust_stack_and_probe): Likewise.
6290 (ix86_emit_probe_stack_range): Likewise.
6291 (ix86_expand_prologue): Likewise.
6293 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6295 PR tree-optimization/82816
6296 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
6297 if the modes of the two types are the same.
6298 (convert_plusminus_to_widen): Likewise.
6300 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6302 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
6304 (usadv16qi): New define_expand.
6305 (usadv8hi): New define_expand.
6307 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
6310 * ipa-inline-transform.c (update_noncloned_frequencies): Always
6312 (inline_transform): Likewise.
6313 * predict.c (counts_to_freqs): Remove useless conditional.
6314 * profile-count.h (profile_count::apply_scale): Move sanity check.
6315 * tree-inline.c (copy_bb): Always scale.
6316 (copy_cfg_body): Likewise.
6318 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
6321 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
6323 (*cmp_ite0): Add enabled_for_depr_it attribute.
6324 (*cmp_ite1): Likewise.
6326 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6328 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
6331 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6333 * tree-vrp.c (vrp_int_const_binop): Return true on success and
6334 return the value by pointer.
6335 (extract_range_from_multiplicative_op_1): Update accordingly.
6336 Return as soon as an operation fails.
6338 2017-11-05 Tom de Vries <tom@codesourcery.com>
6341 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
6342 (DEF_SANITIZER_BUILTIN): ... here.
6343 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
6344 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
6346 2017-11-05 Tom de Vries <tom@codesourcery.com>
6349 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
6351 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
6352 ASM_OUTPUT_BEFORE_CASE_LABEL call.
6353 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
6355 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6356 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6357 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6359 2017-11-05 Tom de Vries <tom@codesourcery.com>
6362 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
6365 2017-11-04 Michael Clark <michaeljclark@mac.com>
6367 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
6368 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
6370 (*addsi3_extended): Likewise.
6371 (*addsi3_extended2): Likewise.
6372 (<optab>si3): Likewise.
6373 (<optab>di3): Likewise.
6374 (<optab><mode>3): Likewise.
6375 (<*optabe>si3_internal): Likewise.
6376 (zero_extendqi<SUPERQI:mode>2): Likewise.
6377 (*add<mode>hi3): Likewise.
6378 (*xor<mode>hi3): Likewise.
6379 (<optab>di3): Likewise.
6380 (*<optab>si3_extend): Likewise.
6381 (*sge<u>_<X:mode><GPR:mode>): Likewise.
6382 (*slt<u>_<X:mode><GPR:mode>): Likewise.
6383 (*sle<u>_<X:mode><GPR:mode>): Likewise.
6385 2017-11-04 Andrew Waterman <andrew@sifive.com>
6387 * config/riscv/riscv.c (riscv_option_override): Conditionally set
6388 TARGET_STRICT_ALIGN based upon -mtune argument.
6390 2017-11-04 Andrew Waterman <andrew@sifive.com>
6392 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
6394 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
6396 * config/i386/i386.c (choose_basereg): Use optional scratch
6397 register and add assertion.
6398 (x86_emit_outlined_ms2sysv_save): Use scratch register when
6399 needed, and don't allocate stack.
6400 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
6401 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
6402 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
6404 2017-11-03 Jeff Law <law@redhat.com>
6406 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
6407 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
6408 to probe at the start of a noreturn function.
6410 2017-11-03 Jakub Jelinek <jakub@redhat.com>
6412 PR tree-optimization/78821
6413 * gimple-ssa-store-merging.c: Update the file comment.
6414 (MAX_STORE_ALIAS_CHECKS): Define.
6415 (struct store_operand_info): New type.
6416 (store_operand_info::store_operand_info): New constructor.
6417 (struct store_immediate_info): Add rhs_code and ops data members.
6418 (store_immediate_info::store_immediate_info): Add rhscode, op0r
6419 and op1r arguments to the ctor, initialize corresponding data members.
6420 (struct merged_store_group): Add load_align_base and load_align
6422 (merged_store_group::merged_store_group): Initialize them.
6423 (merged_store_group::do_merge): Update them.
6424 (merged_store_group::apply_stores): Pick the constant for
6425 encode_tree_to_bitpos from one of the two operands, or skip
6426 encode_tree_to_bitpos if neither operand is a constant.
6427 (class pass_store_merging): Add process_store method decl. Remove
6428 bool argument from terminate_all_aliasing_chains method decl.
6429 (pass_store_merging::terminate_all_aliasing_chains): Remove
6430 var_offset_p argument and corresponding handling.
6431 (stmts_may_clobber_ref_p): New function.
6432 (compatible_load_p): New function.
6433 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
6434 if there is overlap and rhs_code is not INTEGER_CST. For
6435 non-overlapping stores terminate group if rhs is not mergeable.
6436 (get_alias_type_for_stmts): Change first argument from
6437 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
6438 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
6439 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
6441 (get_location_for_stmts): Change first argument from
6442 auto_vec<gimple *> & to vec<gimple *> &.
6443 (struct split_store): Remove orig_stmts data member, add orig_stores.
6444 (split_store::split_store): Create orig_stores rather than orig_stmts.
6445 (find_constituent_stmts): Renamed to ...
6446 (find_constituent_stores): ... this. Change second argument from
6447 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
6448 to info structures rather than the statements.
6449 (split_group): Rename ALLOW_UNALIGNED argument to
6450 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
6451 it. Adjust find_constituent_stores caller.
6452 (imm_store_chain_info::output_merged_store): Handle rhs_code other
6453 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
6454 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
6455 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
6456 (mem_valid_for_store_merging): New function.
6457 (handled_load): New function.
6458 (pass_store_merging::process_store): New method.
6459 (pass_store_merging::execute): Use process_store method. Adjust
6460 terminate_all_aliasing_chains caller.
6462 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6464 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
6465 Return true for more constants, symbols and label references.
6466 (aarch64_valid_floating_const): Remove unused function.
6468 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6471 * config/aarch64/aarch64.c (aarch64_layout_frame):
6472 Undo forcing of LR at bottom of frame.
6474 2017-11-03 Jeff Law <law@redhat.com>
6477 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
6478 for int_registers_saved.
6480 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
6481 extracted from tree-ssa-dom.c.
6482 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
6483 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
6484 (record_equivalences_from_incoming_edge): Add additional argument
6485 to single_pred_edge_ignoring_loop_edges call.
6486 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
6487 (uncprop_dom_walker::before_dom_children): Add additional argument
6488 to single_pred_edge_ignoring_loop_edges call.
6489 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
6490 single_pred_edge_ignoring_loop_edges rather than open coding.
6491 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
6493 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
6495 * match.pd (-(-A)): Rewrite.
6497 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6499 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
6500 (rs6000_emit_int_cmove): New declaration.
6501 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
6502 (rs6000_emit_sISEL): Delete.
6503 (rs6000_emit_int_cmove): Make non-static.
6504 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
6505 instead of rs6000_emit_sISEL.
6507 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
6509 * asan.c (create_cond_insert_point): Maintain profile.
6510 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
6512 * basic-block.h (struct basic_block_def): Remove frequency.
6513 (EDGE_FREQUENCY): Use to_frequency
6514 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
6516 (find_traces): Update to use to_frequency.
6517 (find_traces_1_round): Likewise; use only IPA counts.
6518 (bb_to_key): Likewise.
6519 (connect_traces): Use IPA counts only.
6520 (copy_bb_p): Update to use to_frequency.
6521 (fix_up_crossing_landing_pad): Likewise.
6522 (sanitize_hot_paths): Likewise.
6523 * bt-load.c (basic_block_freq): Likewise.
6524 * cfg.c (init_flow): Set count_max to uninitialized.
6525 (check_bb_profile): Remove frequencies; check counts.
6526 (dump_bb_info): Do not dump frequencies.
6527 (update_bb_profile_for_threading): Update counts only.
6528 (scale_bbs_frequencies_int): Likewise.
6529 (MAX_SAFE_MULTIPLIER): Remove.
6530 (scale_bbs_frequencies_gcov_type): Update counts only.
6531 (scale_bbs_frequencies_profile_count): Update counts only.
6532 (scale_bbs_frequencies): Update counts only.
6533 * cfg.h (struct control_flow_graph): Add count-max.
6534 (update_bb_profile_for_threading): Update prototype.
6535 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
6536 (find_many_sub_basic_blocks): Likewise.
6537 * cfgcleanup.c (try_forward_edges): Likewise.
6538 (try_crossjump_to_edge): Likewise.
6539 * cfgexpand.c (expand_gimple_cond): Likewise.
6540 (expand_gimple_tailcall): Likewise.
6541 (construct_init_block): Likewise.
6542 (construct_exit_block): Likewise.
6543 * cfghooks.c (verify_flow_info): Check consistency of counts.
6544 (dump_bb_for_graph): Do not dump frequencies.
6545 (split_block_1): Do not update frequencies.
6546 (split_edge): Do not update frequencies.
6547 (make_forwarder_block): Do not update frequencies.
6548 (duplicate_block): Do not update frequencies.
6549 (account_profile_record): Do not update frequencies.
6550 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
6551 for global heuristics.
6552 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
6553 (expected_loop_iterations_unbounded): Use counts only.
6554 * cfgloopmanip.c (scale_loop_profile): Simplify.
6555 (create_empty_loop_on_edge): Simplify
6557 (duplicate_loop_to_header_edge): Simplify
6558 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
6559 (update_br_prob_note): Take care of removing note when profile
6561 (relink_block_chain): Do not dump frequency.
6562 (rtl_account_profile_record): Use to_frequency.
6563 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
6564 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
6565 (cgraph_update_edges_for_call_stmt_node): Likewise.
6566 (cgraph_edge::verify_count_and_frequency): Update.
6567 (cgraph_node::verify_node): Temporarily disable frequency verification.
6568 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
6569 to_cgraph_frequency.
6570 (cgraph_edge::rebuild_edges): Convert to ipa counts.
6571 * cgraphunit.c (init_lowered_empty_function): Do not initialize
6573 (cgraph_node::expand_thunk): Update profile.
6574 * except.c (dw2_build_landing_pads): Do not update frequency.
6575 * final.c (compute_alignments): Use to_frequency.
6576 (dump_basic_block_info): Do not dump frequency.
6577 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
6578 (dump_gimple_bb_header): Do not dump frequency.
6579 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
6581 * gimple-streamer-in.c (input_bb): Do not stream frequency.
6582 * gimple-streamer-out.c (output_bb): Do not stream frequency.
6583 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
6584 (init_before_recovery): Do not update frequency.
6585 (sched_create_recovery_edges): Do not update frequency.
6586 * hsa-gen.c (convert_switch_statements): Do not update frequency.
6587 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
6588 (ipa_cp_c_finalize): Set max_count to uninitialized.
6589 * ipa-fnsummary.c (get_minimal_bb): Use counts.
6590 (param_change_prob): Use counts.
6591 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
6593 * ipa-split.c (consider_split): Use to_frequency.
6594 (split_function): Use to_frequency.
6595 * ira-build.c (loop_compare_func): Likewise.
6596 (mark_loops_for_removal): Likewise.
6597 (mark_all_loops_for_removal): Likewise.
6598 * loop-doloop.c (doloop_modify): Do not update frequency.
6599 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
6601 * lto-streamer-in.c (input_function): Update count_max.
6602 * omp-expand.c (expand_omp_taskreg): Update count_max.
6603 * omp-simd-clone.c (simd_clone_adjust): Update profile.
6604 * predict.c (maybe_hot_frequency_p): Use to_frequency.
6605 (maybe_hot_count_p): Use ipa counts only.
6606 (maybe_hot_bb_p): Simplify.
6607 (maybe_hot_edge_p): Simplify.
6608 (probably_never_executed): Do not take frequency argument.
6609 (probably_never_executed_bb_p): Do not pass frequency.
6610 (probably_never_executed_edge_p): Likewise.
6611 (combine_predictions_for_bb): Check that profile is nonzero.
6612 (propagate_freq): Do not set frequency.
6613 (drop_profile): Simplify.
6614 (counts_to_freqs): Simplify.
6615 (expensive_function_p): Use to_frequency.
6616 (propagate_unlikely_bbs_forward): Simplify.
6617 (determine_unlikely_bbs): Simplify.
6618 (estimate_bb_frequencies): Add hack to silence graphite issues.
6619 (compute_function_frequency): Use ipa counts.
6620 (pass_profile::execute): Update.
6621 (rebuild_frequencies): Use counts only.
6622 (force_edge_cold): Use counts only.
6623 * profile-count.c (profile_count::dump): Dump new count types.
6624 (profile_count::differs_from_p): Check compatiblity.
6625 (profile_count::to_frequency): New function.
6626 (profile_count::to_cgraph_frequency): New function.
6627 * profile-count.h (struct function): Declare.
6628 (enum profile_quality): Add profile_guessed_local and
6629 profile_guessed_global0.
6630 (class profile_proability): Decrease number of bits to 29;
6631 update from_reg_br_prob_note and to_reg_br_prob_note.
6632 (class profile_count: Update comment; decrease number of bits
6633 to 61. Check compatibility.
6634 (profile_count::compatible_p): New private member function.
6635 (profile_count::ipa_p): New member function.
6636 (profile_count::operator<): Handle global zero correctly.
6637 (profile_count::operator>): Handle global zero correctly.
6638 (profile_count::operator<=): Handle global zero correctly.
6639 (profile_count::operator>=): Handle global zero correctly.
6640 (profile_count::nonzero_p): New member function.
6641 (profile_count::force_nonzero): New member function.
6642 (profile_count::max): New member function.
6643 (profile_count::apply_scale): Handle IPA scalling.
6644 (profile_count::guessed_local): New member function.
6645 (profile_count::global0): New member function.
6646 (profile_count::ipa): New member function.
6647 (profile_count::to_frequency): Declare.
6648 (profile_count::to_cgraph_frequency): Declare.
6649 * profile.c (OVERLAP_BASE): Delete.
6650 (compute_frequency_overlap): Delete.
6651 (compute_branch_probabilities): Do not use compute_frequency_overlap.
6652 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
6653 * sched-ebb.c (rank): Use counts only.
6654 * shrink-wrap.c (handle_simple_exit): Use counts only.
6655 (try_shrink_wrapping): Use counts only.
6656 (place_prologue_for_one_component): Use counts only.
6657 * tracer.c (find_best_predecessor): Use to_frequency.
6658 (find_trace): Use to_frequency.
6659 (tail_duplicate): Use to_frequency.
6660 * trans-mem.c (expand_transaction): Do not update frequency.
6661 * tree-call-cdce.c: Do not update frequency.
6662 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
6663 (gimple_merge_blocks): Likewise.
6664 (gimple_split_edge): Likewise.
6665 (gimple_duplicate_sese_region): Likewise.
6666 (gimple_duplicate_sese_tail): Likewise.
6667 (move_sese_region_to_fn): Likewise.
6668 (gimple_account_profile_record): Likewise.
6669 (insert_cond_bb): Likewise.
6670 * tree-complex.c (expand_complex_div_wide): Likewise.
6671 * tree-eh.c (lower_resx): Update profile.
6672 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
6674 (initialize_cfun): Do not initialize frequencies
6675 (freqs_to_counts): Delete.
6676 (copy_cfg_body): Ignore count parameter.
6677 (copy_body): Update.
6678 (expand_call_inline): Update count_max.
6679 (optimize_inline_calls): Update count_max.
6680 (tree_function_versioning): Update count_max.
6681 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
6682 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
6684 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
6685 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
6686 (try_peel_loop): Likewise.
6687 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
6689 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
6690 (tree_transform_and_unroll_loop): Do not use frequencies
6691 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
6692 Use reliable prediction only.
6693 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
6694 * tree-ssa-sink.c (select_best_block): Use to_frequency.
6695 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
6696 probability scaling.
6697 * tree-ssa-threadupdate.c (create_block_for_threading): Do
6698 not update frequency
6699 (any_remaining_duplicated_blocks): Likewise.
6700 (update_profile): Likewise.
6701 (estimated_freqs_path): Delete.
6702 (freqs_to_counts_path): Delete.
6703 (clear_counts_path): Delete.
6704 (ssa_fix_duplicate_block_edges): Likewise.
6705 (duplicate_thread_path): Likewise.
6706 * tree-switch-conversion.c (gen_inbound_check): Use counts.
6707 * tree-tailcall.c (decrease_profile): Do not update frequency.
6708 (eliminate_tail_call): Likewise.
6709 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
6710 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
6711 (optimize_mask_stores): Likewise.
6712 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6713 * ubsan.c (ubsan_expand_null_ifn): Update profile.
6714 (ubsan_expand_ptr_ifn): Update profile.
6715 * value-prof.c (gimple_ic): Simplify.
6716 * value-prof.h (gimple_ic): Update prototype.
6717 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
6718 * ipa-inline.c (compute_uninlined_call_time): Be sure that
6720 (want_inline_self_recursive_call_p): Likewise.
6721 (resolve_noninline_speculation): Only cummulate defined counts.
6722 (inline_small_functions): Use nonzero_p.
6723 (ipa_inline): Do not access freed node.
6725 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6727 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6728 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
6730 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
6732 * config/riscv/riscv.c (riscv_legitimize_move): Handle
6733 non-legitimate address.
6735 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6737 * config/rs6000/rs6000.md (*lt0_disi): Delete.
6738 (*lt0_<mode>di, *lt0_<mode>si): New.
6740 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6742 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
6743 TARGET_PAIRED_FLOAT.
6745 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
6746 Jim Wilson <jim.wilson@linaro.org>
6748 * config/aarch64/aarch64-cores.def (saphira): New CPU.
6749 * config/aarch64/aarch64-tune.md: Regenerated.
6750 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
6751 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
6753 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
6755 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
6756 (arc_expand_prologue): Restore blink for millicode.
6757 * config/arc/linux.h (LINK_EH_SPEC): Defined.
6759 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
6762 * config/i386/i386.c (ix86_vector_duplicate_value): Use
6763 gen_vec_duplicate after forcing the scalar into a register.
6765 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
6767 * combine (try_combine): Print the insns input to try_combine to the
6770 2017-11-02 Steve Ellcey <sellcey@cavium.com>
6773 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
6774 Remove second argument from aarch64_process_target_attr call.
6775 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
6777 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
6779 (aarch64_handle_attr_arch): Remove second argument.
6780 (aarch64_handle_attr_cpu): Ditto.
6781 (aarch64_handle_attr_tune): Ditto.
6782 (aarch64_handle_attr_isa_flags): Ditto.
6783 (aarch64_process_one_target_attr): Ditto.
6784 (aarch64_process_target_attr): Ditto.
6785 (aarch64_option_valid_attribute_p): Remove second argument.
6786 on aarch64_process_target_attr call.
6788 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6790 * diagnostic.c: Include "selftest-diagnostic.h".
6791 (selftest::assert_location_text): New function.
6792 (selftest::test_diagnostic_get_location_text): New function.
6793 (selftest::diagnostic_c_tests): Call it.
6795 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6797 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
6798 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
6799 (class selftest::test_diagnostic_context): Move to...
6800 * selftest-diagnostic.c: New file.
6801 * selftest-diagnostic.h: New file.
6803 2017-11-02 James Bowman <james.bowman@ftdichip.com>
6805 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
6806 offset range for FT32B.
6807 * config/ft32/ft32.h: option "mcompress" enables relaxation.
6808 * config/ft32/ft32.md: Add TARGET_NOPM.
6809 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
6810 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
6812 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
6814 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
6816 2017-11-02 Jeff Law <law@redhat.com>
6818 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
6819 virtual keyword on FINAL OVERRIDE members.
6821 * tree-ssa-propagate.h (ssa_propagation_engine): Group
6822 virtuals together. Add virtual destructor.
6823 (substitute_and_fold_engine): Similarly.
6825 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
6827 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
6829 2017-11-02 Richard Biener <rguenther@suse.de>
6831 PR tree-optimization/82795
6832 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
6834 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6836 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
6838 (gcc_GAS_CHECK_FEATURE): Remove.
6839 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
6840 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
6841 * configure: Regenerate.
6843 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
6845 * config/arc/arc.c (hwloop_optimize): Account for empty
6848 2017-11-02 Richard Biener <rguenther@suse.de>
6851 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
6852 Truncate ARRAY_REF index and element size.
6854 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6856 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
6858 2017-11-01 Jeff Law <law@redhat.com>
6860 * tree-ssa-ccp.c (ccp_folder): New class derived from
6861 substitute_and_fold_engine.
6862 (ccp_folder::get_value): New member function.
6863 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
6864 (ccp_fold_stmt): Remove prototype.
6865 (ccp_finalize): Call substitute_and_fold from the ccp_class.
6866 * tree-ssa-copy.c (copy_folder): New class derived from
6867 substitute_and_fold_engine.
6868 (copy_folder::get_value): Renamed from get_value.
6869 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
6870 * tree-vrp.c (vrp_folder): New class derived from
6871 substitute_and_fold_engine.
6872 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
6873 (vrp_folder::get_value): New member function.
6874 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
6875 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
6876 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
6877 provide a class interface to folder/substitute routines.
6878 (ssa_prop_fold_stmt_fn): Remove typedef.
6879 (ssa_prop_get_value_fn): Likewise.
6880 (subsitute_and_fold): Remove prototype.
6881 (replace_uses_in): Likewise.
6882 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
6883 Renamed from replace_uses_in. Call the virtual member function
6884 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
6885 (substitute_and_fold_dom_walker): Remove initialization of
6886 data member entries for calbacks. Add substitute_and_fold_engine
6887 member and initialize it.
6888 (substitute_and_fold_dom_walker::before_dom_children0: Use the
6889 member functions for get_value, replace_phi_args_in c
6890 replace_uses_in, and fold_stmt calls.
6891 (substitute_and_fold_engine::substitute_and_fold): Renamed from
6892 substitute_and_fold. Remove assert. Update ctor call.
6894 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
6895 (ssa_prop_visit_phi_fn): Likewise.
6896 (class ssa_propagation_engine): New class to provide an interface
6898 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
6900 (ssa_prop_visit_phi): Likewise.
6901 (ssa_propagation_engine::simulate_stmt): Moved into class.
6902 Call visit_phi/visit_stmt from the class rather than via
6903 file scoped static variables.
6904 (ssa_propagation_engine::simulate_block): Moved into class.
6905 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
6906 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
6907 set file scoped statics for the visit_stmt/visit_phi callbacks.
6908 * tree-complex.c (complex_propagate): New class derived from
6909 ssa_propagation_engine.
6910 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
6911 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
6912 (tree_lower_complex): Call ssa_propagate via the complex_propagate
6914 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
6915 ssa_propagation_engine.
6916 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
6917 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
6918 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
6919 * tree-ssa-copy.c (copy_prop): New class derived from
6920 ssa_propagation_engine.
6921 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
6922 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
6923 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
6924 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
6925 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
6926 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
6927 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6929 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6931 PR rtl-optimization/82778
6932 PR rtl-optimization/82597
6933 * compare-elim.c (struct comparison): Add in_a_setter field.
6934 (find_comparison_dom_walker::before_dom_children): Remove killed
6935 bitmap and df_simulate_find_defs call, instead walk the defs.
6936 Compute last_setter and initialize in_a_setter. Merge definitions
6937 with first initialization for a few variables.
6938 (try_validate_parallel): Use insn_invalid_p instead of
6939 recog_memoized. Return insn rather than just the pattern.
6940 (try_merge_compare): Fix up comment. Don't uselessly test if
6941 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6943 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6946 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6947 Alan Hayward <alan.hayward@arm.com>
6948 David Sherwood <david.sherwood@arm.com>
6950 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6951 aarch64_hard_regno_nregs to get the number of registers
6954 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6955 Alan Hayward <alan.hayward@arm.com>
6956 David Sherwood <david.sherwood@arm.com>
6958 * config/aarch64/constraints.md (Upl): Rename to...
6960 * config/aarch64/aarch64.md
6961 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6964 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6965 Alan Hayward <alan.hayward@arm.com>
6966 David Sherwood <david.sherwood@arm.com>
6968 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6969 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6972 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6973 Alan Hayward <alan.hayward@arm.com>
6974 David Sherwood <david.sherwood@arm.com>
6976 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6977 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6978 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6980 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6981 of a single element.
6982 * config/aarch64/iterators.md: Add a comment above the permute
6983 unspecs to say that they are generated directly by
6984 aarch64_expand_vec_perm_const.
6985 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6987 2017-11-01 Nathan Sidwell <nathan@acm.org>
6989 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6991 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6993 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6994 and medany code models, and describe what they do.
6996 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6998 Revert accidental duplicate:
7000 * combine.c (can_change_dest_mode): Reject changes in
7001 REGMODE_NATURAL_SIZE.
7003 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
7005 PR rtl-optimization/64682
7006 PR rtl-optimization/69567
7007 PR rtl-optimization/69737
7008 PR rtl-optimization/82683
7009 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
7010 register mentioned in the note, drop the note, unless it came from I3,
7011 in which case it should go to I3 again.
7013 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7015 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
7016 and return false if not.
7017 (clear_bytes_written_by, live_bytes_read): Update accordingly.
7019 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7021 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
7022 range is known to be empty.
7024 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7025 Alan Hayward <alan.hayward@arm.com>
7026 David Sherwood <david.sherwood@arm.com>
7028 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
7029 and CONST_VECTOR_NUNITS instead of computing the number of units from
7030 the byte sizes of the vector and element.
7031 (simplify_binary_operation_1): Likewise.
7032 (simplify_const_binary_operation): Likewise.
7033 (simplify_ternary_operation): Likewise.
7035 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7036 Alan Hayward <alan.hayward@arm.com>
7037 David Sherwood <david.sherwood@arm.com>
7039 * var-tracking.c (INT_MEM_OFFSET): Replace with...
7040 (int_mem_offset): ...this new function.
7041 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
7042 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
7043 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
7046 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7047 Alan Hayward <alan.hayward@arm.com>
7048 David Sherwood <david.sherwood@arm.com>
7050 * lower-subreg.c (interesting_mode_p): New function.
7051 (compute_costs, find_decomposable_subregs, decompose_register)
7052 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
7053 (resolve_clobber, dump_choices): Use it.
7055 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7056 Alan Hayward <alan.hayward@arm.com>
7057 David Sherwood <david.sherwood@arm.com>
7059 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
7061 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7062 Alan Hayward <alan.hayward@arm.com>
7063 David Sherwood <david.sherwood@arm.com>
7065 * alias.c (find_base_value, find_base_term): Only process integer
7066 truncations. Check the precision rather than the size.
7068 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7069 Alan Hayward <alan.hayward@arm.com>
7070 David Sherwood <david.sherwood@arm.com>
7072 * machmode.h (is_narrower_int_mode): New function
7073 * optabs.c (expand_float, expand_fix): Use it.
7074 * dwarf2out.c (rotate_loc_descriptor): Likewise.
7076 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7077 Alan Hayward <alan.hayward@arm.com>
7078 David Sherwood <david.sherwood@arm.com>
7080 * rtl.h (narrower_subreg_mode): New function.
7081 * ira-color.c (update_costs_from_allocno): Use it.
7083 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7084 Alan Hayward <alan.hayward@arm.com>
7085 David Sherwood <david.sherwood@arm.com>
7087 * optabs-query.h (convert_optab_p): New function, split out from...
7088 (convert_optab_handler): ...here.
7089 (widening_optab_handler): Delete.
7090 (find_widening_optab_handler): Remove permit_non_widening parameter.
7091 (find_widening_optab_handler_and_mode): Likewise. Provide an
7092 override that operates on mode class wrappers.
7093 * optabs-query.c (widening_optab_handler): Delete.
7094 (find_widening_optab_handler_and_mode): Remove permit_non_widening
7095 parameter. Assert that the two modes are the same class and that
7096 the "from" mode is narrower than the "to" mode. Use
7097 convert_optab_handler instead of widening_optab_handler.
7098 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
7099 instead of widening_optab_handler.
7100 * expr.c (expand_expr_real_2): Update calls to
7101 find_widening_optab_handler.
7102 * optabs.c (expand_widen_pattern_expr): Likewise.
7103 (expand_binop_directly): Take the insn_code as a parameter.
7104 (expand_binop): Only call find_widening_optab_handler for
7105 conversion optabs; use optab_handler otherwise. Update calls
7106 to find_widening_optab_handler and expand_binop_directly.
7107 Use convert_optab_handler instead of widening_optab_handler.
7108 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
7109 find_widening_optab_handler and use scalar_mode rather than
7111 (convert_plusminus_to_widen): Likewise.
7113 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7114 Alan Hayward <alan.hayward@arm.com>
7115 David Sherwood <david.sherwood@arm.com>
7117 * machmode.h (fixed_size_mode): New class.
7118 * rtl.h (get_pool_mode): Return fixed_size_mode.
7119 * gengtype.c (main): Add fixed_size_mode.
7120 * target.def (get_raw_result_mode): Return a fixed_size_mode.
7121 (get_raw_arg_mode): Likewise.
7122 * doc/tm.texi: Regenerate.
7123 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
7124 * targhooks.c (default_get_reg_raw_mode): Likewise.
7125 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
7126 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
7127 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
7128 (msp430_get_raw_result_mode): Likewise.
7129 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
7130 * dbxout.c (dbxout_parms): Require fixed-size modes.
7131 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
7132 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7133 * omp-low.c (lower_oacc_reductions): Likewise.
7134 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
7135 (simplify_subreg): Update accordingly.
7136 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
7137 (force_const_mem): Update accordingly. Return NULL_RTX for modes
7138 that aren't fixed-size.
7139 (get_pool_mode): Return a fixed_size_mode.
7140 (output_constant_pool_2): Take a fixed_size_mode.
7142 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7143 Alan Hayward <alan.hayward@arm.com>
7144 David Sherwood <david.sherwood@arm.com>
7146 * doc/rtl.texi (vec_series): Document.
7147 (const): Say that the operand can be a vec_series.
7148 * rtl.def (VEC_SERIES): New rtx code.
7149 * rtl.h (const_vec_series_p_1): Declare.
7150 (const_vec_series_p): New function.
7151 * emit-rtl.h (gen_const_vec_series): Declare.
7152 (gen_vec_series): Likewise.
7153 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
7154 (gen_vec_series): Likewise.
7155 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
7156 * simplify-rtx.c (simplify_unary_operation): Handle negations
7158 (simplify_binary_operation_series): New function.
7159 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
7160 (test_vector_ops_series): New function.
7161 (test_vector_ops): Call it.
7162 * config/powerpcspe/altivec.md (altivec_lvsl): Use
7163 gen_const_vec_series.
7164 (altivec_lvsr): Likewise.
7165 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
7167 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7168 Alan Hayward <alan.hayward@arm.com>
7169 David Sherwood <david.sherwood@arm.com>
7171 * doc/rtl.texi (const): Update description of address constants.
7172 Say that vector constants are allowed too.
7173 * common.md (E, F): Use CONSTANT_P instead of checking for
7175 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
7176 checking for CONST_VECTOR.
7177 * expmed.c (make_tree): Use build_vector_from_val for a CONST
7179 * expr.c (expand_expr_real_2): Check for vector modes instead
7180 of checking for CONST_VECTOR.
7181 * rtl.h (const_vec_p): New function.
7182 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
7183 (unwrap_const_vec_duplicate): Handle them here too.
7185 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7186 David Malcolm <dmalcolm@redhat.com>
7187 Alan Hayward <alan.hayward@arm.com>
7188 David Sherwood <david.sherwood@arm.com>
7190 * rtl.h (vec_duplicate_p): New function.
7191 * selftest-rtl.c (assert_rtx_eq_at): New function.
7192 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
7193 (assert_rtx_eq_at): Declare.
7194 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
7195 * selftest-run-tests.c (selftest::run_tests): Call it.
7196 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
7197 (simplify_unary_operation_1): Recursively handle vector duplicates.
7198 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
7200 (simplify_subreg): Handle subregs of vector duplicates.
7201 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
7202 (selftest::simplify_rtx_c_tests): New functions.
7204 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7205 Alan Hayward <alan.hayward@arm.com>
7206 David Sherwood <david.sherwood@arm.com>
7208 * emit-rtl.h (gen_const_vec_duplicate): Declare.
7209 (gen_vec_duplicate): Likewise.
7210 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
7212 (gen_const_vector): ...here.
7213 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
7214 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
7215 whose elements are all equal.
7216 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
7217 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
7218 (simplify_relational_operation): Likewise.
7219 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
7221 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
7222 (aarch64_expand_vector_init): Likewise.
7223 * config/arm/arm.c (neon_vdup_constant): Likewise.
7224 (neon_expand_vector_init): Likewise.
7225 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
7226 (arm_block_set_unaligned_vect): Likewise.
7227 (arm_block_set_aligned_vect): Likewise.
7228 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
7229 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
7230 (expand_vec_perm_even_odd_pack): Likewise.
7231 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
7232 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
7233 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
7234 gen_const_vec_duplicate.
7235 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
7236 * config/mips/mips.c (mips_gen_const_int_vector): Use
7237 gen_const_vec_duplicate.
7238 (mips_expand_vector_init): Use CONST0_RTX.
7239 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
7240 (define_split): Use gen_const_vec_duplicate.
7241 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
7242 (define_split): Use gen_const_vec_duplicate.
7243 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
7244 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
7245 * config/spu/spu.c (spu_const): Likewise.
7247 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7248 Alan Hayward <alan.hayward@arm.com>
7249 David Sherwood <david.sherwood@arm.com>
7251 * combine.c (can_change_dest_mode): Reject changes in
7252 REGMODE_NATURAL_SIZE.
7254 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
7256 * configure.ac (--enable-libssp): New.
7257 (gcc_cv_libc_provides_ssp): Check for explicit setting before
7258 trying to determine target-specific default. Adjust indentation.
7259 * configure: Regenerated.
7260 * doc/install.texi (Configuration): Expand --disable-libssp
7263 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
7265 config/i386/i386.c (ix86_expand_epilogue): Correct stack
7268 2017-10-31 Martin Jambor <mjambor@suse.cz>
7271 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
7273 2017-10-31 David Malcolm <dmalcolm@redhat.com>
7275 * auto-profile.c (autofdo_source_profile::read): Use
7276 UNKNOWN_LOCATION rather than 0.
7277 * diagnostic-core.h (warning_at_rich_loc): Rename to...
7278 (warning_at): ...this overload.
7279 (warning_at_rich_loc_n): Rename to...
7280 (warning_n): ...this overload.
7281 (error_at_rich_loc): Rename to...
7282 (error_at): ...this overload.
7283 (pedwarn_at_rich_loc): Rename to...
7284 (pedwarn): ...this overload.
7285 (permerror_at_rich_loc): Rename to...
7286 (permerror): ...this overload.
7287 (inform_at_rich_loc): Rename to...
7288 (inform): ...this overload.
7289 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
7290 (diagnostic_n_impl_richloc): Rename to...
7291 (diagnostic_n_impl): ...this rich_location *-based decl.
7292 (inform_at_rich_loc): Rename to...
7293 (inform): ...this, and add an assertion.
7294 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
7295 (warning_at_rich_loc): Rename to...
7296 (warning_at): ...this, and add an assertion.
7297 (warning_at_rich_loc_n): Rename to...
7298 (warning_n): ...this, and add an assertion.
7299 (warning_n): Update location_t-based implementation for removal of
7300 location_t-based diagnostic_n_impl.
7301 (pedwarn_at_rich_loc): Rename to...
7302 (pedwarn): ...this, and add an assertion.
7303 (permerror_at_rich_loc): Rename to...
7304 (permerror): ...this, and add an assertion.
7305 (error_n): Update for removal of location_t-based diagnostic_n_impl.
7306 (error_at_rich_loc): Rename to...
7307 (error_at): ...this, and add an assertion.
7308 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
7309 (driver::do_spec_on_infiles): Likewise.
7310 * substring-locations.c (format_warning_va): Update for renaming
7311 of inform_at_rich_loc.
7313 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
7315 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
7316 _Float<N>X built-in functions so that the variant without the
7317 "__builtin_" prefix is only enabled for the GNU C and Objective C
7318 languages when they are in non-strict ANSI/ISO mode.
7319 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
7320 * target.def (floatn_builtin_p): Add a target hook to control
7321 whether _Float<N> and _Float<N>X built-in functions without the
7322 "__builtin_" prefix are enabled, and return true for C and
7323 Objective C in the default hook. Include langhooks.h in
7325 * targhooks.h (default_floatn_builtin_p): Likewise.
7326 * targhooks.c (default_floatn_builtin_p): Likewise.
7327 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
7328 floatn_builtin_p target hook.
7329 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
7331 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
7332 Eric Botcazou <ebotcazou@adacore.com>
7334 PR rtl-optimization/81803
7335 * lra-constraints.c (curr_insn_transform): Also reload the whole
7336 register for a strict subreg no wider than a word if this is for
7337 a WORD_REGISTER_OPERATIONS target.
7339 2017-10-31 Jason Merrill <jason@redhat.com>
7341 * gdbinit.in: Skip over inlines from timevar.h.
7343 2017-10-31 Martin Liska <mliska@suse.cz>
7345 * doc/gcov.texi: Document new option.
7346 * gcov.c (print_usage): Likewise print it.
7347 (process_args): Support the argument.
7348 (format_count): New function.
7349 (format_gcov): Use the function.
7351 2017-10-31 Martin Liska <mliska@suse.cz>
7353 * gcov.c (struct name_map): do not use typedef.
7354 Define operator== and operator<.
7355 (name_search): Remove.
7356 (name_sort): Remove.
7357 (main): Do not allocate names.
7358 (process_file): Add vertical space.
7359 (generate_results): Use std::find.
7360 (release_structures): Do not release memory.
7361 (find_source): Use std::find.
7363 2017-10-31 Martin Liska <mliska@suse.cz>
7365 * gcov.c (struct line_info): Remove it's typedef.
7366 (line_info::line_info): Add proper ctor.
7367 (line_info::has_block): Do not use a typedef.
7368 (struct source_info): Do not use typedef.
7369 (circuit): Likewise.
7370 (get_cycles_count): Likewise.
7371 (output_intermediate_file): Iterate via vector iterator.
7372 (add_line_counts): Use std::vector methods.
7373 (accumulate_line_counts): Likewise.
7374 (output_lines): Likewise.
7376 2017-10-31 Martin Liska <mliska@suse.cz>
7378 * gcov.c (struct source_info): Remove typedef.
7379 (source_info::source_info): Add proper ctor.
7380 (accumulate_line_counts): Use struct, not it's typedef.
7381 (output_gcov_file): Likewise.
7382 (output_lines): Likewise.
7383 (main): Do not allocate an array.
7384 (output_intermediate_file): Use size of vector container.
7385 (process_file): Resize the vector.
7386 (generate_results): Do not preallocate, use newly added vector
7388 (release_structures): Do not release sources.
7389 (find_source): Use vector methods.
7390 (add_line_counts): Do not use typedef.
7392 2017-10-31 Martin Liska <mliska@suse.cz>
7394 * doc/gcov.texi: Document that.
7395 * gcov.c (add_line_counts): Mark lines with a non-executed
7397 (output_line_beginning): Handle such lines.
7398 (output_lines): Pass new argument.
7399 (output_intermediate_file): Print it in intermediate format.
7401 2017-10-31 Martin Liska <mliska@suse.cz>
7403 * color-macros.h: New file.
7404 * diagnostic-color.c: Factor out color related to macros to
7406 * doc/gcov.texi: Document -k option.
7407 * gcov.c (INCLUDE_STRING): Include string.h.
7408 (print_usage): Add -k option.
7409 (process_args): Parse it.
7410 (pad_count_string): New function.
7411 (output_line_beginning): Likewise.
7412 (DEFAULT_LINE_START): New macro.
7413 (output_lines): Support color output.
7415 2017-10-31 Martin Liska <mliska@suse.cz>
7417 PR gcov-profile/82633
7418 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
7419 their interaction with GCOV infrastructure.
7420 * configure.ac: Add -fkeep-{inline,static}-functions to
7422 * configure: Regenerate.
7424 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
7427 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
7429 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
7432 * config/rs6000/rs6000.md (allocate_stack): Force update interval
7433 into a register if it does not fit into an immediate offset field.
7435 2017-10-31 Olivier Hainque <hainque@adacore.com>
7437 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
7439 2017-10-31 Julia Koval <julia.koval@intel.com>
7441 * config.gcc: Add gfniintrin.h.
7442 * config/i386/gfniintrin.h: New.
7443 * config/i386/i386-builtin-types.def
7444 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7445 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7446 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7447 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7448 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7449 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
7450 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
7451 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
7452 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
7453 V64QI_FTYPE_V64QI_V64QI_INT): New types.
7454 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
7455 * config/i386/immintrin.h: Include gfniintrin.h.
7456 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
7458 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
7460 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
7462 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7464 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
7465 (arm_ashldi3_1bit): Remove pattern.
7466 (ashrdi3): Remove shift by 1 expansion.
7467 (arm_ashrdi3_1bit): Remove pattern.
7468 (lshrdi3): Remove shift by 1 expansion.
7469 (arm_lshrdi3_1bit): Remove pattern.
7470 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
7471 cost of ashldi3 by 1.
7472 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
7473 (<shift>di3_neon): Likewise.
7475 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7477 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
7478 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
7479 and (*aarch64_simd_mov<VQ:mode>).
7480 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
7481 pattern alternative.
7482 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
7483 attributes to match pattern alternative.
7485 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
7487 * config.gcc (powerpc*-*-*): Add emmintrin.h.
7488 * config/rs6000/emmintrin.h: New file.
7489 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
7491 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7493 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
7494 * (movdi_vfp_cortexa8): Remove pattern.
7496 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7498 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
7500 (Specific, alpha*-dec-osf5.1): Remove.
7501 (Specific, mips-sgi-irix5): Remove.
7502 (Specific, mips-sgi-irix6): Remove.
7504 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7507 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
7508 arguments to clear_bit_region_be.
7510 2017-10-30 Jim Wilson <wilson@tuliptree.org>
7512 * gimplify.c: Include memmodel.h.
7514 2017-10-30 Martin Jambor <mjambor@suse.cz>
7516 * omp-grid.c (grid_attempt_target_gridification): Also insert a
7517 condition whether loop should be executed at all.
7519 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
7521 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7522 gimple folding of vec_madd() intrinsics.
7523 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
7524 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
7525 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
7527 2017-10-30 Richard Biener <rguenther@suse.de>
7529 PR tree-optimization/82762
7531 2017-10-23 Richard Biener <rguenther@suse.de>
7533 PR tree-optimization/82129
7535 2017-08-01 Richard Biener <rguenther@suse.de>
7537 PR tree-optimization/81181
7538 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7539 (compute_antic): ... end of iteration here.
7541 2017-10-30 Joseph Myers <joseph@codesourcery.com>
7543 * doc/invoke.texi (C Dialect Options): Document -std=c17,
7544 -std=iso9899:2017 and -std=gnu17.
7545 * doc/standards.texi (C Language): Document C17 support.
7546 * doc/cpp.texi (Overview): Mention -std=c17.
7547 (Standard Predefined Macros): Document C11 and C17 values of
7548 __STDC_VERSION__. Do not refer to C99 support as incomplete.
7549 * doc/extend.texi (Inline): Do not list individual options for
7550 standards newer than C99.
7551 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
7553 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
7556 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
7558 * asan.c (asan_finish_file): Align asan globals array by shadow
7561 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7564 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
7565 (struct store_immediate_info): Add bitregion_start and bitregion_end
7567 (store_immediate_info::store_immediate_info): Add brs and bre
7568 arguments and initialize bitregion_{start,end} from those.
7569 (struct merged_store_group): Add bitregion_start, bitregion_end,
7570 align_base and mask fields. Drop unnecessary struct keyword from
7571 struct store_immediate_info. Add do_merge method.
7572 (clear_bit_region_be): Use memset instead of loop storing zeros.
7573 (merged_store_group::do_merge): New method.
7574 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
7575 stores as long as the surrounding bitregions have no gaps.
7576 (merged_store_group::merge_overlapping): Use do_merge.
7577 (merged_store_group::apply_stores): Test that bitregion_{start,end}
7578 is byte aligned, rather than requiring that start and width are
7579 byte aligned. Drop unnecessary struct keyword from
7580 struct store_immediate_info. Allocate and populate also mask array.
7581 Make start of the arrays relative to bitregion_start rather than
7582 start and size them according to bitregion_{end,start} difference.
7583 (struct imm_store_chain_info): Drop unnecessary struct keyword from
7584 struct store_immediate_info.
7585 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
7586 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
7587 struct keyword from struct store_immediate_info.
7588 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
7589 between stores as long as the surrounding bitregions have no gaps.
7591 (struct split_store): Add orig non-static data member.
7592 (split_store::split_store): Initialize orig to false.
7593 (find_constituent_stmts): Return store_immediate_info *, non-NULL
7594 if there is exactly a single original stmt. Change stmts argument
7595 to pointer from reference, if NULL, don't push anything to it. Add
7596 first argument, use it to optimize skipping over orig stmts that
7597 are known to be before bitpos already. Simplify.
7598 (split_group): Return unsigned int count how many stores are or
7599 would be needed rather than a bool. Add allow_unaligned argument.
7600 Change split_stores argument from reference to pointer, if NULL,
7601 only do a dry run computing how many stores would be produced.
7602 Rewritten algorithm to use both alignment and misalign if
7603 !allow_unaligned and handle bitfield stores with gaps.
7604 (imm_store_chain_info::output_merged_store): Set start_byte_pos
7605 from bitregion_start instead of start. Compute allow_unaligned
7606 here, if true, do 2 split_group dry runs to compute which one
7607 produces fewer stores and prefer aligned if equal. Punt if
7608 new count is bigger or equal than original before emitting any
7609 statements, rather than during that. Remove no longer needed
7610 new_ssa_names tracking. Replace num_stmts with
7611 split_stores.length (). Use 32-bit stack allocated entries
7612 in split_stores auto_vec. Try to reuse original store lhs/rhs1
7613 if possible. Handle bitfields with gaps.
7614 (pass_store_merging::execute): Ignore bitsize == 0 stores.
7615 Compute bitregion_{start,end} for the stores and construct
7616 store_immediate_info with that. Formatting fixes.
7618 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
7621 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
7622 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
7624 2017-10-29 Jim Wilson <wilson@tuliptree.org>
7626 * gimplify.c: Include tm_p.h.
7628 * common.opt (gcoff): Re-add as ignored option.
7629 (gcoff1, gcoff2, gcoff3): Likewise.
7631 * Makefile.in (OBJS): Delete sdbout.o.
7632 (GTFILES): Delete $(srcdir)/sdbout.c.
7633 * debug.h: Delete sdb_debug_hooks.
7634 * final.c: Delete sdbout.h include.
7635 (final_scan_insn): Delete SDB_DEBUG check.
7636 (rest_of_clean_state): Likewise.
7637 * output.h: Delete sdb_begin_function_line.
7640 * toplev.c: Delete sdbout.h include.
7641 (process_options): Delete SDB_DEBUG check.
7642 * tree-core.h (tree_type_common): Delete pointer field of
7644 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
7645 TYPE_SYMTAB_POINTER.
7646 * tree.h (TYPE_SYMTAB_POINTER): Delete.
7647 (TYPE_SYMTAB_IS_POINTER): Delete.
7648 (TYPE_SYMTAB_IS_DIE): Renumber.
7649 * xcoffout.c: Refer to former sdbout.c file.
7650 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
7652 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
7653 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
7654 (Debugging Options): Delete -gcoff.
7655 (-gstabs): Delete SDB reference.
7657 (-gcoff@var{level}): Delete.
7658 * doc/passes.texi (Debugging information output): Delete SDB and
7659 sdbout.c references.
7660 * doc/tm.texi: Regenerate.
7661 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
7662 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
7664 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
7665 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
7666 (SDB_DEBUGGING_INFO): Delete.
7667 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
7668 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
7669 * target.def (output_source_filename): Delete COFF reference.
7671 * common.opt (gcoff): Delete.
7672 (gxcoff+): Update Negative chain.
7673 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
7675 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
7676 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
7677 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
7679 (expand_function_start): Change sdb reference to past tense.
7680 (expand_function_end): Change sdb reference to past tense.
7681 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
7682 * opts.c (debug_type_names): Delete coff entry.
7683 (common_handle_option): Delete OPT_gcoff case.
7684 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
7686 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
7687 * config/cris/cris.h: Delete SDB reference in comment.
7688 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
7689 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
7690 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
7691 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
7693 (ix86_expand_prologue): Likewise.
7694 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
7695 * config/ia64/ia64.h: Likewise.
7696 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
7697 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
7699 * config/mmix/mmix.h: Likewise.
7700 * config/nds32/nds32.c: Likewise.
7701 * config/stormy/storym16.h: Likewise.
7702 * config/visium/visium.h: Likewise.
7703 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
7705 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
7707 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
7708 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
7709 FRAME_POINTER_REGNUM point at high end of local var area.
7711 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
7713 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
7714 Move comment around. Do not reset best_edge for a copiable
7715 destination if the copy would cause a partition change.
7716 (better_edge_p): Remove redundant check.
7718 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7720 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
7722 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
7724 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
7725 for math functions that have _Float<N> and _Float<N>X variants.
7726 (mathfn_built_in_2): Add support for math functions that have
7727 _Float<N> and _Float<N>X variants.
7728 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
7729 (expand_builtin_mathfn_ternary): Add support for fma with
7730 _Float<N> and _Float<N>X variants.
7731 (expand_builtin): Likewise.
7732 (fold_builtin_3): Likewise.
7733 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
7734 create math function _Float<N> and _Float<N>X variants as external
7736 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
7737 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
7738 the __builtin_ prefix and if not strict ansi, without the prefix.
7739 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
7740 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
7741 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
7742 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
7743 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
7744 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
7745 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
7746 function signatures for fma _Float<N> and _Float<N>X variants.
7747 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
7748 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
7749 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
7750 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
7751 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
7752 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
7753 * gencfn-macros.c (print_case_cfn): Add support for math functions
7754 that have _Float<N> and _Float<N>X variants.
7755 (print_define_operator_list): Likewise.
7756 (fltfn_suffixes): Likewise.
7758 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
7759 for math functions that have _Float<N> and _Float<N>X variants.
7760 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
7761 and _Float<N>X variants.
7762 (COPYSIGN): Likewise.
7765 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
7766 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
7768 (integer_valued_read_call_p): Likewise.
7769 * fold-const-call.c (fold_const_call_ss): Likewise.
7770 (fold_const_call_sss): Add support for copysign, fmin, and fmax
7771 _Float<N> and _Float<N>X variants.
7772 (fold_const_call_ssss): Add support for fma _Float<N> and
7773 _Float<N>X variants.
7774 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
7775 support for copysign and fma _Float<N> and _Float<N>X variants.
7776 (backprop::process_builtin_call_use): Likewise.
7777 * tree-call-cdce.c (can_test_argument_range); Add support for
7778 sqrt _Float<N> and _Float<N>X variants.
7779 (edom_only_function): Likewise.
7780 (get_no_error_domain): Likewise.
7781 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
7782 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
7783 copysign _Float<N> and _Float<N>X variants.
7784 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
7785 handled by machine independent code.
7786 (FMAF128): Likewise.
7787 * doc/cpp.texi (Common Predefined Macros): Document defining
7788 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
7789 fma _Float<N> and _Float<N>X variants.
7791 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7794 * config/i386/i386-modes.def (CCFPU): Remove definition.
7795 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
7796 (ix86_cc_modes_compatible): Ditto.
7797 (ix86_expand_carry_flag_compare): Ditto.
7798 (ix86_expand_int_movcc): Ditto.
7799 (ix86_expand_int_addcc): Ditto.
7800 (ix86_reverse_condition): Ditto.
7801 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
7802 Return true/false for unordered/ordered fp comparisons.
7803 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
7804 (ix86_prepare_fp_compare_args): Update for rename.
7805 (ix86_expand_fp_compare): Update for rename. Generate unordered
7806 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
7807 (ix86_expand_sse_compare_and_jump): Ditto.
7808 * config/i386/predicates.md (fcmov_comparison_operator):
7809 Remove CCFPU mode handling.
7810 (ix86_comparison_operator): Ditto.
7811 (ix86_carry_flag_operator): Ditto.
7812 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
7813 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
7814 (*cmpu<mode>_cc_i387): Ditto.
7815 (FPCMP): Remove mode iterator.
7816 (unord): Remove mode attribute.
7817 (unord_subst): New define_subst transformation
7818 (unord): New define_subst attribute.
7820 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
7821 (*cmpi<unord>xf_i387): Ditto.
7822 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
7823 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
7824 using unord_subst transformation.
7825 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
7826 (round_saeonly): Also handle CCFP mode.
7827 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
7828 Remove UNSPEC_SAHF unspec handling.
7830 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7832 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
7834 2017-10-27 Jeff Law <law@redhat.com>
7836 * gimple-ssa-sprintf.c: Include domwalk.h.
7837 (class sprintf_dom_walker): New class, derived from dom_walker.
7838 (sprintf_dom_walker::before_dom_children): New function.
7839 (struct call_info): Moved into sprintf_dom_walker class
7840 (compute_formath_length, handle_gimple_call): Likewise.
7841 (sprintf_length::execute): Call the dominator walker rather
7842 than walking the statements.
7844 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
7845 gimple statement locations.
7846 (check_array_bounds): Corresponding changes. Get the statement's
7847 location directly from wi->stmt.
7849 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
7852 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
7854 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7856 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
7857 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
7859 2017-10-27 Jakub Jelinek <jakub@redhat.com>
7862 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
7863 * config/i386/i386.c (maybe_get_pool_constant): Removed.
7864 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
7865 maybe_get_pool_constant.
7866 * config/i386/predicates.md (zero_extended_scalar_load_operand):
7869 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7871 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
7872 2.26 caveat. Update gas and gld versions.
7873 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
7876 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7878 * cgraph.h (set_malloc_flag): Declare.
7879 * cgraph.c (set_malloc_flag_1): New function.
7880 (set_malloc_flag): Likewise.
7881 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
7882 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
7884 (read_ipa_call_summary): Add support for reading is_return_callee.
7885 (write_ipa_call_summary): Stream is_return_callee.
7886 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
7887 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
7888 ipa-prop.h, ipa-fnsummary.h.
7889 (pure_const_names): Change to static.
7890 (malloc_state_e): Define.
7891 (malloc_state_names): Define.
7892 (funct_state_d): Add field malloc_state.
7893 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
7894 (check_retval_uses): New function.
7895 (malloc_candidate_p): Likewise.
7896 (analyze_function): Add support for malloc attribute.
7897 (pure_const_write_summary): Stream malloc_state.
7898 (pure_const_read_summary): Add support for reading malloc_state.
7899 (dump_malloc_lattice): New function.
7900 (propagate_malloc): New function.
7901 (warn_function_malloc): New function.
7902 (ipa_pure_const::execute): Call propagate_malloc and
7903 ipa_free_fn_summary.
7904 (pass_local_pure_const::execute): Add support for malloc attribute.
7905 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
7906 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
7908 2017-10-27 Martin Liska <mliska@suse.cz>
7910 PR gcov-profile/82457
7911 * doc/invoke.texi: Document that one needs a non-strict ISO mode
7912 for fork-like functions to be properly instrumented.
7914 2017-10-27 Richard Biener <rguenther@suse.de>
7917 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
7918 info when we redirected EH.
7920 2017-10-26 Michael Collison <michael.collison@arm.com>
7922 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
7924 (<optab>_trunchf<GPI:mode>2: New pattern.
7925 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
7926 * config/aarch64/iterators.md (wv): New mode attribute.
7927 (vf, VF): New mode attributes.
7928 (vgp, VGP): New mode attributes.
7929 (s): Update attribute with SImode and DImode prefixes.
7931 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7933 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7934 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7935 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7936 (nios2_option_overide): Initialize it. Don't allow R0-relative
7937 addressing with PIC.
7938 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7939 (nios2_symbolic_constant_p): Likewise.
7940 (nios2_legitimate_address_p): Likewise.
7941 (nios2_r0rel_section_name_p): New.
7942 (nios2_symbol_ref_in_r0rel_data_p): New.
7943 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7944 (r0rel_constant_p): New.
7945 (nios2_print_operand_address): Handle r0rel_constant_p.
7946 (nios2_cdx_narrow_form_p): Likewise.
7947 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7948 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7949 (Nios II Options): Document -mr0rel-sec.
7951 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7953 * config/nios2/nios2.c: Include xregex.h.
7954 (nios2_gprel_sec_regex): New.
7955 (nios2_option_overide): Initialize it. Don't allow GP-relative
7956 addressing with PIC.
7957 (nios2_small_section_name_p): Check for regex match.
7958 * config/nios2/nios2.opt (mgprel-sec=): New option.
7959 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7960 (Nios II Options): Document -mgprel-sec.
7962 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7964 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7966 2017-10-26 Tom de Vries <tom@codesourcery.com>
7968 PR tree-optimization/82707
7969 * gimple.c (gimple_copy): Fix unsharing of
7970 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7972 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7974 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7975 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7976 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7977 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7978 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7979 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7980 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7981 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7982 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7983 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7984 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7985 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7986 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7987 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7988 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7989 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7990 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7992 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7994 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7996 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7997 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7998 warning to rs6000.c. Remove the Undocumented flag, since it has
8000 (-mabi=ibmlongdouble): Likewise.
8001 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
8002 already set, set the default format for long double.
8003 (rs6000_debug_reg_global): Print whether long double is IBM or
8005 (rs6000_option_override_internal): Rework setting long double
8006 format. Only warn if the user is changing the long double default
8007 and they did not use -Wno-psabi.
8008 * doc/invoke.texi (PowerPC options): Update the documentation for
8009 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
8011 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8012 Alan Hayward <alan.hayward@arm.com>
8013 David Sherwood <david.sherwood@arm.com>
8015 * rtl.h (wider_subreg_mode): New function.
8016 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
8017 rather than an unsigned int *.
8018 * ira-color.c (regno_max_ref_width): Replace with...
8019 (regno_max_ref_mode): ...this new variable.
8020 (coalesced_pseudo_reg_slot_compare): Update accordingly.
8021 Use wider_subreg_mode.
8022 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
8023 rather than an unsigned int *.
8024 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
8025 (process_alt_operands): Likewise.
8026 (invariant_p): Likewise.
8027 * lra-spills.c (assign_mem_slot): Likewise.
8028 (add_pseudo_to_slot): Likewise.
8029 * lra.c (collect_non_operand_hard_regs): Likewise.
8030 (add_regs_to_insn_regno_info): Likewise.
8031 * reload1.c (regno_max_ref_width): Replace with...
8032 (regno_max_ref_mode): ...this new variable.
8033 (reload): Update accordingly. Update call to
8034 ira_sort_regnos_for_alter_reg.
8035 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
8036 (init_eliminable_invariants): Update to use regno_max_ref_mode.
8037 (scan_paradoxical_subregs): Likewise.
8039 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8041 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
8042 (aarch64_frame): Add emit_frame_chain boolean.
8043 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8044 Move eh_return case to aarch64_layout_frame.
8045 (aarch64_layout_frame): Initialize emit_frame_chain.
8046 (aarch64_expand_prologue): Use emit_frame_chain.
8048 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8050 * config/aarch64/aarch64.c (aarch64_layout_frame):
8051 Ensure LR is always stored at the bottom of the callee-saves.
8052 Remove rarely used frame layout which saves callee-saves at top of
8053 frame, so the store of LR can be used as a valid probe in all cases.
8055 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
8057 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
8058 Improve unaligned TImode/TFmode base/offset split.
8060 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8061 Alan Hayward <alan.hayward@arm.com>
8062 David Sherwood <david.sherwood@arm.com>
8064 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
8065 * combine.c (find_single_use_1): Likewise.
8066 (expand_field_assignment): Likewise.
8067 (move_deaths): Likewise.
8068 * lra-constraints.c (simplify_operand_subreg): Likewise.
8069 (curr_insn_transform): Likewise.
8070 * lra.c (collect_non_operand_hard_regs): Likewise.
8071 (add_regs_to_insn_regno_info): Likewise.
8072 * rtlanal.c (reg_referenced_p): Likewise.
8073 (covers_regno_no_parallel_p): Likewise.
8075 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8077 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
8078 Don't print any bits outside the precision of the value.
8079 * wide-int.cc (test_printing): Add some new tests.
8081 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8083 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
8084 supports -xbrace_comment option.
8085 * configure: Regenerate.
8086 * config.in: Regenerate.
8087 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
8088 (ASM_CPU_SPEC): Use it.
8090 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8092 * target.def (static_rtx_alignment): New hook.
8093 * targhooks.h (default_static_rtx_alignment): Declare.
8094 * targhooks.c (default_static_rtx_alignment): New function.
8095 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
8096 * doc/tm.texi: Regenerate.
8097 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
8098 instead of targetm.constant_alignment. Remove call to
8100 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8101 (cris_preferred_mininum_alignment): New function, split out from...
8102 (cris_constant_alignment): ...here.
8103 (cris_static_rtx_alignment): New function.
8104 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
8106 (ix86_constant_alignment): ...here.
8107 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8108 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8109 (mmix_static_rtx_alignment): New function.
8110 * config/spu/spu.c (spu_static_rtx_alignment): New function.
8111 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8113 2017-10-26 Tamar Christina <tamar.christina@arm.com>
8116 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
8117 Add flag_trapping_math and flag_fp_int_builtin_inexact.
8119 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
8121 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
8122 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
8123 mark as a sign-extending load.
8124 (local_pic_load_u): Define.
8126 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
8129 * fold-const.c (operand_equal_for_comparison_p): Also return true
8130 if ARG0 is a simple variant of ARG1 with narrower precision.
8131 (fold_ternary_loc): Always pass unstripped operands to the predicate.
8133 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
8135 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
8137 * i386.h (processor_costs): Add gather_static, gather_per_elt,
8138 scatter_static, scatter_per_elt.
8139 * x86-tune-costs.h: Add new cost entries.
8141 2017-10-25 Richard Biener <rguenther@suse.de>
8143 * tree-ssa-sccvn.h (vn_eliminate): Declare.
8144 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
8145 class pass_fre): Move to ...
8146 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
8147 class pass_fre): ... here and adjust for statistics.
8149 2017-10-25 Jakub Jelinek <jakub@redhat.com>
8152 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
8153 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
8154 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
8155 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
8158 2017-10-25 Richard Biener <rguenther@suse.de>
8160 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
8161 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
8162 eliminate_push_avail, eliminate_insert): Move inside...
8163 (class eliminate_dom_walker): ... this class in preparation
8165 (fini_eliminate): Remove by merging with ...
8166 (eliminate): ... this function. Adjust for class changes.
8167 (pass_pre::execute): Remove fini_eliminate call.
8168 (pass_fre::execute): Likewise.
8170 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8173 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
8174 (VPERMI2, VPERMI2I): New mode iterators.
8175 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
8176 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
8178 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
8179 mode iterator. Remove 3 old define_insn patterns.
8180 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
8181 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
8182 VPERMI2 mode iterator, remove the other two expanders.
8183 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
8184 to use VPERMI2 mode iterator, add another alternative for vpermi2*
8185 instructions, remove the other two patterns.
8186 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
8187 mode iterator, remove the other two patterns.
8188 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
8189 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
8190 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
8191 and adjust argument order accordingly.
8192 (ix86_expand_vec_perm): Adjust caller.
8193 (expand_vec_perm_1): Likewise.
8194 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
8195 (expand_vec_perm_vpermt2_vpshub2): ... this.
8196 (ix86_expand_vec_perm_const_1): Adjust caller.
8197 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
8200 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
8201 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
8202 (vec_shl_<mode>): Remove unused expander.
8203 (avx512bw_<shift_insn><mode>3): New define_insn.
8204 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
8205 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
8207 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
8210 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
8213 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8215 PR rtl-optimization/82396
8216 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
8217 (autopref_multipass_init): Simplify initialization.
8218 (autopref_rank_data): Simplify sort order.
8219 * gcc/sched-int.h (autopref_multipass_data_): Remove
8220 multi_mem_insn_p, min_offset and max_offset.
8222 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8225 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8226 Check special value of flag_omit_frame_pointer.
8227 (aarch64_can_eliminate): Likewise.
8228 (aarch64_override_options_after_change_1): Simplify handling of
8229 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
8231 2017-10-24 Richard Biener <rguenther@suse.de>
8233 PR tree-optimization/82697
8234 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
8235 zero for conditional load and unconditional store.
8237 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8239 * doc/install.texi: Document bootstrap-cet.
8241 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8244 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
8245 ENDBR instruction at function entrance if function is only
8248 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8251 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
8252 patterns to better describe from which operation the CF is computed.
8253 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
8254 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
8255 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
8256 is 0, use _0 suffixed expanders instead of emitting a comparison
8259 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
8261 * config/i386/i386.md(*movsf_internal, *movdf_internal):
8262 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
8264 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
8267 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
8268 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
8269 * loop-iv.c (iv_get_reaching_def): Likewise.
8270 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
8271 variable is promoted and the partition contains undefined values.
8273 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8275 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
8277 (nios2_address_cost): Define.
8278 (nios2_legitimize_address): Recognize (exp + constant) directly.
8279 (TARGET_ADDRESS_COST): Define.
8281 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8283 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
8284 (nios2_symbolic_memory_operand_p): Declare.
8285 (nios2_split_large_constant): Declare.
8286 (nios2_split_symbolic_memory_operand): Declare.
8287 * config/nios2/nios2.c: Adjust includes.
8288 (nios2_symbolic_constant_allowed): New.
8289 (nios2_symbolic_constant_p): New.
8290 (nios2_plus_symbolic_constant_p): New.
8291 (nios2_valid_addr_expr_p): Recognize addresses involving
8293 (nios2_legitimate_address_p): Likewise, also LO_SUM.
8294 (nios2_symbolic_memory_operand_p): New.
8295 (nios2_large_constant_p): New.
8296 (nios2_split_large_constant): New.
8297 (nios2_split_plus_large_constant): New.
8298 (nios2_split_symbolic_memory_operand): New.
8299 (nios2_legitimize_address): Code refactoring. Handle addresses
8300 involving symbolic constants.
8301 (nios2_emit_move_sequence): Likewise.
8302 (nios2_print_operand): Improve error output.
8303 (nios2_print_operand_address): Handle LO_SUM.
8304 (nios2_cdx_narrow_form_p): Likewise.
8305 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
8306 operands involving symbolic constants.
8307 (movhi_internal, movsi_internal): Likewise.
8308 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
8309 (extendhisi2, extendqi<mode>2): Likewise.
8311 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8313 * tree-pass.h (PROP_rtl_split_insns): Define.
8314 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
8316 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8318 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
8320 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8323 * target.def (const_not_ok_for_debug_p): Default to
8324 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
8325 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
8326 * targhooks.c (default_const_not_ok_for_debug_p): New function.
8327 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
8328 which targetm.const_not_ok_for_debug_p returned true.
8329 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
8331 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
8333 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
8334 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
8335 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
8337 (ix86_const_not_ok_for_debug_p): New function.
8338 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
8339 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
8341 2017-10-23 David Malcolm <dmalcolm@redhat.com>
8344 * system.h: Conditionally include "unique-ptr.h" if
8345 INCLUDE_UNIQUE_PTR is defined.
8346 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
8347 of defining INCLUDE_UNIQUE_PTR before including "system.h".
8349 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
8351 * config/rl78/rl78.md: New define_expand "subdi3".
8353 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
8356 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
8357 DF_REF_INSN if DF_REF_INSN_INFO is false.
8359 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
8361 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
8362 xmm_move instead of sse_move.
8363 (sse_store_index): New function.
8364 (ix86_register_move_cost): Be more sensible about mismatch stall;
8365 model AVX moves correctly; make difference between sse->integer and
8367 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
8368 moves; make difference between SSE and AVX.
8369 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
8370 and zmm_move. Increase size of sse load and store tables;
8371 add unaligned load and store tables; add ssemmx_to_integer.
8372 * x86-tune-costs.h: Update all entries according to real
8373 move latencies from Agner Fog's manual and chip documentation.
8375 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8378 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
8379 * config/i386/constraints.md (Wf): New constraint.
8380 * config/i386/i386.md (UNSPEC_SBB): New unspec.
8381 (cmp<dwi>_doubleword): Removed.
8382 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
8383 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
8384 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
8385 expand with cmp<dwi>_doubleword. For LTU and GEU use
8386 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
8388 * common.opt (gcolumn-info): Enable by default.
8389 * doc/invoke.texi (gcolumn-info): Document new default.
8391 2017-10-23 Richard Biener <rguenther@suse.de>
8393 PR tree-optimization/82672
8394 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
8395 Fold the stmt if we propagated into it.
8397 2017-10-23 Richard Biener <rguenther@suse.de>
8399 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
8400 (bitmap_remove_expr_from_set): ... this. All callers call this
8401 for non-constant values.
8402 (bitmap_set_subtract): Rename to...
8403 (bitmap_set_subtract_expressions): ... this. Adjust and
8405 (bitmap_set_contains_value): Remove superfluous check.
8406 (bitmap_set_replace_value): Inline into single caller ...
8407 (bitmap_value_replace_in_set): ... here and simplify.
8408 (dependent_clean): Merge into ...
8409 (clean): ... this using an overload. Adjust.
8410 (prune_clobbered_mems): Adjust.
8411 (compute_antic_aux): Likewise.
8412 (compute_partial_antic_aux): Likewise.
8414 2017-10-23 Richard Biener <rguenther@suse.de>
8416 PR tree-optimization/82129
8418 2017-08-01 Richard Biener <rguenther@suse.de>
8420 PR tree-optimization/81181
8421 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8422 (compute_antic): ... end of iteration here.
8424 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8426 * target.def (starting_frame_offset): New hook.
8427 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
8428 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
8429 * doc/tm.texi.in: Regenerate.
8430 * hooks.h (hook_hwi_void_0): Declare.
8431 * hooks.c (hook_hwi_void_0): New function.
8432 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
8433 STARTING_FRAME_OFFSET.
8434 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
8435 * reload1.c (reload): Likewise.
8436 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
8437 instead of STARTING_FRAME_OFFSET.
8438 * function.c (try_fit_stack_local): Likewise.
8439 (assign_stack_local_1): Likewise
8440 (instantiate_virtual_regs): Likewise.
8441 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
8442 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
8443 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
8444 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
8445 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
8446 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
8447 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
8448 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
8449 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
8450 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
8451 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
8452 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
8453 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
8454 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
8455 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
8456 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
8457 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
8458 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
8459 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
8460 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
8461 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
8462 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
8463 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
8464 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
8465 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
8466 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
8467 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
8468 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
8469 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
8470 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
8471 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
8472 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
8473 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
8474 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
8475 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
8476 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
8477 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
8478 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
8479 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
8480 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
8481 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
8482 * config/avr/avr.c (avr_starting_frame_offset): Make static and
8483 return a HOST_WIDE_INT.
8484 (avr_builtin_setjmp_frame_value): Use it instead of
8485 STARTING_FRAME_OFFSET.
8486 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8487 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
8488 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
8490 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8491 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
8492 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
8493 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8494 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
8495 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
8496 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8497 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
8498 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
8499 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8500 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
8501 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
8503 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8504 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
8505 * config/mips/mips.c (mips_compute_frame_info): Refer to
8506 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8507 (mips_starting_frame_offset): New function.
8508 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8509 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
8510 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
8511 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
8512 and return a HOST_WIDE_INT.
8513 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8514 (mmix_initial_elimination_offset): Refer to
8515 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8516 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
8517 * config/pa/pa.c (pa_starting_frame_offset): New function.
8518 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
8519 (pa_expand_prologue): Likewise.
8520 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8521 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
8522 !FRAME_GROWS_DOWNWARD handling to...
8523 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8524 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
8525 !FRAME_GROWS_DOWNWARD handling to...
8526 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8527 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
8528 !FRAME_GROWS_DOWNWARD handling to...
8529 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8530 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
8532 (rs6000_starting_frame_offset): New function.
8533 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
8534 !FRAME_GROWS_DOWNWARD handling to...
8535 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8536 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
8537 !FRAME_GROWS_DOWNWARD handling to...
8538 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8539 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
8540 !FRAME_GROWS_DOWNWARD handling to...
8541 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8542 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
8543 (rs6000_starting_frame_offset): New function.
8544 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
8545 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
8546 * config/vax/vax.c (vax_starting_frame_offset): New function.
8547 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
8548 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8549 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
8550 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
8551 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8552 * system.h (STARTING_FRAME_OFFSET): Poison.
8554 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8556 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8557 SCALAR_TYPE_MODE instead of TYPE_MODE.
8559 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8560 Alan Hayward <alan.hayward@arm.com>
8561 David Sherwood <david.sherwood@arm.com>
8563 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
8565 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8566 Alan Hayward <alan.hayward@arm.com>
8567 David Sherwood <david.sherwood@arm.com>
8569 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
8571 2017-10-23 Richard Biener <rguenther@suse.de>
8573 PR tree-optimization/82129
8574 * tree-ssa-pre.c (bitmap_set_and): Remove.
8575 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
8576 canonicalizing expressions in the set to those with lowest
8577 ID rather than taking that from the first edge.
8579 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8581 * combine.c (rtx_equal_for_field_assignment_p): Use
8582 byte_lowpart_offset.
8584 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8585 Alan Hayward <alan.hayward@arm.com>
8586 David Sherwood <david.sherwood@arm.com>
8588 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
8589 to a SUBREG_PROMOTED_VAR.
8591 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8592 Alan Hayward <alan.hayward@arm.com>
8593 David Sherwood <david.sherwood@arm.com>
8595 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
8596 (expand_debug_source_expr): Likewise.
8597 * combine.c (combine_simplify_rtx): Likewise.
8598 * cse.c (fold_rtx): Likewise.
8599 * optabs.c (expand_float): Likewise.
8600 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8601 (simplify_binary_operation_1): Likewise.
8603 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8604 Alan Hayward <alan.hayward@arm.com>
8605 David Sherwood <david.sherwood@arm.com>
8607 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
8608 (record_promoted_value): Likewise.
8609 * expr.c (expand_expr_real_2): Likewise.
8610 * ree.c (update_reg_equal_equiv_notes): Likewise.
8611 (combine_set_extension): Likewise.
8612 * rtlanal.c (low_bitmask_len): Likewise.
8613 * simplify-rtx.c (neg_const_int): Likewise.
8614 (simplify_binary_operation_1): Likewise.
8616 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8617 Alan Hayward <alan.hayward@arm.com>
8618 David Sherwood <david.sherwood@arm.com>
8620 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
8621 * regcprop.c (maybe_mode_change): Likewise.
8622 * reload1.c (alter_reg): Likewise.
8624 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8626 * inchash.h (inchash::hash::add_wide_int): New function.
8627 * lto-streamer-out.c (hash_tree): Use it.
8629 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8631 * inchash.h (inchash::hash::add_wide_int): Rename to...
8632 (inchash::hash::add_hwi): ...this.
8633 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
8634 (polymorphic_call_target_hasher::hash): Likewise.
8635 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
8636 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
8637 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
8638 * lto-streamer-out.c (hash_tree): Likewise.
8639 * optc-save-gen.awk: Likewise.
8640 * tree.c (add_expr): Likewise.
8642 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8645 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
8646 for ordered inequality comparisons even with TARGET_IEEE_FP.
8648 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8651 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
8652 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8653 Expand with cmp<dwi>_doubleword.
8655 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8657 * extend.texi: Add x86 specific to 'nocf_check' attribute.
8658 List CET intrinsics.
8659 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
8660 specific to -fcf-protection option.
8662 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8664 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
8665 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
8666 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
8667 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
8668 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
8669 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
8670 (extra_objs): Add cet.o for Linux/x86 targets.
8671 (tmake_file): Add i386/t-cet for Linux/x86 targets.
8672 * config/i386/cet.c: New file.
8673 * config/i386/cetintrin.h: Likewise.
8674 * config/i386/t-cet: Likewise.
8675 * config/i386/cpuid.h (bit_SHSTK): New.
8676 (bit_IBT): Likewise.
8677 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
8678 pass IBT and SHSTK bits.
8679 * config/i386/i386-builtin-types.def
8680 (VOID_FTYPE_UNSIGNED_PVOID): New.
8681 (VOID_FTYPE_UINT64_PVOID): Likewise.
8682 * config/i386/i386-builtin.def: Add CET intrinsics.
8683 * config/i386/i386-c.c (ix86_target_macros_internal): Add
8684 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
8685 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
8686 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
8688 * config/i386/i386.c (rest_of_insert_endbranch): New.
8689 (pass_data_insert_endbranch): Likewise.
8690 (pass_insert_endbranch): Likewise.
8691 (make_pass_insert_endbranch): Likewise.
8692 (ix86_notrack_prefixed_insn_p): Likewise.
8693 (ix86_target_string): Add -mibt, -mshstk flags.
8694 (ix86_option_override_internal): Add flag_cf_protection
8696 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
8697 (ix86_print_operand): Add 'notrack' prefix output.
8698 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
8699 (ix86_expand_builtin): Expand CET intrinsics.
8700 (x86_output_mi_thunk): Add 'endbranch' instruction.
8701 * config/i386/i386.h (TARGET_IBT): New.
8702 (TARGET_IBT_P): Likewise.
8703 (TARGET_SHSTK): Likewise.
8704 (TARGET_SHSTK_P): Likewise.
8705 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
8706 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
8707 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
8708 (builtin_setjmp_setup): New pattern.
8709 (builtin_longjmp): Likewise.
8710 (rdssp<mode>): Likewise.
8711 (incssp<mode>): Likewise.
8712 (saveprevssp): Likewise.
8713 (rstorssp): Likewise.
8714 (wrss<mode>): Likewise.
8715 (wruss<mode>): Likewise.
8716 (setssbsy): Likewise.
8717 (clrssbsy): Likewise.
8718 (nop_endbr): Likewise.
8719 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
8721 * config/i386/immintrin.h: Include <cetintrin.h>.
8722 * config/i386/linux-common.h
8723 (file_end_indicate_exec_stack_and_cet): New prototype.
8724 (TARGET_ASM_FILE_END): New.
8726 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8728 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
8729 latencies instead of having separate table; make difference between
8730 integer and float costs.
8731 * i386.h (processor_costs): Remove scalar_stmt_cost,
8732 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
8733 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
8735 * x86-tune-costs.h: Remove entries which has been removed in
8736 procesor_costs from all tables; make cond_taken_branch_cost
8737 and cond_not_taken_branch_cost COST_N_INSNS based.
8739 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8741 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
8743 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8745 * config/i386/i386.md (isa): Remove fma_avx512f.
8746 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
8747 <avx512>_fmadd_<mode>_mask3<round_name>,
8748 <avx512>_fmsub_<mode>_mask<round_name>,
8749 <avx512>_fmsub_<mode>_mask3<round_name>,
8750 <avx512>_fnmadd_<mode>_mask<round_name>,
8751 <avx512>_fnmadd_<mode>_mask3<round_name>,
8752 <avx512>_fnmsub_<mode>_mask<round_name>,
8753 <avx512>_fnmsub_<mode>_mask3<round_name>,
8754 <avx512>_fmaddsub_<mode>_mask<round_name>,
8755 <avx512>_fmaddsub_<mode>_mask3<round_name>,
8756 <avx512>_fmsubadd_<mode>_mask<round_name>,
8757 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
8758 (*vec_widen_umult_even_v16si<mask_name>,
8759 *vec_widen_smult_even_v16si<mask_name>): Likewise.
8760 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
8762 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8764 * extend.texi: Add 'nocf_check' documentation.
8765 * gimple.texi: Add second parameter to
8766 gimple_build_call_from_tree.
8767 * invoke.texi: Add -fcf-protection documentation.
8768 * rtl.texi: Add REG_CALL_NOTRACK documenation.
8770 2017-10-20 Richard Biener <rguenther@suse.de>
8772 PR tree-optimization/82473
8773 * tree-vect-loop.c (vectorizable_reduction): Properly get at
8774 the largest input type.
8776 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8778 * c-attribs.c (handle_nocf_check_attribute): New function.
8779 (c_common_attribute_table): Add 'nocf_check' handling.
8780 * gimple-parser.c: Add second argument NULL to
8781 gimple_build_call_from_tree.
8782 * attrib.c (comp_type_attributes): Check nocf_check attribute.
8783 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
8785 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
8786 * common.opt: Add fcf-protection flag.
8787 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
8788 * flag-types.h: Add enum cf_protection_level.
8789 * gimple.c (gimple_build_call_from_tree): Add second parameter.
8790 Add 'nocf_check' attribute propagation to gimple call.
8791 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
8792 (gimple_build_call_from_tree): Update prototype.
8793 (gimple_call_nocf_check_p): New function.
8794 (gimple_call_set_nocf_check): Likewise.
8795 * gimplify.c: Add second argument to gimple_build_call_from_tree.
8796 * ipa-icf.c: Add nocf_check attribute in statement hash.
8797 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
8798 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
8799 * toplev.c (process_options): Add flag_cf_protection handling.
8801 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8803 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
8805 2017-10-20 Richard Biener <rguenther@suse.de>
8807 PR tree-optimization/82603
8808 * tree-if-conv.c (predicate_mem_writes): Make sure to only
8809 remove false predicated stores.
8811 2017-10-20 Richard Biener <rguenther@suse.de>
8813 * graphite-isl-ast-to-gimple.c
8814 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
8815 Remove return value and simplify, dump copied stmt after lhs
8817 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
8818 Reduce dump verbosity.
8819 (gsi_insert_earliest): Likewise.
8820 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
8821 * graphite.c (print_global_statistics): Adjust dumping.
8822 (print_graphite_scop_statistics): Likewise.
8823 (print_graphite_statistics): Do not dump loops here.
8824 (graphite_transform_loops): But here.
8826 2017-10-20 Nicolas Roche <roche@adacore.com>
8828 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
8829 * configure: Regenerate.
8831 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8834 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
8835 functions when optimizing replace GIMPLE_RETURN stmts with
8836 calls to __builtin_unreachable ().
8839 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
8840 for -fsanitize=thread link of executables.
8841 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
8842 link of executables.
8845 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
8847 (<shift_insn><mode>3<mask_name>): Change the last of the 3
8848 define_insns for logical vector shifts to use VI248_AVX512BW
8849 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
8850 condition, useless isa and prefix attributes. Change the first
8851 2 of these define_insns to ...
8852 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
8853 define_insn for avx512vl.
8854 (<shift_insn><mode>3): ... and this, new define_insn without
8855 masking for non-avx512vl.
8858 * config/i386/sse.md (*andnot<mode>3,
8859 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
8860 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
8861 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
8862 not applied use empty suffix even for TARGET_AVX512VL.
8863 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
8864 is applied, supply evex,evex or evex,evex,evex instead of just
8867 2017-10-20 Julia Koval <julia.koval@intel.com>
8869 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
8870 (OPTION_MASK_ISA_GFNI_UNSET): New.
8871 (ix86_handle_option): Handle OPT_mgfni.
8872 * config/i386/cpuid.h (bit_GFNI): New.
8873 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
8874 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
8875 * config/i386/i386.c (ix86_target_string): Add -mgfni.
8876 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
8877 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
8878 * config/i386/i386.opt: Add mgfni.
8880 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
8882 * config/msp430/msp430.c (msp430_option_override): Disable
8883 -fdelete-null-pointer-checks.
8884 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
8886 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8888 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
8889 of x87 and SSE instructions.
8891 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8893 * asan.c (create_cond_insert_point): Do not update edge count.
8894 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
8895 (afdo_propagate_circuit): Likewise.
8896 (afdo_calculate_branch_prob): Likewise.
8897 (afdo_annotate_cfg): Likewise.
8898 * basic-block.h (struct edge_def): Remove count.
8899 (edge_def::count): New accessor.
8900 * bb-reorder.c (rotate_loop): Update.
8901 (find_traces_1_round): Update.
8902 (connect_traces): Update.
8903 (sanitize_hot_paths): Update.
8904 * cfg.c (unchecked_make_edge): Update.
8905 (make_single_succ_edge): Update.
8906 (check_bb_profile): Update.
8907 (dump_edge_info): Update.
8908 (update_bb_profile_for_threading): Update.
8909 (scale_bbs_frequencies_int): Update.
8910 (scale_bbs_frequencies_gcov_type): Update.
8911 (scale_bbs_frequencies_profile_count): Update.
8912 (scale_bbs_frequencies): Update.
8913 * cfganal.c (connect_infinite_loops_to_exit): Update.
8914 * cfgbuild.c (compute_outgoing_frequencies): Update.
8915 (find_many_sub_basic_blocks): Update.
8916 * cfgcleanup.c (try_forward_edges): Update.
8917 (try_crossjump_to_edge): Update
8918 * cfgexpand.c (expand_gimple_cond): Update
8919 (expand_gimple_tailcall): Update
8920 (construct_exit_block): Update
8921 * cfghooks.c (verify_flow_info): Update
8922 (redirect_edge_succ_nodup): Update
8923 (split_edge): Update
8924 (make_forwarder_block): Update
8925 (duplicate_block): Update
8926 (account_profile_record): Update
8927 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8928 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8929 * cfgloopmanip.c (scale_loop_profile): Update.
8931 (lv_adjust_loop_entry_edge): Update.
8932 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8933 (force_nonfallthru_and_redirect): Update.
8934 (purge_dead_edges): Update.
8935 (rtl_flow_call_edges_add): Update.
8936 * cgraphunit.c (init_lowered_empty_function): Update.
8937 (cgraph_node::expand_thunk): Update.
8938 * gimple-pretty-print.c (dump_probability): Update.
8939 (dump_edge_probability): Update.
8940 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8941 * haifa-sched.c (sched_create_recovery_edges): Update.
8942 * hsa-gen.c (convert_switch_statements): Update.
8943 * ifcvt.c (dead_or_predicable): Update.
8944 * ipa-inline-transform.c (inline_transform): Update.
8945 * ipa-split.c (split_function): Update.
8946 * ipa-utils.c (ipa_merge_profiles): Update.
8947 * loop-doloop.c (add_test): Update.
8948 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8949 * lto-streamer-in.c (input_cfg): Update.
8950 (input_function): Update.
8951 * lto-streamer-out.c (output_cfg): Update.
8952 * modulo-sched.c (sms_schedule): Update.
8953 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8954 * predict.c (maybe_hot_edge_p): Update.
8955 (unlikely_executed_edge_p): Update.
8956 (probably_never_executed_edge_p): Update.
8957 (dump_prediction): Update.
8958 (drop_profile): Update.
8959 (propagate_unlikely_bbs_forward): Update.
8960 (determine_unlikely_bbs): Update.
8961 (force_edge_cold): Update.
8962 * profile.c (compute_branch_probabilities): Update.
8963 * reg-stack.c (better_edge): Update.
8964 * shrink-wrap.c (handle_simple_exit): Update.
8965 * tracer.c (better_p): Update.
8966 * trans-mem.c (expand_transaction): Update.
8967 (split_bb_make_tm_edge): Update.
8968 * tree-call-cdce.c: Update.
8969 * tree-cfg.c (gimple_find_sub_bbs): Update.
8970 (gimple_split_edge): Update.
8971 (gimple_duplicate_sese_region): Update.
8972 (gimple_duplicate_sese_tail): Update.
8973 (gimple_flow_call_edges_add): Update.
8974 (insert_cond_bb): Update.
8975 (execute_fixup_cfg): Update.
8976 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8977 * tree-complex.c (expand_complex_div_wide): Update.
8978 * tree-eh.c (lower_resx): Update.
8979 (unsplit_eh): Update.
8980 (cleanup_empty_eh_move_lp): Update.
8981 * tree-inline.c (copy_edges_for_bb): Update.
8982 (freqs_to_counts): Update.
8983 (copy_cfg_body): Update.
8984 * tree-ssa-dce.c (remove_dead_stmt): Update.
8985 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8986 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8987 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8988 (unloop_loops): Update.
8989 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8990 * tree-ssa-loop-split.c (connect_loops): Update.
8991 (split_loop): Update.
8992 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8993 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8994 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8995 * tree-ssa-reassoc.c (branch_fixup): Update.
8996 * tree-ssa-tail-merge.c (replace_block_by): Update.
8997 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8998 (compute_path_counts): Update.
8999 (update_profile): Update.
9000 (recompute_probabilities): Update.
9001 (update_joiner_offpath_counts): Update.
9002 (estimated_freqs_path): Update.
9003 (freqs_to_counts_path): Update.
9004 (clear_counts_path): Update.
9005 (ssa_fix_duplicate_block_edges): Update.
9006 (duplicate_thread_path): Update.
9007 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9008 (case_bit_test_cmp): Update.
9009 (collect_switch_conv_info): Update.
9010 (gen_inbound_check): Update.
9011 (do_jump_if_equal): Update.
9012 (emit_cmp_and_jump_insns): Update.
9013 * tree-tailcall.c (decrease_profile): Update.
9014 (eliminate_tail_call): Update.
9015 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9016 (vect_do_peeling): Update.
9017 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9018 * ubsan.c (ubsan_expand_null_ifn): Update.
9019 (ubsan_expand_ptr_ifn): Update.
9020 * value-prof.c (gimple_divmod_fixed_value): Update.
9021 (gimple_mod_pow2): Update.
9022 (gimple_mod_subtract): Update.
9023 (gimple_ic): Update.
9024 (gimple_stringop_fixed_value): Update.
9026 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9029 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
9031 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
9033 PR rtl-optimization/82395
9034 * ira-color.c (allocno_priority_compare_func): Fix comparison step
9035 based on non_spilled_static_chain_regno_p.
9037 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
9039 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
9040 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
9041 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
9043 2017-10-19 Martin Sebor <msebor@redhat.com>
9045 PR tree-optimization/82596
9046 * tree.c (array_at_struct_end_p): Handle STRING_CST.
9048 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9050 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
9051 (get_mem_refs_of_builtin_call): Likewise.
9052 * builtins.c (expand_builtin_apply): Adjust call to
9053 allocate_dynamic_stack_space.
9054 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
9055 the third argument to allocate_dynamic_stack_space, otherwise -1.
9056 (expand_builtin): Deal with all alloca variants.
9057 (is_inexpensive_builtin): Likewise.
9058 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
9059 * calls.c (special_function_p): Deal with all alloca variants.
9060 (initialize_argument_information): Adjust call to
9061 allocate_dynamic_stack_space.
9062 (expand_call): Likewise.
9063 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
9064 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
9065 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
9066 use it for the stack usage computation.
9067 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
9068 * function.c (gimplify_parameters): Call build_alloca_call_expr.
9069 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
9070 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
9071 (in_loop_p): Remove first argument and useless check.
9072 (pass_walloca::execute): Remove useless test and adjust call to above.
9073 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
9074 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
9075 (gimplify_call_expr): Deal with all alloca variants.
9076 * hsa-gen.c (gen_hsa_alloca): Likewise.
9077 (gen_hsa_insns_for_call): Likewise.
9078 * ipa-pure-const.c (special_builtin_state): Likewise.
9079 * tree-chkp.c (chkp_build_returned_bound): Likewise.
9080 * tree-object-size.c (alloc_object_size): Likewise.
9081 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
9082 (call_may_clobber_ref_p_1): Likewise.
9083 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
9084 (ccp_fold_stmt): Likewise.
9085 (optimize_stack_restore): Likewise.
9086 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9087 (mark_all_reaching_defs_necessary_1): Likewise.
9088 (propagate_necessity): Likewise.
9089 (eliminate_unnecessary_stmts): Likewise.
9090 * tree.c (build_common_builtin_nodes): Build
9091 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
9092 (build_alloca_call_expr): New function.
9093 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
9094 (CASE_BUILT_IN_ALLOCA): Likewise.
9095 (build_alloca_call_expr): Declare.
9096 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
9098 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9101 * dwarf2out.c (new_die_raw): New static inline function.
9102 (new_die): Use it to create the DIE.
9103 (add_AT_external_die_ref): Likewise.
9104 (clone_die): Likewise.
9105 (clone_as_declaration): Likewise.
9106 (dwarf2out_vms_debug_main_pointer): Likewise.
9107 (base_type_die): Likewise. Remove early return for corner cases.
9108 Do not call add_pubtype on the DIE here.
9109 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
9110 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
9111 typedefs for base types with DW_AT_endianity. Make sure a DIE with
9112 native order exists for base types, attach the DIE manually and call
9113 add_pubtype on it. Do not equate a reverse order DIE to the type.
9115 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
9117 * config/arm/arm.c (align_ok_ldrd_strd): New function.
9118 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
9120 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
9122 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9124 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
9125 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
9126 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
9127 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
9128 * opts.c (sanitizer_opts): Add builtin.
9129 * ubsan.c (instrument_builtin): New function.
9130 (pass_ubsan::execute): Call it.
9131 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
9132 * doc/invoke.texi: Document -fsanitize=builtin.
9134 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
9135 builtins, store max (log2 (align), 0) into uchar field instead of
9136 align into uptr field.
9137 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
9138 store uchar 0 field instead of uptr 0 field.
9139 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
9140 instead of passing one address of struct with 2 locations pass
9141 two addresses of structs with 1 location each.
9142 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
9143 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
9144 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
9145 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
9146 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
9147 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
9148 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
9149 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
9151 2017-10-19 Martin Liska <mliska@suse.cz>
9154 * file-find.c (remove_prefix): Remove.
9155 * file-find.h (remove_prefix): Likewise.
9156 * gcc-ar.c: Remove smartness of lookup.
9158 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
9160 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
9161 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
9162 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
9164 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9167 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
9168 (setcc + and to xor + setcc): New peephole2.
9170 2017-10-19 Tom de Vries <tom@codesourcery.com>
9172 * doc/sourcebuild.texi (Test Directives, Variants of
9173 dg-require-support): Add dg-require-stack-size.
9175 2017-10-19 Martin Liska <mliska@suse.cz>
9178 * gimplify.c (gimplify_decl_expr): Do not instrument variables
9179 that have a large alignment.
9180 (gimplify_target_expr): Likewise.
9182 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
9184 PR rtl-optimization/82602
9185 * ira.c (rtx_moveable_p): Return false for volatile asm.
9187 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
9190 * config/i386/i386-modes.def (CCGZ): New CC mode.
9191 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
9192 * config/i386/predicates.md (ix86_comparison_operator):
9194 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9195 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
9196 with double-word subtraction.
9197 (put_condition_code): Handle CCGZmode.
9199 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
9201 * wide-int.cc (debug (const wide_int &)): New.
9202 (debug (const wide_int *)): New.
9203 (debug (const widest_int &)): New.
9204 (debug (const widest_int *)): New.
9206 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
9209 * lra-constraints.c (curr_insn_transform): Use non-input operand
9210 instead of output one for matched reload.
9212 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9214 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
9215 (tree-ssa-loop-ivopts.h): New header file.
9216 (struct builtin_info): New fields.
9217 (classify_builtin_1): Compute and record base and offset parts for
9218 memset builtin partition by calling strip_offset.
9219 (offset_cmp, fuse_memset_builtins): New functions.
9220 (finalize_partitions): Fuse adjacent memset partitions by calling
9222 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
9223 Expose the interface.
9224 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
9226 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9228 PR tree-optimization/82574
9229 * tree-loop-distribution.c (find_single_drs): New parameter. Check
9230 that data reference must be executed exactly once per iteration
9231 against the outermost loop in nest.
9232 (classify_partition): Update call to above function.
9234 2017-10-18 Richard Biener <rguenther@suse.de>
9236 PR tree-optimization/82591
9237 * graphite.c (graphite_transform_loops): Move code gen message
9239 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9240 Here. Handle scop_to_isl_ast failing.
9241 (scop_to_isl_ast): Limit the number of ISL operations.
9243 2017-10-18 Richard Biener <rguenther@suse.de>
9245 * graphite-isl-ast-to-gimple.c
9246 (translate_isl_ast_to_gimple::set_rename): Simplify.
9247 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
9248 (graphite_copy_stmts_from_block): ... here.
9249 (copy_bb_and_scalar_dependences): Simplify.
9250 (add_parameters_to_ivs_params): Canonicalize.
9251 (generate_entry_out_of_ssa_copies): Simplify.
9252 * graphite-sese-to-poly.c (extract_affine_name): Simplify
9253 by passing in ISL dimension.
9254 (parameter_index_in_region_1): Rename to ...
9255 (parameter_index_in_region): ... this.
9256 (extract_affine): Adjust assert, pass down parameter index.
9257 (add_param_constraints): Use range-info when available.
9258 (build_scop_context): Adjust.
9259 * sese.c (new_sese_info): Adjust.
9260 (free_sese_info): Likewise.
9261 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
9262 Remove unused typedefs.
9263 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
9265 2017-10-18 Martin Liska <mliska@suse.cz>
9267 * combine.c (simplify_compare_const): Add gcc_fallthrough.
9269 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9271 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
9272 (s390_sched_init): Do not reset s390_sched_state if we entered the
9273 current basic block via a fallthru edge and all others are unlikely.
9275 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9277 * config/s390/s390.c (NUM_SIDES): New variable.
9278 (LONGRUNNING_THRESHOLD): New variable.
9279 (LATENCY_FACTOR): New variable.
9280 (s390_sched_score): Decrease score for long-running instructions on
9282 (s390_sched_variable_issue): Perform bookkeeping for long-running
9285 2017-10-18 Richard Biener <rguenther@suse.de>
9287 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
9288 Simplify with removal of the parameter rename map.
9289 (set_rename): Likewise.
9290 (should_copy_to_new_region): Likewise.
9291 (graphite_copy_stmts_from_block): Likewise.
9292 (copy_bb_and_scalar_dependences): Remove initialization of
9293 unused copied_bb_map.
9295 (copy_internal_parameters): Likewise.
9296 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
9297 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
9298 Use INTEGRAL_TYPE_P.
9299 (parameter_index_in_region_1): Rename to ...
9300 (assign_parameter_index_in_region): ... this. Assert we have
9301 a parameter we handle.
9302 (scan_tree_for_params): Adjust.
9303 * sese.h (parameter_rename_map_t): Remove.
9304 (struct sese_info_t): Remove unused parameter_rename_map and
9305 copied_bb_map members.
9306 * sese.c (new_sese_info): Adjust.
9307 (free_sese_info): Likewise.
9309 2017-10-18 Martin Liska <mliska@suse.cz>
9312 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
9313 on an abnormal edge.
9315 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
9317 * doc/invoke.texi (ffunction-sections and fdata-sections):
9320 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9322 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
9323 the use statement can throw internally.
9325 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9327 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
9328 any RTX present on the RHS of a SET.
9329 * compare-elim.c (try_eliminate_compare): Restore comment.
9331 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9333 * langhooks.h (struct lang_hooks): Document that tree_size langhook
9334 may be also called on tcc_type nodes.
9335 * langhooks.c (lhd_tree_size): Likewise.
9337 2017-10-17 David Malcolm <dmalcolm@redhat.com>
9339 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
9340 format_warning_at_substring.
9341 (maybe_warn): Convert source_range * param to a location_t. Pass
9342 UNKNOWN_LOCATION rather than NULL to fmtwarn.
9343 (format_directive): Remove code to extract source_ranges and
9344 source_range * in favor of just a location_t.
9345 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
9347 * substring-locations.c (format_warning_va): Convert
9348 source_range * param to a location_t.
9349 (format_warning_at_substring): Likewise.
9350 * substring-locations.h (format_warning_va): Likewise.
9351 (format_warning_at_substring): Likewise.
9353 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
9355 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
9357 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
9358 and scatter/gather ops.
9360 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
9361 vec_gather_load and vec_scatter_store.
9362 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
9363 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
9364 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
9365 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
9366 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
9367 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
9369 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
9371 * reg-stack.c (compare_for_stack_reg): Add bool argument.
9372 Detect FTST instruction and handle its register pops. Only pop
9373 second operand if can_pop_second_op is true.
9374 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
9375 set can_pop_second_op to false in the compare_for_stack_reg call.
9377 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
9378 output_fp_compare for stack register operands.
9379 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
9380 instructions here. Do not emit stack register pops here. Assert
9381 that FCOMPP pops next to top stack register. Rewrite function.
9383 2017-10-17 Nathan Sidwell <nathan@acm.org>
9386 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
9387 use DECL_ASSEMBLER_NAME_RAW.
9390 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
9393 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
9394 Wilco Dijkstra <wilco.dijkstra@arm.com>
9396 * builtins.c (expand_builtin_update_setjmp_buf): Add a
9397 converstion to Pmode from the buf_addr.
9399 2017-10-17 Richard Biener <rguenther@suse.de>
9401 * graphite-dependences.c (scop_get_reads_and_writes): Change
9402 output parameters to references.
9404 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
9406 PR 71026/tree-optimization
9407 * fold-const.c (distribute_real_division): Removed.
9408 (fold_binary_loc): Remove calls to distribute_real_divison.
9410 2017-10-17 Richard Biener <rguenther@suse.de>
9412 * graphite-scop-detection.c
9413 (scop_detection::stmt_has_simple_data_refs_p): Always use
9414 the full nest as region.
9415 (try_generate_gimple_bb): Likewise.
9416 * sese.c (scalar_evolution_in_region): Simplify now that
9417 SCEV can handle instantiation in regions.
9418 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
9419 in the non-loop part of a function if requested.
9421 2017-10-17 Richard Biener <rguenther@suse.de>
9423 PR tree-optimization/82563
9424 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
9426 (graphite_regenerate_ast_isl): Call it.
9427 * graphite-scop-detection.c (build_scops): Remove entry edge split.
9429 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9431 PR tree-optimization/82549
9432 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
9433 Formatting fixes. Instead of calling make_bit_field_ref with negative
9436 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
9438 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
9439 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
9440 _mm_maskz_reduce_ss): New.
9441 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
9442 __builtin_ia32_reducess_mask): Ditto..
9443 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
9444 * config/i386/sse.md (reduces<mode>): Renamed to ...
9445 (reduces<mode><mask_scalar_name>): ... this.
9446 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
9448 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
9449 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
9451 2017-10-16 David Malcolm <dmalcolm@redhat.com>
9453 * Makefile.in (OBJS): Add unique-ptr-tests.o.
9454 * selftest-run-tests.c (selftest::run_tests): Call
9455 selftest::unique_ptr_tests_cc_tests.
9456 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
9457 * unique-ptr-tests.cc: New file.
9459 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
9462 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9464 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9465 (make_hard_regno_born, make_hard_regno_dead): Update
9466 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
9468 2017-10-16 Jeff Law <law@redhat.com>
9470 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9472 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
9474 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
9476 2017-10-16 Olivier Hainque <hainque@adacore.com>
9478 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
9479 with_cpu if we were configured for an e500v2 target cpu name.
9481 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
9483 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
9484 * doc/invoke.texi: Document +nodsp as a valid extension for
9487 2017-10-16 Martin Liska <mliska@suse.cz>
9489 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
9490 (test_set_range): Likewise.
9491 (test_range_functions): Rename to ...
9492 (test_bit_in_range): ... this.
9493 (sbitmap_c_tests): Add new test.
9495 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9497 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
9499 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
9500 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
9502 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9504 * config/aarch64/aarch64-builtins.c
9505 (aarch64_types_quadopu_lane_qualifiers): New.
9506 (TYPES_QUADOPU_LANE): New.
9507 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
9508 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
9509 (aarch64_<sur>dot_laneq<vsi2qi>): New.
9510 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
9511 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
9512 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
9513 (Vdottype, DOTPROD): New.
9514 (sur): Add SDOT and UDOT.
9516 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9518 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
9519 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
9520 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
9522 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
9523 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
9524 Enable TARGET_DOTPROD.
9525 (cortex-a75.cortex-a55): Likewise.
9526 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
9528 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9530 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
9531 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
9533 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
9535 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
9536 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
9537 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
9538 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
9539 * config/arm/types.md (neon_dot, neon_dot_q): New.
9540 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
9542 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9544 * config/arm/arm.h (TARGET_DOTPROD): New.
9545 * config/arm/arm.c (arm_arch_dotprod): New.
9546 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
9547 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
9548 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
9549 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
9550 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
9551 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
9552 * doc/invoke.texi (armv8.2-a): Document dotprod
9554 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9556 * i386.c (ix86_vec_cost): New function.
9557 (ix86_rtx_costs): Handle vector operations better.
9558 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
9559 * x86-tune-costs.h: Add new costs to all tables.
9561 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9563 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
9565 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
9566 divsd, sqrtss and sqrtsd
9567 * x86-tune-costs.h: Add new entries to all costs.
9568 (znver1_cost): Fix to match real instruction latencies.
9570 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9571 Michael Collison <michael.collison@arm.com>
9573 * compare-elim.c: Include emit-rtl.h.
9574 (can_merge_compare_into_arith): New function.
9575 (try_validate_parallel): Likewise.
9576 (try_merge_compare): Likewise.
9577 (try_eliminate_compare): Call the above when no previous clobber
9579 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
9582 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9586 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
9587 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
9591 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
9592 to be any operand_equal_p operands. For & (B - 1) require
9593 B to be power of 2. Recognize
9594 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
9596 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9599 * optabs.c (expand_memory_blockage): Fix call of
9600 targetm.have_memory_blockage.
9602 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9605 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
9606 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
9607 objects to extra_objs instead of overwriting it.
9609 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9611 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
9612 Use any_fp_register_operand as operand[3] predicate. Simplify
9613 equality test for operands[2] and operands[4] memory location.
9614 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
9615 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
9616 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
9617 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
9618 any_fp_register_operand as operand[1] predicate. Simplify
9619 equality test for operands[0] and operands[3] memory location.
9620 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
9621 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
9622 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
9624 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9626 * target-insns.def: Add memory_blockage.
9627 * optabs.c (expand_memory_blockage): New function.
9628 (expand_asm_memory_barrier): Rename ...
9629 (expand_asm_memory_blockage): ... to this.
9630 (expand_mem_thread_fence): Call expand_memory_blockage
9631 instead of expand_asm_memory_barrier.
9632 (expand_mem_singnal_fence): Ditto.
9633 (expand_atomic_load): Ditto.
9634 (expand_atomic_store): Ditto.
9635 * doc/md.texi (Standard Pattern Names For Generation):
9636 Document memory_blockage instruction pattern.
9638 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
9640 * config/rl78/rl78.c (rl78_emit_libcall): New function.
9641 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
9642 * config/rl78/rl78.md: New define_expand "adddi3".
9644 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9646 * cfghooks.c (verify_flow_info): Disable check that all probabilities
9649 2017-10-13 Jeff Law <law@redhat.com>
9651 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
9653 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9656 * internal-fn.c (expand_mul_overflow): If both operands have
9657 the same highpart of -1 or 0 and the topmost bit of lowpart
9658 is different, overflow is if res <= 0 rather than res < 0.
9660 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
9662 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
9663 TARGET_P9_VECTOR code for unaligned_load case.
9665 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9667 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
9669 2017-10-13 Nathan Sidwell <nathan@acm.org>
9671 * tree-core.h (tree_contains_struct): Make bool.
9672 * tree.c (tree_contains_struct): Likewise.
9673 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
9674 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
9675 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
9676 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
9677 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
9679 2017-10-13 Richard Biener <rguenther@suse.de>
9681 * graphite-isl-ast-to-gimple.c
9682 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
9683 parameters and dominance check.
9684 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
9685 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
9686 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
9687 Do not update SSA form here or do intermediate IL verification.
9688 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
9689 (graphite_initialize): Remove check on the number of loops in
9690 the function and inline into graphite_transform_loops.
9691 (graphite_finalize): Inline into graphite_transform_loops.
9692 (graphite_transform_loops): Perform SSA update and IL verification
9694 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
9696 2017-10-13 Richard Biener <rguenther@suse.de>
9698 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
9699 graphite_expression_type_precision): Avoid global constructor
9701 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
9702 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
9703 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
9704 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
9705 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
9706 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
9708 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
9711 * config/i386/i386.h (ix86_red_zone_size): New.
9712 * config/i386/i386.md (push peephole2s): Replace
9713 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
9715 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9716 Alan Hayward <alan.hayward@arm.com>
9717 David Sherwood <david.sherwood@arm.com>
9719 * combine.c (can_change_dest_mode): Reject changes in
9720 REGMODE_NATURAL_SIZE.
9722 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9723 Alan Hayward <alan.hayward@arm.com>
9724 David Sherwood <david.sherwood@arm.com>
9726 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
9727 (expand_debug_source_expr): Likewise.
9728 * combine.c (combine_simplify_rtx): Likewise.
9729 * cse.c (fold_rtx): Likewise.
9730 * fwprop.c (canonicalize_address): Likewise.
9731 * targhooks.c (default_shift_truncation_mask): Likewise.
9733 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9734 Alan Hayward <alan.hayward@arm.com>
9735 David Sherwood <david.sherwood@arm.com>
9737 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
9738 (widened_mode): Likewise.
9739 (expand_unop): Likewise.
9740 * ree.c (transform_ifelse): Likewise.
9741 (merge_def_and_ext): Likewise.
9742 (combine_reaching_defs): Likewise.
9743 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9745 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9746 Alan Hayward <alan.hayward@arm.com>
9747 David Sherwood <david.sherwood@arm.com>
9749 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
9750 * combine.c (gen_lowpart_for_combine): Likewise.
9751 * dwarf2out.c (rtl_for_decl_location): Likewise.
9752 * final.c (alter_subreg): Likewise.
9753 * rtlhooks.c (gen_lowpart_general): Likewise.
9754 (gen_lowpart_if_possible): Likewise.
9756 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9757 Alan Hayward <alan.hayward@arm.com>
9758 David Sherwood <david.sherwood@arm.com>
9760 * calls.c (expand_call): Use subreg_lowpart_offset.
9761 * cse.c (cse_insn): Likewise.
9762 * regcprop.c (copy_value): Likewise.
9763 (copyprop_hardreg_forward_1): Likewise.
9765 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9768 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
9769 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
9770 =Q constraints to +Q and into insn condition add check
9771 that operands[0] and operands[1] are equal.
9772 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
9773 =Q constraints to +Q and into insn condition add check
9774 that operands[0] is equal to either operands[1] or operands[2].
9777 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
9778 instead of handling MINUS_EXPR twice (once for each argument),
9779 canonicalize operand order and handle just once, use rtype where
9780 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
9783 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
9784 any values of __C while still being pattern recognizable as a simple
9787 2017-10-13 Richard Biener <rguenther@suse.de>
9789 PR tree-optimization/82451
9791 2017-10-02 Richard Biener <rguenther@suse.de>
9793 PR tree-optimization/82355
9794 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9795 a mapping for the enclosing loop but avoid generating one for
9797 (copy_bb_and_scalar_dependences): Remove premature codegen
9798 error on PHIs in blocks duplicated into multiple places.
9799 * graphite-scop-detection.c
9800 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9801 in the region use it as loop and nest to analyze the DR in.
9802 (try_generate_gimple_bb): Likewise.
9803 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9804 (add_loop_constraints): For blocks in a loop not in the region
9805 create a dimension with a single iteration.
9806 * sese.h (gbb_loop_at_index): Remove assert.
9808 * cfgloop.c (loop_preheader_edge): For the loop tree root
9809 return the single successor of the entry block.
9810 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9811 Reset the SCEV hashtable and niters.
9812 * graphite-scop-detection.c
9813 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
9814 assert that we only have POLYNOMIAL_CHREC that vary in loops
9815 contained in the region.
9816 (scop_detection::graphite_can_represent_expr): Adjust.
9817 (scop_detection::stmt_has_simple_data_refs_p): For loops
9818 not in the region set loop to NULL. The nest is now the
9819 entry edge to the region.
9820 (try_generate_gimple_bb): Likewise.
9821 * sese.c (scalar_evolution_in_region): Adjust for
9822 instantiate_scev change.
9823 * tree-data-ref.h (graphite_find_data_references_in_stmt):
9824 Make nest parameter the edge into the region.
9825 (create_data_ref): Likewise.
9826 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
9827 entry edge into a region and adjust instantiate_scev calls.
9828 (create_data_ref): Likewise.
9829 (graphite_find_data_references_in_stmt): Likewise.
9830 (find_data_references_in_stmt): Pass the loop preheader edge
9831 from the nest argument.
9832 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
9833 parameter the edge into the region.
9834 (instantiate_parameters): Use the loop preheader edge as entry.
9835 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
9837 (get_instantiated_value_entry): Make instantiate_below parameter
9838 the edge into the region.
9839 (instantiate_scev_name): Likewise. Adjust dominance checks,
9840 when we cannot use loop-based instantiation instantiate by
9841 walking use-def chains.
9842 (instantiate_scev_poly): Adjust.
9843 (instantiate_scev_binary): Likewise.
9844 (instantiate_scev_convert): Likewise.
9845 (instantiate_scev_not): Likewise.
9846 (instantiate_array_ref): Remove.
9847 (instantiate_scev_3): Likewise.
9848 (instantiate_scev_2): Likewise.
9849 (instantiate_scev_1): Likewise.
9850 (instantiate_scev_r): Do not blindly handle N-operand trees.
9851 Do not instantiate array-refs. Handle all constants and invariants.
9852 (instantiate_scev): Make instantiate_below parameter
9853 the edge into the region.
9854 (resolve_mixers): Use the loop preheader edge for the region
9855 parameter to instantiate_scev_r.
9856 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
9858 2017-10-13 Richard Biener <rguenther@suse.de>
9860 PR tree-optimization/82525
9861 * graphite-isl-ast-to-gimple.c
9862 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
9864 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
9865 Fail code generation when we cannot represent the isl integer.
9866 (binary_op_to_tree): Elide modulo operations that are no-ops
9867 in the type we code generate. Remove now superfluous code
9870 2017-10-13 Richard Biener <rguenther@suse.de>
9872 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
9873 (scop_detection::harmful_loop_in_region): Remove premature
9874 IV type restriction.
9875 (scop_detection::graphite_can_represent_scev): We can handle
9876 pointer IVs just fine.
9878 2017-10-13 Alan Modra <amodra@gmail.com>
9880 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
9881 "Clobbers and Scratch Registers". Add paragraph on
9882 alternative to clobbers for scratch registers and OpenBLAS
9885 2017-10-13 Alan Modra <amodra@gmail.com>
9887 * doc/extend.texi (Clobbers): Correct vax example. Delete old
9888 example of a memory input for a string of known length. Move
9889 commentary out of table. Add a number of new examples
9890 covering array memory inputs.
9892 2017-10-12 Martin Liska <mliska@suse.cz>
9894 PR tree-optimization/82493
9895 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
9896 (test_range_functions): New function.
9897 (sbitmap_c_tests): Likewise.
9898 * selftest-run-tests.c (selftest::run_tests): Run new tests.
9899 * selftest.h (sbitmap_c_tests): New function.
9901 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9903 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9905 * config/rs6000/amo.h: Fix spacing issue.
9907 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9910 * config/i386/i386.md (*ashl<mode>3_mask_1,
9911 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
9912 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
9915 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9917 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
9918 (profile_probability): Set max_probability
9919 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
9921 * profile-count.c (profile_probability::differs_from_p): Do not
9922 rely on max_probaiblity == 10000
9924 2017-10-12 Jeff Law <law@redhat.com>
9926 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9929 2017-10-12 Martin Sebor <msebor@redhat.com>
9933 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9934 (handle_alias_pairs): Call it.
9935 * common.opt (-Wattribute-alias): New option.
9936 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9937 * doc/invoke.texi (-Wattribute-alias): Document.
9939 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9942 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9944 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9946 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9947 (make_hard_regno_born, make_hard_regno_dead): Update
9948 bb_killed_pseudos and bb_gen_pseudos.
9950 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9952 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9954 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9956 * config/alpha/alpha.c (alpha_split_conditional_move):
9957 Use std::swap instead of manually swapping.
9958 (alpha_stdarg_optimize_hook): Ditto.
9959 (alpha_canonicalize_comparison): Ditto.
9961 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9963 * tree-loop-distribution.c (struct builtin_info): New struct.
9964 (struct partition): Refactor fields into struct builtin_info.
9965 (partition_free): Free struct builtin_info.
9966 (build_size_arg_loc, build_addr_arg_loc): Delete.
9967 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9968 information from struct builtin_info.
9969 (find_single_drs): New function refactored from classify_partition.
9970 Also moved builtin validity checks to this function.
9971 (compute_access_range, alloc_builtin): New functions.
9972 (classify_builtin_st, classify_builtin_ldst): New functions.
9973 (classify_partition): Refactor code into functions find_single_drs,
9974 classify_builtin_st and classify_builtin_ldst.
9975 (distribute_loop): Don't do runtime alias check when distributing
9977 (find_seed_stmts_for_distribution): New function.
9978 (pass_loop_distribution::execute): Refactor code finding seed
9979 stmts into above function. Support distribution for the innermost
9980 two-level loop nest. Adjust dump information.
9982 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9984 * tree-loop-distribution.c: Adjust the general comment.
9985 (NUM_PARTITION_THRESHOLD): New macro.
9986 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9987 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9988 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9989 in call to build_partition_graph.
9990 (finalize_partitions): New parameter. Make loop distribution more
9991 conservative by fusing more partitions.
9992 (distribute_loop): Don't do runtime alias check in case of loop nest
9994 (find_seed_stmts_for_distribution): New function.
9995 (prepare_perfect_loop_nest): New function.
9996 (pass_loop_distribution::execute): Refactor code finding seed stmts
9997 and loop nest into above functions. Support loop nest distribution.
9998 Adjust dump information accordingly.
10000 2017-10-12 Bin Cheng <bin.cheng@arm.com>
10002 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
10003 and set PTYPE_SEQUENTIAL for merged partition.
10005 2017-10-12 Richard Biener <rguenther@suse.de>
10007 PR tree-optimization/69728
10009 2017-09-19 Richard Biener <rguenther@suse.de>
10011 PR tree-optimization/69728
10012 * graphite-sese-to-poly.c (schedule_error): New global.
10013 (add_loop_schedule): Handle empty domain by failing the
10015 (build_original_schedule): Handle schedule_error.
10017 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
10018 domain by returning an unchanged schedule.
10020 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10022 * genrecog.c (validate_pattern): For VEC_SELECT verify that
10023 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
10025 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
10027 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
10030 2017-10-12 Jakub Jelinek <jakub@redhat.com>
10033 * expr.c (store_field): Don't optimize away bitsize == 0 store
10034 from CALL_EXPR with addressable return type.
10036 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10038 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
10039 * config/rs6000/rs6000.md (sel): Delete mode attribute.
10040 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
10041 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
10042 TARGET_ISEL instead of TARGET_ISEL<sel>.
10044 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
10046 * config/rs6000/rs6000.c
10047 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
10049 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
10051 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
10053 (reg_or_cint_operand): Fix comment.
10054 (reg_or_zero_operand): New predicate.
10055 * config/rs6000/rs6000-protos.h (output_isel): Delete.
10056 * config/rs6000/rs6000.c (output_isel): Delete.
10057 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
10058 instead of reg_or_cint_operand. Output instruction directly (not via
10060 (isel_unsigned_<mode>): Ditto.
10061 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
10062 gpc_reg_operand. Add an instruction alternative for this. Output
10063 instruction directly.
10064 (*isel_reversed_unsigned_<mode>): Ditto.
10066 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10068 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
10069 (TARGET_CANONICALIZE_COMPARISON): Define.
10071 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
10074 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10075 Check whether the dest is REG before adding REG_EQUIV note.
10077 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10080 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10082 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10083 (make_hard_regno_born, make_hard_regno_dead): Update
10084 bb_killed_pseudos and bb_gen_pseudos.
10086 2017-10-11 Nathan Sidwell <nathan@acm.org>
10088 * incpath.h (enum incpath_kind): Name enum, prefix values.
10089 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
10090 * incpath.c (heads, tails): Use INC_MAX.
10091 (add_env_var_paths, add_standard_paths): Use incpath_kind.
10092 (merge_include_chains, split_quote_chain,
10093 register_include_chains): Update incpath_kind names.
10094 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
10095 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
10097 (add_framework_path, darwin_register_objc_includes): Likewise.
10098 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
10100 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10102 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
10103 Do not use float_operator operator predicate.
10104 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
10105 * config/i386/predicates.md (float_operator): Remove predicate.
10107 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10109 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
10110 (*jccxf_i387): Ditto.
10111 (*jcc<mode>_i387): Ditto.
10112 (*jccu<mode>_i387): Ditto.
10113 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
10114 (*jcc_*_i387 splitters): Remove.
10115 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
10116 * config/i386/i386.c (ix86_split_fp_branch): Remove.
10117 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
10120 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10122 * profile-count.h (slow_safe_scale_64bit): New function.
10123 (safe_scale_64bit): New inline.
10124 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
10125 * profile-count.c: Include wide-int.h
10126 (slow_safe_scale_64bit): New.
10128 2017-10-11 Nathan Sidwell <nathan@acm.org>
10130 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
10131 HAS_DECL_ASSEMBLER_NAME_P.
10132 * gimple-expr.c (gimple_decl_printable_name: Check
10133 HAS_DECL_ASSEMBLER_NAME_P too.
10134 * ipa-utils.h (type_in_anonymous_namespace_p): Check
10135 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
10136 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
10137 * passes.c (rest_of_decl_compilation): Check
10138 HAS_DECL_ASSEMBLER_NAME_P too.
10139 * recog.c (verify_changes): Likewise.
10140 * tree-pretty-print.c (dump_decl_name): Likewise.
10141 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
10143 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
10144 (SET_DECL_ASSEMBLER_NAME): Use it.
10145 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
10146 (COPY_DECL_ASSEMBLER_NAME): Likewise.
10147 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
10149 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10151 * config.gcc (i386, x86_64): Add extra objects.
10152 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
10153 (ix86_min_insn_size): Declare.
10154 (ix86_issue_rate): Declare.
10155 (ix86_adjust_cost): Declare.
10156 (ia32_multipass_dfa_lookahead): Declare.
10157 (ix86_macro_fusion_p): Declare.
10158 (ix86_macro_fusion_pair_p): Declare.
10159 (ix86_bd_has_dispatch): Declare.
10160 (ix86_bd_do_dispatch): Declare.
10161 (ix86_core2i7_init_hooks): Declare.
10162 (ix86_atom_sched_reorder): Declare.
10163 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
10164 (COSTS_N_BYTES): Move to x86-tune-costs.h.
10165 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
10166 (rip_relative_addr_p): Rename to ...
10167 (ix86_rip_relative_addr_p): ... this one; export.
10168 (memory_address_length): Update.
10169 (ix86_issue_rate): Move to x86-tune-sched.c.
10170 (ix86_flags_dependent): Move to x86-tune-sched.c.
10171 (ix86_agi_dependent): Move to x86-tune-sched.c.
10172 (exact_dependency_1): Move to x86-tune-sched.c.
10173 (exact_store_load_dependency): Move to x86-tune-sched.c.
10174 (ix86_adjust_cost): Move to x86-tune-sched.c.
10175 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
10176 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
10177 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
10178 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
10179 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
10180 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
10181 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
10182 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
10183 (min_insn_size): Rename to ...
10184 (ix86_min_insn_size): ... this one; export.
10185 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
10186 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
10187 (core2i7_first_cycle_multipass_backtrack): Move to
10188 x86-tune-sched-core.c.
10189 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
10190 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
10191 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
10192 (ix86_avoid_jump_mispredicts): Update.
10193 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
10194 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
10195 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
10196 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
10197 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
10198 (MAX_INSN): Move to ix86-tune-sched-bd.c.
10199 (MAX_IMM): Move to ix86-tune-sched-bd.c.
10200 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
10201 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
10202 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
10203 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
10204 (MAX_STORE): Move to ix86-tune-sched-bd.c.
10205 (BIG): Move to ix86-tune-sched-bd.c.
10206 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
10207 (enum insn_path): Move to ix86-tune-sched-bd.c.
10208 (get_mem_group): Move to ix86-tune-sched-bd.c.
10209 (is_cmp): Move to ix86-tune-sched-bd.c.
10210 (dispatch_violation): Move to ix86-tune-sched-bd.c.
10211 (is_branch): Move to ix86-tune-sched-bd.c.
10212 (is_prefetch): Move to ix86-tune-sched-bd.c.
10213 (init_window): Move to ix86-tune-sched-bd.c.
10214 (allocate_window): Move to ix86-tune-sched-bd.c.
10215 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
10216 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
10217 (process_end_window): Move to ix86-tune-sched-bd.c.
10218 (allocate_next_window): Move to ix86-tune-sched-bd.c.
10219 (find_constant): Move to ix86-tune-sched-bd.c.
10220 (get_num_immediates): Move to ix86-tune-sched-bd.c.
10221 (has_immediate): Move to ix86-tune-sched-bd.c.
10222 (get_insn_path): Move to ix86-tune-sched-bd.c.
10223 (get_insn_group): Move to ix86-tune-sched-bd.c.
10224 (count_num_restricted): Move to ix86-tune-sched-bd.c.
10225 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
10226 (add_insn_window): Move to ix86-tune-sched-bd.c.
10227 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
10228 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
10229 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
10230 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
10231 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
10232 (do_dispatch): Move to ix86-tune-sched-bd.c.
10233 (has_dispatch): Move to ix86-tune-sched-bd.c.
10234 * config/i386/t-i386: Add new object files.
10235 * config/i386/x86-tune-costs.h: New file.
10236 * config/i386/x86-tune-sched-atom.c: New file.
10237 * config/i386/x86-tune-sched-bd.c: New file.
10238 * config/i386/x86-tune-sched-core.c: New file.
10239 * config/i386/x86-tune-sched.c: New file.
10241 2017-10-11 Liu Hao <lh_mouse@126.com>
10243 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
10244 the generic version below instead.
10245 (should_colorize): Recognize Windows consoles as terminals
10247 * pretty-print.c [__MINGW32__] (write_all): New function.
10248 [__MINGW32__] (find_esc_head): Likewise.
10249 [__MINGW32__] (find_esc_terminator): Likewise.
10250 [__MINGW32__] (eat_esc_sequence): Likewise.
10251 [__MINGW32__] (mingw_ansi_fputs): New function that handles
10253 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
10256 2017-10-11 Richard Biener <rguenther@suse.de>
10258 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
10259 Properly call analyze_scalar_evolution with the loop of the stmt.
10261 2017-10-11 Richard Biener <rguenther@suse.de>
10263 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
10264 * tree-core.h (tree_base): Add chrec_var union member.
10265 * tree.h (CHREC_VAR): Remove.
10266 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
10267 * tree-chrec.h (build_polynomial_chrec): Adjust.
10268 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
10269 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
10271 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
10273 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
10274 * match.pd: ... here.
10275 ((T) X == (T) Y): Relax condition.
10277 2017-10-11 Bin Cheng <bin.cheng@arm.com>
10279 PR tree-optimization/82472
10280 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
10282 (break_alias_scc_partitions): Update postorder number.
10284 2017-10-11 Martin Liska <mliska@suse.cz>
10287 * opts.c (parse_no_sanitize_attribute): Do not use error_value
10289 * opts.h (parse_no_sanitize_attribute): Remove last argument.
10291 2017-10-11 Martin Liska <mliska@suse.cz>
10293 * print-rtl.c (print_insn): Move declaration of idbuf
10294 to same scope as name.
10296 2017-10-11 Martin Liska <mliska@suse.cz>
10301 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10302 volatile arguments.
10304 2017-10-11 Martin Liska <mliska@suse.cz>
10307 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10308 volatile arguments.
10310 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
10312 * config.gcc (default_gnu_indirect_function): Default to yes for
10313 arm*-*-linux* with glibc.
10315 2017-10-11 Richard Biener <rguenther@suse.de>
10317 * tree-scalar-evolution.c (get_scalar_evolution): Handle
10318 default-defs and types we do not want to analyze.
10319 (interpret_loop_phi): Replace unreachable code with an assert.
10320 (compute_scalar_evolution_in_loop): Remove and inline ...
10321 (analyze_scalar_evolution_1): ... here, replacing condition with
10322 what makes the intent clearer. Remove handling of cases
10323 get_scalar_evolution now handles.
10325 2017-10-10 Jim Wilson <wilson@tuliptree.org>
10327 PR rtl-optimization/81434
10328 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
10329 comment for main loop. In sched_group_found if, also add checks for
10330 pass and min_cost_group.
10332 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
10334 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
10335 (rs6000_insn_cost): New function.
10336 * config/rs6000/rs6000.md (cost): New attribute.
10338 2017-10-10 Jakub Jelinek <jakub@redhat.com>
10339 H.J. Lu <hongjiu.lu@intel.com>
10343 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
10344 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
10345 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
10346 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
10347 to OPTION_MASK_ISA_AVX512VL - builtins that have both
10348 OPTION_MASK_ISA_MMX and some other bit set require both
10349 mmx and the ISAs without the mmx bit.
10350 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
10351 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
10352 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
10353 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
10354 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
10355 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
10356 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
10357 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
10358 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
10359 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
10360 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
10361 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
10362 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
10363 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
10364 Add OPTION_MASK_ISA_MMX.
10366 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
10368 * config.gcc (armv7*-*-freebsd*): New target.
10369 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
10371 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
10373 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
10374 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
10377 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
10379 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
10380 a template parameter.
10381 (WIDE_INT_REF_FOR): Update accordingly.
10382 * tree.h (wi::int_traits <const_tree>): Delete.
10383 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
10384 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
10385 (wi::tree_to_wide_ref): New typedef.
10386 (wi::to_wide): New function.
10387 * calls.c (get_size_range): Use wi::to_wide when operating on
10388 trees as wide_ints.
10389 * cgraph.c (cgraph_node::create_thunk): Likewise.
10390 * config/i386/i386.c (ix86_data_alignment): Likewise.
10391 (ix86_local_alignment): Likewise.
10392 * dbxout.c (stabstr_O): Likewise.
10393 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
10394 * expr.c (const_vector_from_tree): Likewise.
10395 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
10396 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
10397 (fold_negate_expr_1, int_const_binop_1, const_binop)
10398 (fold_convert_const_int_from_real, optimize_bit_field_compare)
10399 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
10400 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
10401 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
10402 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
10403 (fold_not_const, round_up_loc): Likewise.
10404 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
10405 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
10406 (alloca_call_type): Likewise.
10407 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
10408 * godump.c (go_output_typedef): Likewise.
10409 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
10410 * internal-fn.c (get_min_precision): Likewise.
10411 * ipa-cp.c (ipcp_store_vr_results): Likewise.
10412 * ipa-polymorphic-call.c
10413 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
10414 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
10415 (ipa_modify_call_arguments): Likewise.
10416 * match.pd: Likewise.
10417 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
10418 * print-tree.c (print_node_brief, print_node): Likewise.
10419 * stmt.c (expand_case): Likewise.
10420 * stor-layout.c (layout_type): Likewise.
10421 * tree-affine.c (tree_to_aff_combination): Likewise.
10422 * tree-cfg.c (group_case_labels_stmt): Likewise.
10423 * tree-data-ref.c (dr_analyze_indices): Likewise.
10424 (prune_runtime_alias_test_list): Likewise.
10425 * tree-dump.c (dequeue_and_dump): Likewise.
10426 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
10427 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
10428 * tree-pretty-print.c (dump_generic_node): Likewise.
10429 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
10430 (simple_iv_with_niters): Likewise.
10431 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
10432 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
10433 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
10434 * tree-ssa-loop-niter.c (split_to_var_and_offset)
10435 (refine_value_range_using_guard, number_of_iterations_ne_max)
10436 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
10437 (get_cst_init_from_scev, record_nonwrapping_iv)
10438 (scev_var_range_cant_overflow): Likewise.
10439 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
10440 * tree-ssa-pre.c (compute_avail): Likewise.
10441 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
10442 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
10443 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
10444 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
10445 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
10446 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
10447 (dump_case_nodes, try_switch_expansion): Likewise.
10448 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
10449 (vect_do_peeling): Likewise.
10450 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
10451 * tree-vect-stmts.c (vectorizable_load): Likewise.
10452 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
10453 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
10454 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
10455 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
10456 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
10457 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
10458 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
10459 (evrp_dom_walker::before_dom_children): Likewise.
10460 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
10461 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
10462 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
10463 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
10464 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
10465 (get_range_pos_neg): Likewise.
10466 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
10467 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
10468 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
10469 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
10470 * config/avr/avr.c (avr_fold_builtin): Likewise.
10471 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
10472 * config/msp430/msp430.c (msp430_attr): Likewise.
10473 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
10474 * config/powerpcspe/powerpcspe-c.c
10475 (altivec_resolve_overloaded_builtin): Likewise.
10476 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
10477 (rs6000_expand_ternop_builtin): Likewise.
10478 * config/rs6000/rs6000-c.c
10479 (altivec_resolve_overloaded_builtin): Likewise.
10480 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
10481 (rs6000_expand_ternop_builtin): Likewise.
10482 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
10484 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10486 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
10487 when copying loop nest with only one inner loop.
10489 2017-10-10 Richard Biener <rguenther@suse.de>
10491 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
10492 blocks if SCEV is active.
10493 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
10495 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
10496 (scev_initialize): Assert we are not yet initialized.
10498 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10500 * tree-loop-distribution.c (generate_loops_for_partition): Remove
10501 inner loop's exit stmt by making it always exit the loop, otherwise
10502 we would generate an infinite empty loop.
10504 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10506 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
10507 renaming variables in new preheader if it's deleted.
10509 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10511 * tree-loop-distribution.c (struct partition): Remove unused field
10512 loops of the structure.
10513 (partition_alloc, partition_free): Ditto.
10514 (build_rdg_partition_for_vertex): Ditto.
10516 2017-10-09 Jeff Law <law@redhat.com>
10518 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
10519 return type to match prototype and documentation.
10521 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10523 * config/rs6000/rs6000.c (processor_costs): Move to ...
10524 * config/rs6000/rs6000.h: ... here.
10525 (rs6000_cost): Declare.
10527 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
10529 * except.c (setjmp_fn): New global variable.
10530 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
10531 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
10532 if DONT_USE_BUILTIN_SETJMP is defined.
10534 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10536 * target.def (insn_cost): New hook.
10537 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
10538 * doc/tm.texi: Regenerate.
10539 * rtlanal.c (insn_cost): Use the new hook.
10541 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10543 * combine.c (combine_validate_cost): Compute the new insn_cost,
10544 not just pattern_cost.
10545 (try_combine): Adjust comment.
10547 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10549 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
10551 * combine.c (uid_insn_cost): Adjust comment.
10552 (combine_validate_cost): Adjust comment. Use pattern_cost instead
10554 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
10555 * dse.c (find_shift_sequence): Ditto.
10556 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
10557 (bb_valid_for_noce_process_p): Use pattern_cost.
10558 * rtl.h (insn_rtx_cost): Delete.
10559 (pattern_cost): New prototype.
10560 (insn_cost): New prototype.
10561 * rtlanal.c (insn_rtx_cost): Rename to...
10562 (pattern_cost): ... this.
10565 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
10567 * config/i386/i386.md (*jcc_2): Remove insn pattern.
10568 (*jcc<mode>_0_r_i387): Ditto.
10569 (*jccxf_r_i387): Ditto.
10570 (*jcc<mode>_r_i387): Ditto.
10571 (*jccu<mode>_r_i387): Ditto.
10572 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
10573 (*jcc): Rename from *jcc_1.
10575 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10577 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
10578 deferred rescans after the lvx/stvx recombination pre-pass.
10580 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
10582 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
10583 memory operation instruction support.
10584 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
10585 (rs6000-ibm-aix[789]*): Likewise.
10586 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
10587 Document new functions.
10589 2017-10-09 Richard Biener <rguenther@suse.de>
10591 PR tree-optimization/82397
10592 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
10593 equality only for semantically equal trees.
10595 2017-10-09 Richard Biener <rguenther@suse.de>
10597 PR tree-optimization/82449
10598 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
10599 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
10600 allow constant addresses.
10601 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
10604 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10606 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
10609 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10612 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
10615 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10618 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
10620 2017-10-09 Jakub Jelinek <jakub@redhat.com>
10623 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
10624 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
10626 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
10628 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
10629 (WI_BINARY_PREDICATE_RESULT): Likewise.
10630 (wi::binary_traits::operator_result): New type.
10631 (wi::binary_traits::predicate_result): Likewise.
10632 (generic_wide_int::operator~, unary generic_wide_int::operator-)
10633 (generic_wide_int::operator==, generic_wide_int::operator!=)
10634 (generic_wide_int::operator&, generic_wide_int::and_not)
10635 (generic_wide_int::operator|, generic_wide_int::or_not)
10636 (generic_wide_int::operator^, generic_wide_int::operator+
10637 (binary generic_wide_int::operator-, generic_wide_int::operator*):
10639 (operator~, unary operator-, operator==, operator!=, operator&)
10640 (operator|, operator^, operator+, binary operator-, operator*): New
10642 * expr.c (get_inner_reference): Use wi::bit_and_not.
10643 * fold-const.c (fold_binary_loc): Likewise.
10644 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
10645 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
10646 (bit_value_binop): Likewise.
10647 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
10648 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
10649 (extract_range_from_binary_expr_1): Likewise.
10650 (masked_increment): Likewise.
10651 (simplify_bit_ops_using_ranges): Likewise.
10653 2017-10-09 Martin Jambor <mjambor@suse.cz>
10656 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
10657 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
10658 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
10660 (hsa_fixup_mov_insn_type): New function.
10661 (hsa_op_with_type::get_in_type): Use it.
10662 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
10663 immediates in an assert.
10664 (hsa_op_with_type::extend_int_to_32bit): New method.
10665 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
10666 types. Convert to dest type if necessary.
10667 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
10668 (reg_for_gimple_ssa): Pass false as min32int to
10669 hsa_type_for_scalar_tree_type.
10670 (gen_hsa_addr): Fixup type when creating addresable temporary.
10671 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
10672 (gen_hsa_unary_operation): Extend operands and convert to dest type if
10673 necessary. Call hsa_fixup_mov_insn_type.
10674 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
10675 extend operands and convert to dest type if necessary.
10676 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
10677 to dest type if necessary.
10678 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
10679 if conversion nt necessary and size matches.
10680 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
10681 to dest type if necessary.
10682 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
10683 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
10685 (gen_hsa_clrsb): Likewise.
10686 (gen_hsa_ffs): Likewise.
10687 (gen_hsa_divmod): Extend operands and convert to dest type if
10689 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
10691 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10693 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
10694 Remove empty default arguments. Use a brace block as output
10696 (conditional return): Ditto.
10698 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
10699 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
10701 (group_ending_nop): Ditto.
10702 (doloop_end): Ditto.
10703 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
10704 (splitters for those): Ditto.
10706 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10708 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
10709 a conditional jump (and the compare for it) so that pc_rtx is the
10711 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
10712 for the deleted and renamed ctr<mode>_internal[234] patterns.
10713 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
10714 Delete second conditional return pattern.
10715 (ctr<mode>_internal2): Delete this second bdnz pattern.
10716 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
10717 (ctr<mode>_internal4): Delete this second bdz pattern.
10719 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10721 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
10722 (always_initialized_rtx_for_ssa_name_p): New predicate.
10723 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
10724 (finish_out_of_ssa): Free new field of SA.
10725 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
10726 * tree-ssa-coalesce.c: Include tree-ssa.h.
10727 (get_parm_default_def_partitions): Remove extern keyword.
10728 (get_undefined_value_partitions): New function.
10729 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
10730 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
10731 uninitialized bits.
10732 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
10734 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10736 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
10738 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10740 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
10741 for targets that preffer 128bit.
10743 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10745 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
10747 2017-10-08 Olivier Hainque <hainque@adacore.com>
10749 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
10750 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
10752 (thumb_set_return_address): Likewise.
10754 2017-10-08 Olivier Hainque <hainque@adacore.com>
10756 * common/config/arm/arm-common.c (arm_except_unwind_info):
10757 Handle DWARF2_UNWIND_INFO.
10759 2017-10-07 Michael Collison <michael.collison@arm.com>
10761 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
10764 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
10766 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
10767 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
10768 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
10769 defined, force the creation of a new block for a dispatch label.
10771 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
10773 * invoke.texi (Wsuggest-attribute=cold): Document.
10774 * common.opt (Wsuggest-attribute=cold): New
10775 * ipa-pure-const.c (warn_function_cold): New function.
10776 * predict.c (compute_function_frequency): Use it.
10777 * predict.h (warn_function_cold): Declare.
10779 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
10781 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
10784 2017-10-06 Martin Liska <mliska@suse.cz>
10786 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
10787 keyword for member functions.
10788 (struct sanopt_tree_couple): New struct.
10789 (struct sanopt_tree_couple_hash): New function.
10790 (struct sanopt_ctx): Add new hash_map.
10791 (has_dominating_ubsan_ptr_check): New function.
10792 (record_ubsan_ptr_check_stmt): Likewise.
10793 (maybe_optimize_ubsan_ptr_ifn): Likewise.
10794 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
10795 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
10797 2017-10-06 Sudakshina Das <sudi.das@arm.com>
10800 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
10801 aarch64_simd_valid_immediate on CONST_VECTORs.
10802 (aarch64_reg_or_bic_imm): Likewise.
10804 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
10806 PR rtl-optimization/82396
10807 * haifa-sched.c (ready_sort_real): Disable qsort checking.
10809 2017-10-06 Sebastian Pop <sebpop@gmail.com>
10811 * graphite-dependences.c (scop_get_reads): Move code to...
10812 (scop_get_must_writes): Move code to...
10813 (scop_get_may_writes): Move code to...
10814 (scop_get_reads_and_writes): ... here.
10815 (scop_get_dependences): Call scop_get_reads_and_writes.
10817 2017-10-06 Jakub Jelinek <jakub@redhat.com>
10819 PR tree-optimization/82434
10820 * fold-const.h (can_native_encode_type_p,
10821 can_native_encode_string_p): Remove.
10822 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
10823 don't encode anything, just return what would be otherwise returned.
10824 (native_encode_fixed, native_encode_complex, native_encode_vector):
10826 (native_encode_string): Likewise. Inline by hand
10827 can_native_encode_string_p.
10828 (can_native_encode_type_p): Remove.
10829 (can_native_encode_string_p): Remove.
10830 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
10831 STRING_CSTs using can_native_encode_string_p, test all
10832 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
10833 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
10834 argument from native_encode_expr.
10835 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
10836 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
10839 2017-10-06 Richard Biener <rguenther@suse.de>
10841 PR tree-optimization/82397
10842 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
10843 operand_equal_p but rely on data_ref_compare_tree for detecting
10845 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
10846 to match up with dr_group_sort_cmp.
10848 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10851 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
10853 * config/s390/s390-builtin-types.def: Regenerate.
10855 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10858 * config/s390/s390-builtin-types.def: Regenerate.
10859 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
10860 Change flag from B_VXE to B_VX.
10861 (s390_vec_min_dbl): Remove B_VXE flag.
10863 2017-10-06 Richard Biener <rguenther@suse.de>
10865 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
10866 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
10867 translate_isl_ast_to_gimple::is_valid_rename,
10868 translate_isl_ast_to_gimple::get_rename,
10869 translate_isl_ast_to_gimple::get_def_bb_for_const,
10870 translate_isl_ast_to_gimple::get_new_name,
10871 translate_isl_ast_to_gimple::collect_all_ssa_names,
10872 translate_isl_ast_to_gimple::copy_loop_phi_args,
10873 translate_isl_ast_to_gimple::collect_all_ssa_names,
10874 translate_isl_ast_to_gimple::copy_loop_phi_args,
10875 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
10876 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
10877 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
10878 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
10879 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
10880 translate_isl_ast_to_gimple::copy_cond_phi_args,
10881 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
10882 translate_isl_ast_to_gimple::edge_for_new_close_phis,
10883 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
10884 translate_isl_ast_to_gimple::rename_uses,
10885 translate_isl_ast_to_gimple::rename_all_uses): Remove.
10886 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
10887 (set_rename_for_each_def): Likewise.
10888 (graphite_copy_stmts_from_block): Handle debug stmt resetting
10889 here. Handle rewriting SCEV analyzable uses here.
10890 (copy_bb_and_scalar_dependences): Generate code for PHI
10892 (graphite_regenerate_ast_isl): Adjust.
10893 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
10894 (add_write, add_read): New functions.
10895 (build_cross_bb_scalars_def): Use it and simplify.
10896 (build_cross_bb_scalars_use): Likewise.
10897 (graphite_find_cross_bb_scalar_vars): Inline into...
10898 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
10899 simulating out-of-SSA. Compute liveout and add dependencies.
10900 (build_scops): Force an empty entry block.
10901 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
10903 (sese_build_liveouts): Declare.
10904 (sese_trivially_empty_bb_p): Likewise.
10905 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
10906 compute liveout and debug_liveout.
10907 (sese_bad_liveouts_use): Remove.
10908 (sese_reset_debug_liveouts_bb): Likewise.
10909 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
10910 (sese_build_liveouts): Build liveout and debug_liveout and store
10912 (new_sese_info): Adjust.
10913 (free_sese_info): Likewise.
10914 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
10915 do not build liveout here.
10916 (move_sese_in_condition): Adjust region entry.
10917 (scev_analyzable_p): Match up with chrec_apply requirements.
10918 (sese_trivially_empty_bb_p): New.
10919 * tree-into-ssa.c (get_reaching_def): Properly support generating
10920 default-defs for incremental rewrite of anonymous names.
10922 2017-10-06 Richard Biener <rguenther@suse.de>
10924 * graphite-sese-to-poly.c (extract_affine): For casts increasing
10925 precision do not perform modulo reduction.
10927 2017-10-06 Richard Biener <rguenther@suse.de>
10929 PR tree-optimization/82436
10930 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10931 conservatively choose the vectorization factor when checking
10932 whether we can perform the required load permutation.
10933 (vect_transform_slp_perm_load): Assert when we may not fail.
10935 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10937 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10938 message for incompatible -msdata=* and -mcall-* options.
10940 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10942 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10943 rate for post-reload scheduling.
10945 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10947 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10949 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10951 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10952 to improve monte carlo in scimark.
10954 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10956 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10957 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10958 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10959 pentium4_cost, nocona_cost): Set reassociation width to 1.
10960 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10961 width to 2 for fp operations and 1 otherwise.
10962 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10964 (atom_cost): Set reassociation width to 2.
10965 (slm_cost, generic_cost): Set fp reassociation width
10966 to 2 and 1 otherwise.
10967 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10968 (core_cost): Set fp reassociation width to 4 and vector to 2.
10969 (ix86_reassociation_width): Rewrite using cost table; special case
10970 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10971 and TARGET_AVX128_OPTIMAL.
10972 * config/i386/i386.h (processor_costs): Add
10973 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10974 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10975 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10976 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10977 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10978 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10980 2017-10-05 Nathan Sidwell <nathan@acm.org>
10982 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10984 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10986 * config/arm/arm.c (arm_test_fpu_data): New.
10987 (arm_run_selftests): Call arm_test_fpu_data.
10989 2017-10-04 Nathan Sidwell <nathan@acm.org>
10991 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10993 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10995 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10997 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10998 check type for aarch64_simd_valid_immediate.
10999 (aarch64_output_simd_mov_immediate): Update prototype.
11000 (aarch64_simd_valid_immediate): Update prototype.
11001 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
11002 support for ORR-immediate.
11003 (and<mode>3): modified pattern to add support for BIC-immediate.
11004 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
11005 now checks for valid immediate for BIC and ORR based on new enum
11007 (aarch64_output_simd_mov_immediate): Function now used to output
11008 BIC/ORR imm as well based on new enum argument.
11009 * config/aarch64/constraints.md (Do): New vector immediate constraint.
11011 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
11012 (aarch64_reg_or_bic_imm): Likewise.
11014 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11016 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
11017 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
11019 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11023 PR rtl-optimization/82396
11024 * haifa-sched.c (autopref_multipass_init): Simplify
11026 (autopref_rank_data): Simplify sort order.
11027 * sched-int.h (autopref_multipass_data_): Remove
11028 multi_mem_insn_p, min_offset and max_offset.
11030 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11032 * doc/sourcebuild.texi: Document vect_peeling_profitable.
11034 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11036 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
11037 vect_doubleint_cvt.
11039 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11041 * doc/sourcebuild.texi: Document vect_long_mult.
11043 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
11045 PR tree-optimization/82413
11046 * fold-const.c (build_range_check): Use widest_int when comparing
11047 the maximum ETYPE value with HIGH.
11049 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
11051 PR rtl-optimization/82396
11052 * haifa-sched.c (autopref_multipass_init): Simplify
11054 (autopref_rank_data): Simplify sort order.
11055 * sched-int.h (autopref_multipass_data_): Remove
11056 multi_mem_insn_p, min_offset and max_offset.
11058 2017-10-04 Jakub Jelinek <jakub@redhat.com>
11060 PR tree-optimization/82381
11061 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
11062 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
11065 PR tree-optimization/82374
11066 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
11067 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
11068 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
11069 current_function_decl to the new decl.
11071 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
11073 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
11074 helper macro for IEEE float128 hardware built-in functions.
11075 (SQRTF128_ODD): Add built-in functions with the round-to-odd
11077 (TRUNCF128_ODD): Likewise.
11078 (ADDF128_ODD): Likewise.
11079 (SUBF128_ODD): Likewise.
11080 (MULF128_ODD): Likewise.
11081 (DIVF128_ODD): Likewise.
11082 (FMAF128_ODD): Likewise.
11083 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
11084 UNSPEC_TRUNC_ROUND_TO_ODD.
11085 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
11086 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
11087 floating point round to odd instructions.
11088 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
11089 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
11090 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
11091 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
11092 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
11093 (trunc<mode>sf2_hw): Change the truncate with round to odd
11094 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
11095 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
11096 to odd hardware instructions.
11097 (sub<mode>3_odd): Likewise.
11098 (mul<mode>3_odd): Likewise.
11099 (div<mode>3_odd): Likewise.
11100 (sqrt<mode>2_odd): Likewise.
11101 (fma<mode>4_odd): Likewise.
11102 (fms<mode>4_odd): Likewise.
11103 (nfma<mode>4_odd): Likewise.
11104 (nfms<mode>4_odd): Likewise.
11105 (trunc<mode>df2_odd): Change the truncate with round to odd
11106 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
11108 * doc/extend.texi (PowerPC built-in functions): Update documentation
11109 for existing IEEE float128-bit built-in functions. Add built-in
11110 functions that generate the IEEE 128-bit floating point round to
11113 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
11115 PR rtl-optimization/77729
11116 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
11117 to (X&(C1&~C2))|C2 transformations.
11119 2017-10-03 Martin Jambor <mjambor@suse.cz>
11121 PR tree-optimization/82363
11122 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
11123 mismatch, mark lacc written regardless of racc.
11125 2017-10-03 Jakub Jelinek <jakub@redhat.com>
11127 PR tree-optimization/82381
11128 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
11129 stmt_to_insert nor wheather SSA_NAMEs are default defs.
11130 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
11131 fallthrough into reassoc_stmt_dominates_stmt_p.
11134 * combine.c (combine_instructions): Don't combine in unreachable
11137 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
11140 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
11141 function to not use the have_cpu variable. Do not set cpu_index,
11142 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
11143 or the default cpu.
11144 (rs6000_valid_attribute_p): Remove duplicate initializations of
11145 old_optimize and func_optimize.
11146 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
11147 (rs6000_activate_target_options): Make global.
11148 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
11151 2017-10-02 Jakub Jelinek <jakub@redhat.com>
11153 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
11154 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
11155 Set *poffset to 0 and *psize and *pmax_size to -1 if
11156 *poffset + *psize overflows in HOST_WIDE_INT.
11158 PR tree-optimization/82387
11159 PR tree-optimization/82388
11160 PR tree-optimization/82389
11161 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
11162 instead of live_bytes non-NULL.
11164 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
11167 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
11170 2017-10-02 Richard Biener <rguenther@suse.de>
11172 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
11173 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
11174 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
11176 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11178 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
11179 requested precision matches the type's.
11180 * calls.c (alloc_max_size): Calculate the new candidate size as
11181 a widest_int and use wi::to_widest when comparing it with the
11182 current candidate size.
11183 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
11184 zero rather than integer_zero_node.
11185 * match.pd: Check for a no-op conversion before using wi::add
11186 rather than after. Use tree_to_uhwi when summing small shift
11187 counts into an unsigned int.
11189 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11190 Alan Hayward <alan.hayward@arm.com>
11191 David Sherwood <david.sherwood@arm.com>
11194 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
11195 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
11196 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
11197 POINTER_AND_FP_REGS.
11199 2017-10-02 Richard Biener <rguenther@suse.de>
11201 PR tree-optimization/82355
11202 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11203 a mapping for the enclosing loop but avoid generating one for
11204 the loop tree root.
11205 (copy_bb_and_scalar_dependences): Remove premature codegen
11206 error on PHIs in blocks duplicated into multiple places.
11207 * graphite-scop-detection.c
11208 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11209 in the region use it as loop and nest to analyze the DR in.
11210 (try_generate_gimple_bb): Likewise.
11211 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11212 (add_loop_constraints): For blocks in a loop not in the region
11213 create a dimension with a single iteration.
11214 * sese.h (gbb_loop_at_index): Remove assert.
11216 2017-10-01 Kevin Buettner <kevinb@redhat.com>
11218 * omp-expand.c (adjust_context_scope): New function.
11219 (expand_parallel_call): Call adjust_context_scope.
11221 2017-10-01 Jeff Law <law@redhat.com>
11223 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
11224 dom_opt_dom_walker class with direct access to private members.
11225 Add comments. Call test_for_singularity.
11226 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
11227 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
11228 m_dummy_cond anymore.
11229 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
11231 (pass_dominator:execute): Build the dummy_cond here and pass it
11232 to the dom_opt_dom_walker ctor.
11233 (test_for_singularity): New function.
11235 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
11236 Maya Rashish <coypu@sdf.org>
11238 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
11239 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
11240 (alpha*-*-netbsd*) Use nbsd_tm_file.
11241 (arm*-*-netbsdelf*) Likewise.
11242 (i[34567]86-*-netbsdelf*) Likewise.
11243 (x86_64-*-netbsd*) Likewise.
11244 (mips*-*-netbsd*) Likewise.
11245 (powerpc-*-netbsd*) Likewise.
11246 (sh*-*-netbsd*) Likewise.
11247 (sparc-*-netbsdelf*) Likewise.
11248 (sparc64-*-netbsd*) Likewise.
11249 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
11251 (vax-*-netbsdelf*) Likewise.
11252 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
11253 (UINT_FAST8_TYPE) Likewise.
11254 (INT_FAST16_TYPE) Check CHAR_FAST16.
11255 (UINT_FAST16_TYPE) Likewise.
11257 2017-09-30 Jakub Jelinek <jakub@redhat.com>
11260 * config/i386/i386.md
11261 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
11262 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
11263 *divmodsi4_zext_2): New define_insn_and_split.
11264 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
11265 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
11266 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
11267 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
11268 New define_insn_and_split.
11269 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
11270 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
11271 operands[1] having DImode when mode is SImode.
11273 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
11274 always SImode for DIV and MOD in REG_EQUAL notes.
11276 2017-09-29 Yury Gribov <tetra2005@gmail.com>
11278 PR middle-end/82319
11279 * match.pd: Fix handling of NaNs in pattern.
11281 2017-09-29 Jeff Law <law@redhat.com>
11283 * sbitmap.c (bitmap_bit_in_range_p): New function.
11284 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
11285 * tree-ssa-dse.c (live_bytes_read): New function.
11286 (dse_classify_store): Ignore reads of dead bytes.
11288 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
11289 typos and whitespace errors.
11290 * config/i386/predicates.md (address_no_seg_operand): Likewise.
11291 * config/s390/s390.c (s390_emit_prologue): Likewise.
11293 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11296 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
11297 with a symbol for LRA.
11299 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11301 PR rtl-optimization/82338
11302 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
11304 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
11306 * genmodes.c (calc_wider_mode): Suppress qsort macro.
11307 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
11308 (qsort_chk): Declare.
11309 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
11310 (qsort_chk): New function.
11312 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11314 PR tree-optimization/82337
11315 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
11316 phi definition if the PHI result appears in an abnormal PHI.
11317 (find_basis_for_base_expr): Don't record a basis if the LHS of the
11318 basis appears in an abnormal PHI.
11320 2017-09-29 Richard Biener <rguenther@suse.de>
11322 * graphite-isl-ast-to-gimple.c
11323 (translate_isl_ast_to_gimple::set_codegen_error): New function.
11324 (binary_op_to_tree): Use it.
11325 (get_rename_from_scev): Likewise.
11326 (copy_loop_phi_nodes): Likewise.
11327 (copy_bb_and_scalar_dependences): Likewise.
11328 (translate_pending_phi_nodes): Likewise.
11330 2017-09-29 Jakub Jelinek <jakub@redhat.com>
11333 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
11334 for movabsq $(i32 << shift), r64.
11336 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
11338 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
11339 index when encoding %esp as %rsp to avoid 0x67 prefix.
11341 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
11343 * config/i386/i386.md (*movsf_internal, *movdf_internal):
11344 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
11346 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
11348 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
11349 Extensions with more than 16 double VFP registers.
11350 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
11351 to_clear_mask and all code related to it. Replace the remaining
11352 entry by a sbitmap and adapt code accordingly.
11354 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
11356 * brig-builtins.def: Change pure attributes to const.
11358 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11360 * config.gcc (default_gnu_indirect_function): Default to yes for
11361 sparc*-*-linux* with glibc.
11363 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11365 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
11366 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
11367 when creating .init_array and .fini_array sections with priority
11370 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
11373 * config/aarch64/aarch64.c
11374 (aarch64_builtin_support_vector_misalignment): Always return false
11375 when misalignment is unknown.
11377 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11379 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
11380 this function to return false if the definition used by the swap
11381 instruction is artificial, or if the memory address from which the
11382 constant value is loaded is not represented by a base address held
11383 in a register or if the base address register is a frame or stack
11384 pointer. Additionally, return false if the base address of the
11385 loaded constant is a SYMBOL_REF but is not considered to be a
11387 (replace_swapped_load_constant): New function.
11388 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
11389 loaded constant vector with a load of a swapped constant vector.
11391 2017-09-27 Carl Love <cel@us.ibm.com>
11393 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
11394 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
11395 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
11398 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
11400 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
11401 first, always call autopref_rank_data otherwise.
11403 2017-09-27 Richard Biener <rguenther@suse.de>
11405 * graphite-scop-detection.c (find_scop_parameters): Move
11406 loop bound handling ...
11407 (gather_bbs::before_dom_children): ... here, avoiding the need
11408 to build scop_info->loop_nest.
11409 (record_loop_in_sese): Remove.
11410 * sese.h (sese_info_t::loop_nest): Remove.
11411 * sese.c (new_sese_info): Do not allocate loop_nest.
11412 (free_sese_info): Do not free loop_nest.
11414 2017-09-27 Jakub Jelinek <jakub@redhat.com>
11417 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
11418 lhs from calls if the lhs has addressable type.
11420 2017-09-27 Richard Biener <rguenther@suse.de>
11422 * graphite.h (scop::max_alias_set): New member.
11423 * graphite-scop-detection.c: Remove references to non-existing
11424 --param in comments.
11425 (build_alias_sets): Record the maximum alias set used for drs.
11426 (build_scops): Support zero as unlimited for
11427 --param graphite-max-arrays-per-scop.
11428 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
11429 and inline into ...
11430 (build_poly_sr_1): ... here. Compute alias set based on the
11431 maximum alias set used for drs rather than
11432 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
11434 2017-09-27 Richard Biener <rguenther@suse.de>
11436 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
11437 --param loop-block-tile-size=0 to disable tiling.
11439 2017-09-27 Richard Biener <rguenther@suse.de>
11441 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
11442 (graphite-max-nb-scop-params): Document special value zero.
11443 * domwalk.h (dom_walker::STOP): New symbolical constant.
11444 (dom_walker::dom_walker): Add optional parameter for bb to
11446 (dom_walker::~dom_walker): Declare.
11447 (dom_walker::before_dom_children): Document STOP return value.
11448 (dom_walker::m_user_bb_to_rpo): New member.
11449 (dom_walker::m_bb_to_rpo): Likewise.
11450 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
11451 mapping here if not provided by the user.
11452 (dom_walker::~dom_walker): Free bb to RPO mapping if not
11453 provided by the user.
11454 (dom_walker::STOP): Define.
11455 (dom_walker::walk): Do not compute bb to RPO mapping here.
11456 Support STOP return value from before_dom_children to stop
11458 * graphite-optimize-isl.c (optimize_isl): If the schedule
11459 is the same still generate code if -fgraphite-identity
11460 or -floop-parallelize-all are given.
11461 * graphite-scop-detection.c: Include cfganal.h.
11462 (gather_bbs::gather_bbs): Get and pass through bb to RPO
11464 (gather_bbs::before_dom_children): Return STOP for BBs
11466 (build_scops): Compute bb to RPO mapping and pass it to
11467 the domwalk. Treat --param graphite-max-nb-scop-params=0
11468 as not limiting the number of params.
11469 * graphite.c (graphite_initialize): Remove limit on the
11470 number of basic-blocks in a function.
11471 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
11472 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
11473 default value of 10.
11475 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11477 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
11478 Adjust code to eliminate needing to do the shift right 32-bits
11479 operation after XSCVDPSPN.
11481 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11483 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
11484 ((X / Y) != 0 -> X >= Y): Likewise.
11486 2017-09-26 Carl Love <cel@us.ibm.com>
11488 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
11489 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
11490 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
11491 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
11492 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
11493 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
11494 definitions and overloading.
11495 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
11496 statement for P9V_BUILTIN_XST_LEN_R.
11497 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
11498 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
11499 define_expand and define_insn for the instructions and builtins.
11500 * doc/extend.texi: Update the built-in documentation file for the new
11501 built-in functions.
11502 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
11503 define_insn for the instructions
11505 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
11508 * gcc/config/netbsd-protos.h: New file.
11509 * gcc/config/netbsd.c: New file.
11510 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
11511 * gcc/config/t-netbsd: New file.
11512 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
11513 (tmake_file) Add t-netbsd.
11514 (extra_objs) Add netbsd.o.
11516 2017-09-26 Janus Weil <janus@gcc.gnu.org>
11520 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
11522 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11524 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
11525 sign extension from a vector register to a GPR by doing a 32-bit
11526 direct move and then an EXTSW.
11527 (extendsi<mode>2 splitter): Likewise.
11528 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
11529 right or vector extract after doing XSCVDPSPN. Use
11530 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
11532 (movdi_from_sf_zero_ext): Likewise.
11533 (reload_gpr_from_vsxsf): Likewise.
11534 (p8_mfvsrd_4_disf): Delete, no longer used.
11535 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
11536 then needing to move the SFmode to a GPR to use the XSCVDPSP
11537 instruction instead of FRSP and XSCVDPSPN.
11538 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
11539 it is adjacent to the other XSCVSPDP insns.
11540 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
11541 SFmode to be in traditional Altivec registers.
11542 (vsx_xscvdpspn): Eliminate useless alternative constraint.
11543 (vsx_xscvspdpn): Likewise.
11544 (vsx_xscvspdpn_scalar): Likewise.
11546 2017-09-26 Martin Jambor <mjambor@suse.cz>
11548 * tree-sra.c (compare_access_positions): Put integral types first,
11549 stabilize sorting of integral types, remove conditions putting
11550 non-full-precision integers last.
11551 (sort_and_splice_var_accesses): Disable scalarization if a
11552 non-integert would be represented by a non-full-precision integer.
11554 2017-09-26 Joseph Myers <joseph@codesourcery.com>
11556 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
11557 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
11558 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
11559 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
11560 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
11561 conditionals inside the function instead of around it. Call
11562 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
11563 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
11565 2017-09-26 Richard Biener <rguenther@suse.de>
11567 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
11569 (scop_detection::build_scop_breadth): ... this. Removed.
11570 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
11571 (scop_detection::harmful_stmt_in_bb): Likewise.
11572 (scop_detection::graphite_can_represent_stmt): Likewise.
11573 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
11574 (scop_detection::can_represent_loop): Remove recursion, fold in ...
11575 (scop_detection::can_represent_loop_1): ... this. Removed.
11576 (scop_detection::harmful_loop_in_region): Simplify after inlining
11577 the above and remove more quadraticness.
11578 (build_scops): Adjust.
11579 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
11582 2017-09-26 Jakub Jelinek <jakub@redhat.com>
11585 * config/i386/i386.c (ix86_print_operand_address_as): Only test
11586 REGNO (base) == SP_REG if base is a REG.
11588 PR middle-end/35691
11589 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
11590 if it is different SSA_NAME.
11591 (optimize_range_tests_cmp_bitwise): New function.
11592 (optimize_range_tests): Call it.
11594 2017-09-26 Richard Biener <rguenther@suse.de>
11596 PR tree-optimization/82321
11597 * graphite.c (canonicalize_loop_closed_ssa): Properly check
11598 for the def being inside the loop.
11600 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11602 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
11604 * config/s390/s390-builtins.def: Fix constraint on op4.
11606 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11608 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
11609 independent expanders.
11610 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
11611 ("vec_ordered", "vec_unordered"): New expanders.
11613 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11615 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
11618 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11620 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
11621 vec_unpacks_lo_v16qi.
11622 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
11624 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11626 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
11627 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
11628 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
11630 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11632 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
11634 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
11636 ("*vec_slb<mode>"): New insn pattern.
11637 ("vec_shr_<mode>"): New expander.
11638 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
11639 and force the shift count operand to V16QImode.
11640 ("vec_srb<mode>"): Set shift count mode to V16QI.
11642 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11644 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
11645 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
11646 ("vec_widen_smult_hi_<mode>"): New expander definitions.
11648 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
11651 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
11653 2017-09-26 Richard Biener <rguenther@suse.de>
11655 PR tree-optimization/82320
11656 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
11659 2017-09-25 Jeff Law <law@redhat.com>
11661 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
11662 prototype for new argument.
11663 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
11664 mostly extracted from rs6000_emit_allocate_stack.
11665 (rs6000_emit_probe_stack_range_stack_clash): New function.
11666 (rs6000_emit_allocate_stack): Call
11667 rs6000_emit_probe_stack_range_stack_clash as needed.
11668 (rs6000_emit_probe_stack_range): Add additional argument
11669 to call to gen_probe_stack_range{si,di}.
11670 (output_probe_stack_range): New.
11671 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
11672 (output_probe_stack_range_stack_clash): New.
11673 (rs6000_emit_prologue): Emit notes into dump file as requested.
11674 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
11675 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
11676 Add additional operand and pass it to output_probe_stack_range.
11678 2017-09-25 Bin Cheng <bin.cheng@arm.com>
11680 PR tree-optimization/82163
11681 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
11682 (checking_verify_loop_closed_ssa): New parameter.
11683 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
11684 (check_loop_closed_ssa_stmt): Delete.
11685 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
11686 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
11687 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
11690 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
11692 * brig-builtins.def: Treat HSAIL barrier builtins as
11693 setjmp/longjump style functions.
11695 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11697 * target.def (constant_alignment): New hook.
11698 * defaults.h (CONSTANT_ALIGNMENT): Delete.
11699 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
11700 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
11701 * doc/tm.texi: Regenerate.
11702 * targhooks.h (default_constant_alignment): Declare.
11703 (constant_alignment_word_strings): Likewise.
11704 * targhooks.c (default_constant_alignment): New function.
11705 (constant_alignment_word_strings): Likewise.
11706 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
11707 instead of CONSTANT_ALIGNMENT.
11708 * varasm.c (align_variable, get_variable_align, build_constant_desc)
11709 (force_const_mem): Likewise.
11710 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
11711 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
11712 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
11713 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11714 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
11716 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
11717 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11718 constant_alignment_word_strings.
11719 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
11720 (CONSTANT_ALIGNMENT): Likewise.
11721 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11722 (arm_constant_alignment): New function.
11723 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
11724 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11725 constant_alignment_word_strings.
11726 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
11727 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11728 constant_alignment_word_strings.
11729 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
11730 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11731 (cris_constant_alignment): New function.
11732 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
11733 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11734 (epiphany_constant_alignment): New function.
11735 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
11736 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11737 constant_alignment_word_strings.
11738 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
11739 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11740 constant_alignment_word_strings.
11741 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
11742 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11743 constant_alignment_word_strings.
11744 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
11745 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
11746 * config/i386/i386.c (ix86_constant_alignment): Make static.
11747 Use the same interface as the target hook.
11748 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11749 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
11750 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11751 constant_alignment_word_strings.
11752 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
11753 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
11754 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11755 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
11756 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11757 constant_alignment_word_strings.
11758 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
11759 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11760 constant_alignment_word_strings.
11761 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
11762 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11763 constant_alignment_word_strings.
11764 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
11765 * config/microblaze/microblaze.c (microblaze_constant_alignment):
11767 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11768 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
11769 * config/mips/mips.c (mips_constant_alignment): New function.
11770 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11771 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
11772 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
11773 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11774 (mmix_constant_alignment): Make static. Use the same interface
11775 as the target hook.
11776 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
11777 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11778 constant_alignment_word_strings.
11779 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
11780 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11781 constant_alignment_word_strings.
11782 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
11783 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11784 constant_alignment_word_strings.
11785 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
11786 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11787 (rs6000_constant_alignment): New function.
11788 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
11789 * config/riscv/riscv.c (riscv_constant_alignment): New function.
11790 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11791 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
11792 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11793 (rs6000_constant_alignment): New function.
11794 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
11795 * config/s390/s390.c (s390_constant_alignment): New function.
11796 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11797 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
11798 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11799 constant_alignment_word_strings.
11800 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
11801 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11802 (sparc_constant_alignment): New function.
11803 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
11804 * config/spu/spu.c (spu_constant_alignment): New function.
11805 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11806 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
11807 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11808 constant_alignment_word_strings.
11809 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
11810 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11811 constant_alignment_word_strings.
11812 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
11813 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11814 constant_alignment_word_strings.
11815 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
11816 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11817 (visium_constant_alignment): New function.
11818 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
11819 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11820 (xtensa_constant_alignment): New function.
11821 * system.h (CONSTANT_ALIGNMENT): Poison.
11823 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
11825 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11826 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
11827 (rs6000_builtin_valid_without_lhs): New helper function.
11828 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11829 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
11831 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11833 * target.h (vec_perm_indices): Use unsigned short rather than
11835 (auto_vec_perm_indices): Likewise.
11836 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
11837 Use unsigned int rather than unsigned char.
11838 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
11840 2017-09-25 Richard Biener <rguenther@suse.de>
11842 * cfgloop.h (sort_sibling_loops): Declare.
11843 * cfgloop.c (sort_sibling_loops_cmp): New helper.
11844 (sort_sibling_loops): New function sorting the sibling loop list
11846 * graphite.c (graphite_transform_loops): Sort sibling loops.
11848 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
11850 * target.def (vec_perm_const_ok): Change sel parameter to
11852 * optabs-query.c (can_vec_perm_p): Update accordingly.
11853 * doc/tm.texi: Regenerate.
11854 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
11855 auto_vec_perm_indices and remove separate nelt field.
11856 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
11857 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
11858 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
11859 (aarch64_expand_vec_perm_const): Update accordingly.
11860 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
11861 to vec_perm_indices.
11862 * config/arm/arm.c (expand_vec_perm_d): Change perm to
11863 auto_vec_perm_indices and remove separate nelt field.
11864 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
11865 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
11866 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
11868 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
11869 to vec_perm_indices.
11870 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
11871 sel to vec_perm_indices.
11872 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
11873 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
11874 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
11876 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
11879 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
11882 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
11883 on the FUNCTION_DECL function context if it has a DIE that is a
11886 2017-09-25 Richard Biener <rguenther@suse.de>
11888 PR tree-optimization/82285
11889 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
11892 2017-09-25 Tom de Vries <tom@codesourcery.com>
11896 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
11899 2017-09-25 Richard Biener <rguenther@suse.de>
11901 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
11902 ISL errors other than isl_error_quota happen. Dump if the
11903 schedule is the same.
11904 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
11905 errors instead of aborting inside ISL.
11907 2017-09-25 Iain Sandoe <iain@codesourcery.com>
11910 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
11911 of libgcc_eh for m64.
11912 * config/i386/darwin64.h: Likewise.
11914 2017-09-25 Richard Biener <rguenther@suse.de>
11916 PR middle-end/82144
11917 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
11918 attribute for incomplete types nor twice for complete ones.
11920 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
11923 * config/i386/i386.c (ix86_print_operand_address_as): Encode
11924 %esp as %rsp to avoid 0x67 prefix if there is no index or base
11927 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11930 * config/i386/i386.opt (mprefer-avx256): Use
11931 ix86_target_flags variable.
11932 * config/i386/i386.c (ix86_target_string): Move
11933 -mprefer-avx256 to flag2_opts.
11935 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11937 PR middle-end/35691
11938 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11939 and x != -1 | y != -1 into (x & y) != -1.
11941 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11943 * config.gcc: Add new case statement to set
11944 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11945 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11946 s390x-*-linux* case statements. Added aarch64 to the list of
11947 supported architectures.
11949 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11951 PR tree-optimization/82289
11952 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11953 STMT_VINFO_RELEVANT_P.
11955 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11956 Alan Hayward <alan.hayward@arm.com>
11957 David Sherwood <david.sherwood@arm.com>
11959 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11960 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11961 (extract_range_from_binary_expr_1): Don't call
11962 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11964 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11965 Alan Hayward <alan.hayward@arm.com>
11966 David Sherwood <david.sherwood@arm.com>
11968 * target.def (preferred_vector_alignment): New hook.
11969 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11971 * doc/tm.texi: Regenerate.
11972 * targhooks.h (default_preferred_vector_alignment): Declare.
11973 * targhooks.c (default_preferred_vector_alignment): New function.
11974 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11976 (DR_TARGET_ALIGNMENT): New macro.
11977 (aligned_access_p): Update commentary.
11978 (vect_known_alignment_in_bytes): New function.
11979 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11981 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11982 Calculate the misalignment based on the target alignment rather than
11984 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11985 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11986 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11987 the target alignment, rather than masking the element misalignment
11988 with the number of elements in a vector. Also use the target
11989 alignment when calculating the maximum number of peels.
11990 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11991 instead of TYPE_ALIGN_UNIT.
11992 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11993 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11994 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11995 (vect_create_data_ref_ptr): Likewise.
11996 (vect_setup_realignment): Realign by ANDing with
11997 -DR_TARGET_MISALIGNMENT.
11998 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11999 the number of peels based on DR_TARGET_ALIGNMENT.
12000 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
12001 with the guaranteed alignment boundary when deciding whether
12003 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
12004 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
12005 (ensure_base_align): Remove stmt_info parameter. Get the
12006 target base alignment from DR_TARGET_ALIGNMENT.
12007 (vectorizable_store): Update call accordingly. Interpret
12008 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
12010 (vectorizable_load): Likewise.
12012 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
12013 Alan Hayward <alan.hayward@arm.com>
12014 David Sherwood <david.sherwood@arm.com>
12016 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
12017 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
12018 (vect_enhance_data_refs_alignment): Likewise.
12020 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12022 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
12023 error. Only quit immediately if parsing is complete.
12024 (BEGIN): Initialize fatal_err and parse_done.
12025 (begin fpu, end fpu): Check number of arguments.
12026 (begin arch, end arch): Likewise.
12027 (begin cpu, end cpu): Likewise.
12028 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
12029 (optalias): Likewise.
12031 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
12033 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
12034 * config/arm/arm-isa.h: Delete. Move definitions to ...
12035 * arm-cpus.in: ... here. Use new feature and fgroup values.
12036 * config/arm/arm.c (arm_option_override): Use lower case for feature
12038 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
12039 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
12040 * config/arm/parsecpu.awk (END): Add new command 'isa'.
12042 (print_isa_bits_for): New function.
12043 (gen_isa): New function.
12044 (gen_comm_data): Use print_isa_bits_for.
12045 (define feature): New keyword.
12046 (define fgroup): New keyword.
12047 * config/arm/t-arm (TM_H): Remove.
12048 (GTM_H): Add arm-isa.h.
12049 (arm-isa.h): Add rule to generate file.
12050 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
12051 case for feature bit names.
12053 2017-09-22 Richard Biener <rguenther@suse.de>
12055 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
12057 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
12058 print of no dependency loops ...
12059 * graphite.c (graphite_transform_loops): ... here.
12060 (canonicalize_loop_closed_ssa_form): Work from inner to outer
12062 (same_close_phi_node, remove_duplicate_close_phi,
12063 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
12064 (canonicalize_loop_closed_ssa): ... here and simplify.
12065 * graphite-optimize-isl.c: Include tree-vectorizer.h.
12066 (optimize_isl): Use dump_printf_loc to tell when we stopped
12067 optimizing because of an ISL timeout.
12069 2017-09-22 Richard Biener <rguenther@suse.de>
12071 PR tree-optimization/82291
12072 * tree-if-conv.c (predicate_mem_writes): Make sure to
12073 remove writes in blocks predicated with false.
12075 2017-09-22 Richard Biener <rguenther@suse.de>
12077 * sese.c: Include cfganal.h.
12078 (if_region_set_false_region): Remove.
12079 (create_if_region_on_edge): Likewise.
12080 (move_sese_in_condition): Re-implement without destroying
12083 2017-09-22 Richard Biener <rguenther@suse.de>
12085 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
12086 Verify both BBs contain loop PHI nodes before dispatching to
12087 copy_loop_phi_args.
12088 (graphite_regenerate_ast_isl): Do not recompute dominators,
12089 do not verify three times. Restructure for clarity.
12090 * graphite-scop-detection.c (same_close_phi_node,
12091 remove_duplicate_close_phi, make_close_phi_nodes_unique,
12092 defined_in_loop_p, canonicalize_loop_closed_ssa,
12093 canonicalize_loop_closed_ssa_form): Simplify, remove excess
12094 checking and SSA rewrite, move to ...
12095 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
12096 (graphite_initialize): Do not pass in ctx, do not reset the
12097 SCEV cache, compute only dominators.
12098 (graphite_transform_loops): Allocate ISL ctx after
12099 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
12100 Maintain post-dominators only around build_scops.
12101 * sese.c (if_region_set_false_region): Make static. Free
12102 and recompute dominators.
12103 (move_sese_in_condition): Assert we don't get called with
12104 post-dominators computed.
12105 * sese.h (if_region_set_false_region): Remove.
12107 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
12109 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
12110 mode attribute for TARGET_AVX512VL.
12112 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
12114 * config/i386/i386.opt (mprefer-avx256): New option.
12115 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
12117 (ix86_preferred_simd_mode): Return 256-bit AVX modes
12118 for TARGET_PREFER_AVX256.
12119 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
12121 2017-09-21 Jeff Law <law@redhat.com>
12123 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
12124 Fix dump output if the only stack space is for pushed registers.
12126 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12128 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
12131 2017-09-21 Martin Sebor <msebor@redhat.com>
12134 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
12135 code (in C++) or code that triggers warnings.
12137 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
12139 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
12141 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
12143 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
12144 * sched-rgn.c: Ditto.
12145 * sel-sched-ir.c: Ditto.
12147 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
12149 * toplev.h (set_random_seed): Adjust return type.
12150 * toplev.c (init_local_tick): Move eager initialization of random_seed
12151 to get_random_seed. Adjust comment.
12152 (init_random_seed): Inline to get_random_seed, delete.
12153 (get_random_seed): Initialize random_seed lazily.
12154 (set_random_seed): Do not return previous value.
12155 (print_switch_value): Do not call get_random_seed.
12157 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
12159 * cgraph.c (delete_function_version): New, broken out from...
12160 (cgraph_node::delete_function_version): ...here. Rename to
12161 cgraph_node::delete_function_version_by_decl. Update all uses.
12162 (cgraph_node::remove): Call delete_function_version.
12164 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12167 * tree-inline.c (expand_call_inline): Emit clobber stmts for
12168 VAR_DECLs to which addressable non-volatile parameters are mapped
12169 and for id->retvar after the return value assignment. Clear
12170 id->retval and id->retbnd after inlining.
12172 2017-09-21 Richard Biener <rguenther@suse.de>
12174 PR tree-optimization/82276
12175 PR tree-optimization/82244
12176 * tree-vrp.c (build_assert_expr_for): Set
12177 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
12179 (remove_range_assertions): Revert earlier change.
12181 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
12184 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
12186 2017-09-21 Richard Biener <rguenther@suse.de>
12188 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12189 Restore valid IL after code generation errors.
12190 * graphite.c (graphite_transform_loops): Diagnose code
12191 generation issues as MSG_MISSED_OPTIMIZATION and continue
12192 with processing SCOPs.
12194 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12195 Alan Hayward <alan.hayward@arm.com>
12196 David Sherwood <david.sherwood@arm.com>
12198 * calls.c (compute_argument_addresses): Use simplify_gen_binary
12199 rather than choosing between plus_constant and gen_rtx_<CODE>.
12200 * expr.c (emit_push_insn): Likewise.
12201 (expand_expr_real_2): Likewise.
12203 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12204 Alan Hayward <alan.hayward@arm.com>
12205 David Sherwood <david.sherwood@arm.com>
12207 * loop-unroll.c (split_iv): Call copy_rtx on the step.
12209 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12210 Alan Hayward <alan.hayward@arm.com>
12211 David Sherwood <david.sherwood@arm.com>
12213 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
12214 calling tree_to_uhwi.
12216 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12217 Alan Hayward <alan.hayward@arm.com>
12218 David Sherwood <david.sherwood@arm.com>
12220 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
12221 INTEGER_CST rather than a negative test for ADDR_EXPR.
12223 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12224 Alan Hayward <alan.hayward@arm.com>
12225 David Sherwood <david.sherwood@arm.com>
12227 * tree-vrp.c (extract_range_from_binary_expr_1): Check
12228 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
12230 2017-09-21 Richard Biener <rguenther@suse.de>
12232 PR tree-optimization/71351
12233 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
12234 graphite_create_new_loop_guard): Remove, fold remaining parts
12236 (translate_isl_ast_node_for): ... here and simplify.
12238 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12241 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
12242 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
12243 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
12244 alternative always use QI mode, for -Os imov (=R,R) alternative
12245 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
12248 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12249 Jeff Law <law@redhat.com>
12251 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
12252 (allocate_stack_space): New function, partially extracted from
12253 s390_emit_prologue.
12254 (s390_emit_prologue): Track offset to most recent stack probe.
12255 Code to allocate space moved into allocate_stack_space.
12256 Dump actions when no stack is allocated.
12257 (s390_prologue_plus_offset): New function.
12258 (s390_emit_stack_probe): Likewise.
12260 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
12262 * common.opt (Wa, Wl, Wp, g, gz=): Add
12264 (gno-column-info): Remove.
12265 (gcolumn-info): Drop RejectNegative.
12266 (gno-): New prefix.
12267 (gno-record-gcc-switches): Remove.
12268 (grecord-gcc-switches): Drop RejectNegative.
12269 (gno-split-dwarf): Remove.
12270 (gsplit-dwarf): Drop RejectNegative.
12271 (gno-strict-dwarf): Remove.
12272 (gstrict-dwarf): Drop RejectNegative.
12273 * config/darwin.opt (gfull, gused): Add RejectNegative.
12274 * dwarf2out.c (gen_producer_string): Drop
12275 gno-record-gcc-switches handler.
12276 * optc-gen.awk: Add g to prefixes with negative forms.
12277 * opts-common.c (remapping_prefix_p): New.
12278 (find_opt): Check it.
12279 (generate_canonical_option): Test g prefix.
12280 (option_map): Add -gno- mapping.
12281 (add_misspelling_candidates): Check remapping_prefix_p.
12283 2017-09-20 Jeff Law <law@redhat.com>
12285 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
12286 thinko in stack clash protection support.
12288 * explow.c (compute_stack_clash_protection_loop_data): Use
12289 CONST_INT_P instead of explicit test. Verify object is a
12290 CONST_INT_P before looking at INTVAL.
12291 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
12292 instead of explicit test.
12294 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
12297 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
12298 address instead of to r1 and r11.
12300 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
12302 * config.gcc: Support "knm".
12303 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
12304 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12306 * config/i386/i386.c (m_KNM): Define.
12307 (processor_target_table): Add "knm".
12309 (ix86_option_override_internal): Add "knm".
12310 (ix86_issue_rate): Add PROCESSOR_KNM.
12311 (ix86_adjust_cost): Ditto.
12312 (ia32_multipass_dfa_lookahead): Ditto.
12313 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
12314 (fold_builtin_cpu): Add M_INTEL_KNM.
12315 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
12316 (processor_type): Add PROCESSOR_KNM.
12317 * config/i386/x86-tune.def: Add m_KNM.
12318 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
12320 2017-09-20 Richard Biener <rguenther@suse.de>
12322 PR tree-optimization/80213
12323 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
12324 are allowed in empty BBs as well.
12325 (canonicalize_loop_closed_ssa): Also look for other complex
12327 (scop_detection::get_sese): Include the loop-closed PHI block
12329 (scop_detection::merge_sese): Remove code adding extra blocks.
12330 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
12331 (build_scops): Assert the final returned scop is invalid.
12333 2017-09-20 Richard Biener <rguenther@suse.de>
12335 PR tree-optimization/82264
12336 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
12338 (vn_phi_lookup): Likewise.
12339 (vn_phi_insert): Likewise.
12341 2017-09-20 Jakub Jelinek <jakub@redhat.com>
12343 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
12344 that fits into uhwi or shwi, add DW_AT_const_value regardless
12345 of early_dwarf without going through RTL, using add_AT_unsigned
12348 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
12349 (DEBUG_LTO_ABBREV_SECTION): Likewise.
12350 (DEBUG_LTO_MACINFO_SECTION): Likewise.
12351 (DEBUG_MACRO_SECTION): Likewise.
12352 (DEBUG_LTO_MACRO_SECTION): Likewise.
12353 (DEBUG_STR_DWO_SECTION): Likewise.
12354 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
12355 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
12356 (DEBUG_LTO_DWO_LINE_SECTION): Define.
12357 (DEBUG_LTO_LINE_STR_SECTION): Define.
12358 (init_sections_and_labels): Initialize debug_line_str_section
12359 variable. Initialize debug_loc_section for -gdwarf-5 to
12360 DEBUG_LOCLISTS_SECTION. Formatting fixes.
12362 2017-09-20 Richard Biener <rguenther@suse.de>
12364 * graphite-sese-to-poly.c (extract_affine): Properly handle
12365 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
12367 2017-09-20 Richard Biener <rguenther@suse.de>
12369 PR tree-optimization/81373
12370 * graphite-scop-detection.c (build_cross_bb_scalars_def):
12371 Force SESE live-out defs to be handled even if they are
12374 2017-09-19 Jeff Law <law@redhat.com>
12376 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
12377 nothing for stack adjustments with REG_STACK_CHECK.
12378 * sched-deps.c (parse_add_or_inc): Reject insns with
12379 REG_STACK_CHECK from dependency breaking.
12380 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
12381 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
12382 * reg-notes.def (STACK_CHECK): New note.
12384 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
12385 (ix86_expand_prologue): Dump stack clash info as needed.
12386 Call ix86_adjust_stack_and_probe_stack_clash as needed.
12388 * function.c (dump_stack_clash_frame_info): New function.
12389 * function.h (dump_stack_clash_frame_info): Prototype.
12390 (enum stack_clash_probes): New enum.
12392 * config/alpha/alpha.c (alpha_expand_prologue): Also check
12393 flag_stack_clash_protection.
12394 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
12395 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
12396 (arm_frame_pointer_required): Likewise.
12397 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
12398 (ia64_expand_prologue): Likewise.
12399 * config/mips/mips.c (mips_expand_prologue): Likewise.
12400 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
12401 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12402 (sparc_flat_expand_prologue): Likewise.
12403 * config/spu/spu.c (spu_expand_prologue): Likewise.
12405 * explow.c: Include "params.h".
12406 (anti_adjust_stack_and_probe_stack_clash): New function.
12407 (get_stack_check_protect): Likewise.
12408 (compute_stack_clash_protection_loop_data): Likewise.
12409 (emit_stack_clash_protection_loop_start): Likewise.
12410 (emit_stack_clash_protection_loop_end): Likewise.
12411 (allocate_dynamic_stack_space): Use get_stack_check_protect.
12412 Use anti_adjust_stack_and_probe_stack_clash.
12413 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
12414 (emit_stack_clash_protection_loop_start): Likewise.
12415 (emit_stack_clash_protection_loop_end): Likewise.
12416 * rtl.h (get_stack_check_protect): Prototype.
12417 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
12418 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
12419 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
12421 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
12423 * doc/tm.texi: Rebuilt.
12424 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
12425 get_stack_check_protect.
12426 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
12427 * config/arm/arm.c (arm_expand_prologue): Likewise.
12428 (arm_frame_pointer_required): Likewise.
12429 * config/i386/i386.c (ix86_expand_prologue): Likewise.
12430 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
12431 * config/mips/mips.c (mips_expand_prologue): Likewise.
12432 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
12433 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
12434 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12435 (sparc_flat_expand_prologue): Likewise.
12437 * common.opt (-fstack-clash-protection): New option.
12438 * flag-types.h (enum stack_check_type): Note difference between
12439 -fstack-check= and -fstack-clash-protection.
12440 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
12441 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
12442 * toplev.c (process_options): Issue warnings/errors for cases
12443 not handled with -fstack-clash-protection.
12444 * doc/invoke.texi (-fstack-clash-protection): Document new option.
12445 (-fstack-check): Note additional problem with -fstack-check=generic.
12446 Note that -fstack-check is primarily for Ada and refer users
12447 to -fstack-clash-protection for stack-clash-protection.
12448 Document new params for stack clash protection.
12450 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
12452 * config/i386/i386.c (ix86_split_long_move): Do not handle
12453 address used for LEA in a special way.
12455 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
12457 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
12459 2017-09-19 Martin Sebor <msebor@redhat.com>
12462 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
12463 of incompatible types.
12465 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
12467 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12468 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
12469 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12470 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
12472 2017-09-19 Richard Biener <rguenther@suse.de>
12474 PR tree-optimization/82244
12475 * tree-vrp.c (remove_range_assertions): Do not propagate
12476 a constant to abnormals but replace the assert with a copy.
12478 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
12480 PR rtl-optimization/57878
12481 PR rtl-optimization/68988
12482 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
12483 avoidance test involving non_reload_pseudos. Move frequency test
12484 below the general fragmentation avoidance test.
12486 2017-09-19 Richard Biener <rguenther@suse.de>
12488 PR tree-optimization/69728
12489 * graphite-sese-to-poly.c (schedule_error): New global.
12490 (add_loop_schedule): Handle empty domain by failing the
12492 (build_original_schedule): Handle schedule_error.
12494 2017-09-19 Richard Biener <rguenther@suse.de>
12496 * graphite-scop-detection.c (scop_detection::can_represent_loop):
12497 Do not iterate to sibling loops but only to siblings of inner
12500 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
12503 * config/m68k/m68k.md (moveq feeding equality comparison): Check
12504 that the registers are different.
12506 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
12508 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
12509 to processor_model and "amdfam17h" to arch_names_table.
12510 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
12512 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12515 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
12517 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12518 Alan Hayward <alan.hayward@arm.com>
12519 David Sherwood <david.sherwood@arm.com>
12521 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
12523 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
12525 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
12526 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
12527 vect_schedule_slp_instance.
12528 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
12529 Update call to vect_slp_analyze_node_operations. Simplify return
12531 (vect_slp_analyze_bb_1): Update call accordingly.
12532 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
12533 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
12534 (vect_schedule_slp): Update call accordingly.
12536 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12537 Alan Hayward <alan.hayward@arm.com>
12538 David Sherwood <david.sherwood@arm.com>
12540 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
12541 with types that aren't in fact scalar.
12543 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12545 * tree-vect-slp.c (vect_record_max_nunits): New function,
12547 (vect_build_slp_tree_1): ...here.
12548 (vect_build_slp_tree_2): Call it for phis too.
12550 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12552 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
12553 to vect_get_vec_def_for_operand when getting the mask operand.
12555 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12556 Alan Hayward <alan.hayward@arm.com>
12557 David Sherwood <david.sherwood@arm.com>
12559 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
12562 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12563 Alan Hayward <alan.hayward@arm.com>
12564 David Sherwood <david.sherwood@arm.com>
12566 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
12567 calculation for vector booleans.
12569 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12570 Alan Hayward <alan.hayward@arm.com>
12571 David Sherwood <david.sherwood@arm.com>
12573 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
12575 (vect_transform_stmt): ...here.
12576 (vect_analyze_stmt): Use it instead of calling
12577 vectorizable_live_operation directly.
12579 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
12581 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
12582 non-SIMT targets in acc vector loops.
12584 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
12586 * configure.ac: Add arc and check if assembler supports gdwarf2.
12587 * configure: Regenerate.
12589 2017-09-18 Richard Biener <rguenther@suse.de>
12591 PR tree-optimization/82220
12592 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
12593 epilogue niters from the min_profitable_iters compute.
12595 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12598 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
12599 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
12600 (ix86_init_pic_reg): Revert 2017-09-01 changes.
12602 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
12605 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
12606 switching to a new text section.
12608 2017-09-18 Richard Biener <rguenther@suse.de>
12610 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
12612 (build_alias_set): Reject aliases with no access function.
12614 2017-09-18 Richard Biener <rguenther@suse.de>
12616 PR tree-optimization/79622
12617 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
12619 (build_cross_bb_scalars_use): Likewise.
12621 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
12623 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
12625 2017-09-18 Alan Modra <amodra@gmail.com>
12628 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
12629 stack_pointer_rtx for count 0. Update comments. Break up
12630 large rtl expression.
12632 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
12634 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
12635 Increase to 20 bytes.
12636 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
12637 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
12638 or avx version of the stub.
12640 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
12643 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
12644 compute the minimum stack alignment. Also update preferred stack
12645 boundary for leaf functions.
12647 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
12649 PR tree-optimization/82228
12650 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
12653 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12655 * common/config/nds32/nds32-common.c
12656 (nds32_option_optimization_table): Refine formatting.
12657 (nds32_option_optimization_table): Use -fsched-pressure and
12658 -fomit-frame-pointer for specific optimization level.
12660 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12662 * config/nds32/nds32.c: Refine formatting and comments.
12663 * config/nds32/nds32.h: Likewise.
12664 * config/nds32/nds32.md: Likewise.
12665 * config/nds32/nds32-cost.c: Likewise.
12666 * config/nds32/nds32-isr.c: Likewise.
12667 * config/nds32/nds32-md-auxiliary.c: Likewise.
12668 * config/nds32/nds32-multiple.md: Likewise.
12669 * config/nds32/nds32-predicates.c: Likewise.
12671 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
12672 Jakub Jelinek <jakub@redhat.com>
12674 Add support for -std=c++2a.
12675 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
12677 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
12679 2017-09-15 Steve Ellcey <sellcey@cavium.com>
12682 * doc/extend.texi (Common Function Attributes): Add
12683 references to ARM, AArch64, and S/390 specific attributes.
12684 (Function Specific Option Pragmas): Add AArch64 and S/390
12685 to list of back ends that support the target pragma.
12687 2017-09-15 Nathan Sidwell <nathan@acm.org>
12689 * doc/standards.texi: Fix C++17 description. Update URLs for
12692 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
12694 * common.opt (Wcast-align=strict): New warning option.
12695 * doc/invoke.texi: Document -Wcast-align=strict.
12697 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
12699 * cgraph.h (cgraph_thunk_info): Add comments.
12700 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
12701 assert for VIRTUAL_* arguments stricter.
12703 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
12705 PR tree-optimization/71026
12706 * match.pd: Move RDIV patterns from fold-const.c
12707 * fold-const.c (distribute_real_division): Removed.
12708 (fold_binary_loc): Remove calls to distribute_real_divison.
12710 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12712 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
12713 c++1z and gnu++1z as deprecated. Change other references to
12714 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
12715 Change -Wc++1z-compat to -Wc++17-compat.
12716 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
12717 * dwarf2out.c (highest_c_language): Handle C++17.
12718 (gen_compile_unit_die): Likewise.
12720 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12722 PR rtl-optimization/82192
12723 * combine.c (make_extraction): Don't look through non-paradoxical
12724 SUBREGs or TRUNCATE if pos + len is or might be bigger than
12727 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12728 Alan Hayward <alan.hayward@arm.com>
12729 David Sherwood <david.sherwood@arm.com>
12731 * target.def (function_arg_offset): New hook.
12732 * targhooks.h (default_function_arg_offset): Declare.
12733 * targhooks.c (default_function_arg_offset): New function.
12734 * function.c (locate_and_pad_parm): Use
12735 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
12736 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
12737 (TARGET_FUNCTION_ARG_OFFSET): ...this.
12738 * doc/tm.texi: Regenerate.
12739 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
12740 * config/spu/spu.c (spu_function_arg_offset): New function.
12741 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
12742 * system.h (FUNCTION_ARG_OFFSET): Poison.
12744 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12745 Alan Hayard <alan.hayward@arm.com>
12746 David Sherwood <david.sherwood@arm.com>
12748 * target.def (truly_noop_truncation): New hook.
12749 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
12750 than TRULY_NOOP_TRUNCATION.
12751 * hooks.h (hook_bool_uint_uint_true): Declare.
12752 * hooks.c (hook_bool_uint_uint_true): New function.
12753 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
12754 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
12755 * doc/tm.texi: Regenerate.
12756 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
12757 rather than TRULY_NOOP_TRUNCATION in comments.
12758 (simplify_comparison): Likewise.
12759 (record_truncated_value): Likewise.
12760 * expmed.c (extract_bit_field_1): Likewise.
12761 (extract_split_bit_field): Likewise.
12762 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
12763 instead of TRULY_NOOP_TRUNCATION.
12764 * function.c (assign_parm_setup_block): Likewise.
12765 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
12766 * rtlhooks.c: Include target.h.
12767 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
12768 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
12769 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
12770 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
12771 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
12772 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
12773 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
12774 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
12775 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
12776 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
12777 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
12778 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
12779 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
12780 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
12781 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
12782 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
12783 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
12784 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
12785 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
12786 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
12787 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
12788 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
12789 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
12790 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
12791 * config/mips/mips.c (mips_truly_noop_truncation): New function.
12792 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12793 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12794 rather than TRULY_NOOP_TRUNCATION in comments.
12795 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
12796 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
12797 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
12798 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
12799 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
12800 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
12801 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
12802 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
12803 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
12804 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
12805 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
12806 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12807 rather than TRULY_NOOP_TRUNCATION in comments.
12808 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
12809 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
12810 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
12811 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
12812 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
12813 TRULY_NOOP_TRUNCATION condition.
12814 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
12815 (TRULY_NOOP_TRUNCATION): Delete.
12816 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
12817 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
12818 * config/spu/spu.c (spu_truly_noop_truncation): New function.
12819 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12820 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
12821 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
12822 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
12823 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12824 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12825 rather than TRULY_NOOP_TRUNCATION in comments.
12826 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
12827 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
12828 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
12829 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
12830 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
12831 * system.h (TRULY_NOOP_TRUNCATION): Poison.
12833 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
12836 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
12837 (*cmp_ior): Likewise.
12838 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
12839 (*ior_scc_scc_cmp): Likewise.
12840 (*and_scc_scc): Likewise.
12841 (*and_scc_scc_cmp): Likewise.
12843 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12844 Alan Hayard <alan.hayward@arm.com>
12845 David Sherwood <david.sherwood@arm.com>
12847 * target.def (can_change_mode_class): New hook.
12848 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
12849 (hard_regno_nregs): Likewise.
12850 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
12851 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
12852 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
12853 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
12854 (LOAD_EXTEND_OP): Update accordingly.
12855 * doc/tm.texi: Regenerate.
12856 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
12857 CANNOT_CHANGE_MODE_CLASS.
12858 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
12859 (REG_CAN_CHANGE_MODE_P): ...this new macro.
12860 * combine.c (simplify_set): Update accordingly.
12861 * emit-rtl.c (validate_subreg): Likewise.
12862 * recog.c (general_operand): Likewise.
12863 * regcprop.c (mode_change_ok): Likewise.
12864 * reload1.c (choose_reload_regs): Likewise.
12865 (inherit_piecemeal_p): Likewise.
12866 * rtlanal.c (simplify_subreg_regno): Likewise.
12867 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
12868 instead of CANNOT_CHANGE_MODE_CLASS.
12869 (reload_cse_simplify_operands): Likewise.
12870 * reload.c (push_reload): Use targetm.can_change_mode_class
12871 instead of CANNOT_CHANGE_MODE_CLASS.
12872 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
12873 REG_CANNOT_CHANGE_MODE_P.
12874 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12875 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
12876 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12877 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12878 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12879 (arm_can_change_mode_class): New function.
12880 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12881 than CANNOT_CHANGE_MODE_CLASS in comments.
12882 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12883 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
12884 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
12885 (ix86_can_change_mode_class): ...this new function, inverting the
12886 sense of the return value.
12887 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12888 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12889 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12890 (ia64_can_change_mode_class): New function.
12891 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12892 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
12893 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
12894 (m32c_can_change_mode_class): ...this new function, inverting the
12895 sense of the return value.
12896 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12897 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12898 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
12899 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
12900 (mips_can_change_mode_class): ...this new function, inverting the
12901 sense of the return value.
12902 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12903 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12904 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12905 (msp430_can_change_mode_class): New function.
12906 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12907 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
12908 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12909 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12910 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12911 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
12912 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12913 (pa_cannot_change_mode_class): Replace with...
12914 (pa_can_change_mode_class): ...this new function, inverting the
12915 sense of the return value.
12916 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12917 than CANNOT_CHANGE_MODE_CLASS in comments.
12918 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12919 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
12920 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12921 (pdp11_cannot_change_mode_class): Replace with...
12922 (pdp11_can_change_mode_class): ...this new function, inverting the
12923 sense of the return value.
12924 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12925 * config/powerpcspe/powerpcspe-protos.h
12926 (rs6000_cannot_change_mode_class_ptr): Delete.
12927 * config/powerpcspe/powerpcspe.c
12928 (rs6000_cannot_change_mode_class_ptr): Delete.
12929 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12930 (rs6000_option_override_internal): Assign to
12931 targetm.can_change_mode_class instead of
12932 rs6000_cannot_change_mode_class_ptr.
12933 (rs6000_cannot_change_mode_class): Replace with...
12934 (rs6000_can_change_mode_class): ...this new function, inverting the
12935 sense of the return value.
12936 (rs6000_debug_cannot_change_mode_class): Replace with...
12937 (rs6000_debug_can_change_mode_class): ...this new function.
12938 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12939 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12940 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12941 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12942 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12944 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12945 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12946 (rs6000_option_override_internal): Assign to
12947 targetm.can_change_mode_class instead of
12948 rs6000_cannot_change_mode_class_ptr.
12949 (rs6000_cannot_change_mode_class): Replace with...
12950 (rs6000_can_change_mode_class): ...this new function, inverting the
12951 sense of the return value.
12952 (rs6000_debug_cannot_change_mode_class): Replace with...
12953 (rs6000_debug_can_change_mode_class): ...this new function.
12954 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12955 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12956 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12957 (s390_can_change_mode_class): ...this new function, inverting the
12958 sense of the return value.
12959 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12960 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12961 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12962 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12963 (sh_cannot_change_mode_class): Replace with...
12964 (sh_can_change_mode_class): ...this new function, inverting the
12965 sense of the return value.
12966 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12967 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12968 (sparc_can_change_mode_class): New function.
12969 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12970 * config/spu/spu.c (spu_can_change_mode_class): New function.
12971 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12972 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12973 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12974 (visium_can_change_mode_class): New function.
12975 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12977 2017-09-15 Richard Biener <rguenther@suse.de>
12979 PR tree-optimization/82217
12980 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12981 but not undefined case.
12983 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12986 * postreload.c (reload_cse_simplify_operands): Skip
12987 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12989 2017-09-15 Richard Biener <rguenther@suse.de>
12991 PR tree-optimization/68823
12992 * graphite-scop-detection.c (build_alias_set): If we have a
12993 possible dependence check whether we can handle them by just
12994 looking at the DRs DR_ACCESS_FNs.
12995 (build_scops): If build_alias_set fails, fail the SCOP.
12997 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12999 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
13000 to support float128 built-in functions that require the ISA 3.0
13002 (BU_FLOAT128_3_HW): Likewise.
13003 (SQRTF128): Add support for the IEEE 128-bit square root and fma
13004 built-in functions.
13005 (FMAF128): Likewise.
13007 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13008 support for built-in functions that need the ISA 3.0 IEEE 128-bit
13009 floating point instructions.
13010 (rs6000_invalid_builtin): Likewise.
13011 (rs6000_builtin_mask_names): Likewise.
13012 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
13013 (RS6000_BTM_FLOAT128_HW): Likewise.
13014 (RS6000_BTM_COMMON): Likewise.
13015 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
13017 * doc/extend.texi (RS/6000 built-in functions): Document the
13018 IEEE 128-bit floating point square root and fused multiply-add
13019 built-in functions.
13021 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
13023 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
13024 reg (r2) isn't in the set of registers defined in the prologue.
13026 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13027 Alan Hayward <alan.hayward@arm.com>
13028 David Sherwood <david.sherwood@arm.com>
13030 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
13031 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
13032 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
13033 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
13034 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
13036 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
13037 max_vectorization_factor.
13038 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
13040 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13041 Alan Hayward <alan.hayward@arm.com>
13042 David Sherwood <david.sherwood@arm.com>
13044 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
13045 (vect_worthwhile_without_simd_p): Declare.
13046 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
13047 (vectorizable_reduction): Use it.
13048 * tree-vect-stmts.c (vectorizable_shift): Likewise.
13049 (vectorizable_operation): Likewise.
13051 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13052 Alan Hayward <alan.hayward@arm.com>
13053 David Sherwood <david.sherwood@arm.com>
13055 * tree-vectorizer.h (vect_get_num_copies): New function.
13056 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
13057 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13058 (vectorizable_induction): Likewise.
13059 (vectorizable_live_operation): Likewise.
13060 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13061 (vectorizable_bswap): Likewise.
13062 (vectorizable_call): Likewise.
13063 (vectorizable_conversion): Likewise.
13064 (vectorizable_assignment): Likewise.
13065 (vectorizable_shift): Likewise.
13066 (vectorizable_operation): Likewise.
13067 (vectorizable_store): Likewise.
13068 (vectorizable_load): Likewise.
13069 (vectorizable_condition): Likewise.
13070 (vectorizable_comparison): Likewise.
13071 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
13073 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13074 Alan Hayward <alan.hayward@arm.com>
13075 David Sherwood <david.sherwood@arm.com>
13077 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
13078 of vect_init_vector.
13080 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13081 Alan Hayward <alan.hayward@arm.com>
13082 David Sherwood <david.sherwood@arm.com>
13084 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
13085 an inline wrapper that provides a location.
13086 (gimple_build_vector): Likewise.
13087 * gimple-fold.c (gimple_build_vector_from_val): New function.
13088 (gimple_build_vector): Likewise.
13089 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
13090 functions to build the initial value. Always return a gimple value.
13091 (get_initial_defs_for_reduction): Likewise. Only compute
13093 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
13094 vect_init_vector on the results from get_initial_def(s)_for_reduction.
13095 (vectorizable_induction): Use gimple_build_vector rather than
13098 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13099 Alan Hayward <alan.hayward@arm.com>
13100 David Sherwood <david.sherwood@arm.com>
13102 * target.h (vec_perm_indices): New typedef.
13103 (auto_vec_perm_indices): Likewise.
13104 * optabs-query.h: Include target.h
13105 (can_vec_perm_p): Take a vec_perm_indices *.
13106 * optabs-query.c (can_vec_perm_p): Likewise.
13107 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
13108 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13109 * tree-vect-generic.c (lower_vec_perm): Likewise.
13110 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
13111 (vect_grouped_load_supported): Likewise.
13112 (vect_shift_permute_load_chain): Likewise.
13113 (vect_permute_store_chain): Use auto_vec_perm_indices.
13114 (vect_permute_load_chain): Likewise.
13115 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
13116 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
13117 Update uses of can_vec_perm_p.
13118 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
13119 mode with a number of elements. Take a vec_perm_indices *.
13120 (vect_create_epilog_for_reduction): Update accordingly.
13121 Use auto_vec_perm_indices.
13122 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
13123 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
13124 (vect_transform_slp_perm_load): Likewise.
13125 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
13126 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
13127 (vect_gen_perm_mask_checked): Likewise.
13128 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
13129 (vect_gen_perm_mask_checked): Likewise.
13130 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
13131 (vectorizable_store): Likewise.
13132 (vectorizable_load): Likewise.
13133 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
13134 (vectorizable_bswap): Likewise.
13136 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13137 Alan Hayward <alan.hayward@arm.com>
13138 David Sherwood <david.sherwood@arm.com>
13140 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
13141 * tree.c (build_vector): Likewise.
13142 (build_vector_from_ctor): Update accordingly.
13143 (build_vector_from_val): Likewise.
13144 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13145 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13146 * tree-vect-generic.c (add_rshift): Likewise.
13147 (expand_vector_divmod): Likewise.
13148 (optimize_vector_constructor): Likewise.
13149 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
13150 (vect_transform_slp_perm_load): Likewise.
13151 (vect_schedule_slp_instance): Likewise.
13152 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
13153 (vectorizable_call): Likewise.
13154 (vect_gen_perm_mask_any): Likewise. Add elements in order.
13155 * expmed.c (make_tree): Likewise.
13156 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
13157 a vector passed to build_vector.
13158 (fold_convert_const): Likewise.
13159 (exact_inverse): Likewise.
13160 (fold_ternary_loc): Likewise.
13161 (fold_relational_const): Likewise.
13162 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
13163 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
13164 (const_unop): Likewise. Store the reduction accumulator in a
13165 variable rather than an array.
13166 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
13167 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
13168 the new vector, rather than constructing it after the input arrays.
13169 (native_interpret_vector): Use auto_vec<tree> when building
13170 a vector passed to build_vector. Add elements in order.
13171 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
13172 auto_vec<tree> when building a vector passed to build_vector.
13173 (vect_create_epilog_for_reduction): Likewise.
13174 (vectorizable_induction): Likewise.
13175 (get_initial_def_for_reduction): Likewise. Fix indentation of
13177 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
13179 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
13180 passed to build_vector.
13182 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13183 Alan Hayward <alan.hayward@arm.com>
13184 David Sherwood <david.sherwood@arm.com>
13186 * tree-core.h (tree_base::u): Add an "nelts" field.
13187 (tree_vector): Use VECTOR_CST_NELTS as the length.
13188 * tree.c (tree_size): Likewise.
13189 (make_vector): Initialize VECTOR_CST_NELTS.
13190 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
13191 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
13192 TYPE_VECTOR_SUBPARTS.
13193 * expr.c (const_vector_mask_from_tree): Consistently use "units"
13194 as the number of units, setting it from VECTOR_CST_NELTS.
13195 (const_vector_from_tree): Likewise.
13196 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
13197 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
13198 (fold_negate_expr_1): Likewise.
13199 (fold_convert_const): Likewise.
13200 (const_binop): Likewise. Differentiate the number of output and
13202 (const_unop): Likewise.
13203 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
13204 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
13205 in cases that did the opposite.
13207 2017-09-14 Richard Biener <rguenther@suse.de>
13209 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
13212 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
13214 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
13216 2017-09-14 Jakub Jelinek <jakub@redhat.com>
13219 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
13220 if and where to split a bb, except for splitting before debug insn
13221 sequences followed by non-label real insn. Delete debug insns
13222 in between basic blocks.
13224 * combine.c (make_compound_operation_int): Formatting fixes.
13226 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
13227 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13228 * config/netbsd.h (LINK_EH_SPEC): Likewise.
13229 * config/sol2.h (LINK_EH_SPEC): Likewise.
13230 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13231 * config/s390/linux.h (LINK_SPEC): Likewise.
13232 * config/freebsd.h (LINK_EH_SPEC): Likewise.
13233 * config/openbsd.h (LINK_EH_SPEC): Likewise.
13234 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13235 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
13236 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
13237 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13238 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
13239 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
13241 2017-09-13 Jakub Jelinek <jakub@redhat.com>
13243 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
13245 (ENDFILE_LINUX_SPEC): Likewise.
13246 (LINK_EH_SPEC): Likewise.
13247 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
13248 (LINK_OS_LINUX_SPEC32): Likewise.
13249 (LINK_OS_LINUX_SPEC64): Likewise.
13250 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
13251 (LINK_OS_LINUX_SPEC): Likewise.
13253 2017-09-13 Martin Liska <mliska@suse.cz>
13255 PR middle-end/82154
13256 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
13257 CASE_HIGH is NULL_TREE.
13259 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13260 Alan Hayward <alan.hayward@arm.com>
13261 David Sherwood <david.sherwood@arm.com>
13263 * target.def (secondary_memory_needed): New hook.
13264 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
13265 instead of SECONDARY_MEMORY_NEEDED.
13266 (secondary_memory_needed_mode): Likewise.
13267 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
13268 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
13269 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
13270 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
13271 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
13272 * doc/tm.texi: Regenerate.
13273 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
13274 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
13275 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13276 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
13277 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
13278 * config/i386/i386.c (inline_secondary_memory_needed): Put the
13279 mode argument first and change the reg_class arguments to reg_class_t.
13280 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
13281 Make static. Update the call to inline_secondary_memory_needed.
13282 (ix86_register_move_cost): Update the call to
13283 inline_secondary_memory_needed.
13284 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13285 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
13287 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
13288 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13290 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
13291 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
13292 * config/mips/mips.c (mips_secondary_memory_needed): Make static
13293 and match hook interface. Add comment from mips.h.
13294 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13295 * config/mmix/mmix.md (truncdfsf2): Refer to
13296 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13298 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
13299 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
13300 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13301 (pa_secondary_memory_needed): New function.
13302 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
13303 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
13304 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13305 (pdp11_secondary_memory_needed): Make static and match hook interface.
13306 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
13307 * config/powerpcspe/powerpcspe-protos.h
13308 (rs6000_secondary_memory_needed_ptr): Delete.
13309 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
13311 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13312 (rs6000_option_override_internal): Assign to
13313 targetm.secondary_memory_needed rather than
13314 rs6000_secondary_memory_needed_ptr.
13315 (rs6000_secondary_memory_needed): Match hook interface.
13316 (rs6000_debug_secondary_memory_needed): Likewise.
13317 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
13318 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
13319 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
13320 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13321 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
13322 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
13324 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
13325 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13326 (rs6000_option_override_internal): Assign to
13327 targetm.secondary_memory_needed rather than
13328 rs6000_secondary_memory_needed_ptr.
13329 (rs6000_secondary_memory_needed): Match hook interface.
13330 (rs6000_debug_secondary_memory_needed): Likewise.
13331 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
13332 * config/s390/s390.c (s390_secondary_memory_needed): New function.
13333 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13334 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
13335 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13336 (sparc_secondary_memory_needed): New function.
13337 * lra-constraints.c (check_and_process_move): Refer to
13338 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13340 (curr_insn_transform): Likewise.
13341 (process_alt_operands): Use targetm.secondary_memory_needed
13342 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13343 (check_secondary_memory_needed_p): Likewise.
13344 (choose_split_class): Likewise.
13345 * reload.c: Unconditionally include code that was previously
13346 conditional on SECONDARY_MEMORY_NEEDED.
13347 (push_secondary_reload): Use targetm.secondary_memory_needed
13348 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13349 (push_reload): Likewise.
13350 * reload1.c: Unconditionally include code that was previously
13351 conditional on SECONDARY_MEMORY_NEEDED.
13352 (choose_reload_regs): Use targetm.secondary_memory_needed
13353 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13354 (gen_reload): Likewise.
13355 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
13357 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13358 Alan Hayward <alan.hayward@arm.com>
13359 David Sherwood <david.sherwood@arm.com>
13361 * target.def (secondary_memory_needed_mode): New hook:
13362 * targhooks.c (default_secondary_memory_needed_mode): Declare.
13363 * targhooks.h (default_secondary_memory_needed_mode): New function.
13364 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
13365 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
13366 * doc/tm.texi: Regenerate.
13367 * lra-constraints.c (check_and_process_move): Use
13368 targetm.secondary_memory_needed_mode instead of
13369 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
13370 (curr_insn_transform): Likewise.
13371 * reload.c (get_secondary_mem): Likewise.
13372 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13373 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
13375 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13376 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13377 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
13378 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13379 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
13381 * config/powerpcspe/powerpcspe-protos.h
13382 (rs6000_secondary_memory_needed_mode): Delete.
13383 * config/powerpcspe/powerpcspe.c
13384 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13385 (rs6000_secondary_memory_needed_mode): Make static.
13386 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13387 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
13389 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13391 (rs6000_secondary_memory_needed_mode): Make static.
13392 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13393 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
13394 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13395 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13396 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13398 (sparc_secondary_memory_needed_mode): New function.
13399 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
13401 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
13403 * config/aarch64/constraints.md (Umq): New constraint.
13404 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
13406 (mov<mode>): Update condition.
13408 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13410 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
13411 when bitposition is the same.
13413 2017-09-13 Richard Biener <rguenther@suse.de>
13415 * dwarf2out.c (output_die_symbol): Remove.
13416 (output_die): Do not output a DIEs symbol.
13418 2017-09-13 Richard Biener <rguenther@suse.de>
13420 PR middle-end/82128
13421 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
13422 default-def to avoid breaking iterator update with the weird
13423 interaction with cgraph_update_edges_for_call_stmt_node.
13425 2017-09-13 Richard Biener <rguenther@suse.de>
13427 * tree-cfg.c (verify_gimple_assign_binary): Add verification
13428 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
13429 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
13430 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
13432 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13434 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
13435 Disable pc relative literal load irrespective of
13436 TARGET_FIX_ERR_A53_84341 for default.
13438 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
13440 * config/sparc/sparc.c (output_return): Output the source location of
13441 the insn in the delay slot, if any.
13442 (output_sibcall): Likewise.
13444 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
13447 * common.opt (-static-pie): New alias.
13448 (shared): Negate static-pie.
13449 (-no-pie): Update help text.
13451 (static-pie): New option.
13452 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
13453 -static-pie support.
13454 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
13455 (LINK_EH_SPEC): Likewise.
13456 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13457 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13458 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13459 * gcc.c (LINK_COMMAND_SPEC): Likewise.
13460 (init_gcc_specs): Likewise.
13461 (init_spec): Likewise.
13462 (display_help): Update help message for -pie.
13463 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
13466 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
13468 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
13469 (movdi_aarch64): Likewise.
13470 (movti_aarch64): Likewise.
13472 2017-09-12 Simon Wright <simon@pushface.org>
13475 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
13476 calculation of the minor version, always output as 0.
13478 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13481 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13482 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
13483 on it early, rather than manual conversion late. For
13484 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
13485 instead of performing manual conversion.
13487 2017-09-12 Carl Love <cel@us.ibm.com>
13489 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
13490 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
13492 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13493 VMULOSW): Add definitions.
13494 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13495 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13496 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13497 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13498 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13500 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13502 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
13504 (movti_aarch64): Likewise.
13505 (movdf_aarch64): Likewise.
13506 (movtf_aarch64): Likewise.
13507 (load_pairdi): Likewise.
13508 (store_pairdi): Likewise.
13509 (load_pairdf): Likewise.
13510 (store_pairdf): Likewise.
13511 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
13512 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
13513 (ldr_got_small_<mode>): Likewise.
13514 (ldr_got_small_28k_<mode>): Likewise.
13515 (ldr_got_tiny): Likewise.
13516 * config/aarch64/iterators.md (ldst_sz): New.
13517 (ldpstp_sz): Likewise.
13518 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
13520 (thunderx_load): Split load_8 to load_16.
13521 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
13523 (thunderx2t99_storepair_basic): Split store_8 to store_16.
13524 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
13525 (xgene1_store_pair): Split store_8 to store_16.
13526 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
13527 (falkor_st_0_st_sd): Split store_8 to store_16.
13529 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13531 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
13532 and store1/2/3/4 to store_4/8/12/16.
13533 * config/aarch64/aarch64.md: Update for rename.
13534 * config/arm/arm.md: Likewise.: Likewise.
13535 * config/arm/arm.c: Likewise.
13536 * config/arm/thumb1.md: Likewise.
13537 * config/arm/thumb2.md: Likewise.
13538 * config/arm/vfp.md: Likewise.
13539 * config/arm/arm-generic.md: Likewise.
13540 * config/arm/arm1020e.md: Likewise.
13541 * config/arm/arm1026ejs.md: Likewise.
13542 * config/arm/arm1136jfs.md: Likewise.
13543 * config/arm/arm926ejs.md: Likewise.
13544 * config/arm/cortex-a15.md: Likewise.
13545 * config/arm/cortex-a17.md: Likewise.
13546 * config/arm/cortex-a5.md: Likewise.
13547 * config/arm/cortex-a53.md: Likewise.
13548 * config/arm/cortex-a57.md: Likewise.
13549 * config/arm/cortex-a7.md: Likewise.
13550 * config/arm/cortex-a8.md: Likewise.
13551 * config/arm/cortex-a9.md: Likewise.
13552 * config/arm/cortex-m4.md: Likewise.
13553 * config/arm/cortex-m7.md: Likewise.
13554 * config/arm/cortex-r4.md: Likewise.
13555 * config/arm/exynos-m1.md: Likewise.
13556 * config/arm/fa526.md: Likewise.
13557 * config/arm/fa606te.md: Likewise.
13558 * config/arm/fa626te.md: Likewise.
13559 * config/arm/fa726te.md: Likewise.
13560 * config/arm/fmp626.md: Likewise.
13561 * config/arm/iwmmxt.md: Likewise.
13562 * config/arm/ldmstm.md: Likewise.
13563 * config/arm/marvell-pj4.md: Likewise.
13564 * config/arm/xgene1.md: Likewise.
13565 * config/aarch64/thunderx.md: Likewise.
13566 * config/aarch64/thunderx2t99.md: Likewise.
13567 * config/aarch64/falkor.md: Likewise.
13569 2017-09-12 Martin Liska <mliska@suse.cz>
13571 * attribs.c (private_lookup_attribute): New function.
13572 * attribs.h (private_lookup_attribute): Declared here.
13573 (lookup_attribute): Called from this place.
13575 2017-09-12 Richard Biener <rguenther@suse.de>
13577 PR tree-optimization/82157
13578 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
13579 stmts with side-effects.
13581 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13582 Alan Hayward <alan.hayward@arm.com>
13583 David Sherwood <david.sherwood@arm.com>
13585 * target.def (hard_regno_nregs): New hook.
13586 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
13587 * targhooks.h (default_hard_regno_nregs): Declare.
13588 * targhooks.c (default_hard_regno_nregs): New function.
13589 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
13590 (TARGET_HARD_REGNO_NREGS): ...this hook.
13591 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
13592 (CLASS_MAX_NREGS): Likewise.
13593 * doc/tm.texi: Regenerate.
13594 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
13595 instead of HARD_REGNO_NREGS.
13596 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
13597 HARD_REGNO_NREGS in the comment.
13598 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
13599 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
13600 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
13601 Return an unsigned int.
13602 (TARGET_HARD_REGNO_NREGS): Redefine.
13603 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
13604 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
13605 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13606 (arc_hard_regno_nregs): New function.
13607 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
13608 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
13609 (arm_hard_regno_nregs): New function.
13610 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
13611 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
13612 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
13613 (TARGET_HARD_REGNO_NREGS): Redefine.
13614 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
13615 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
13616 (HARD_REGNO_NREGS): Delete.
13617 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
13618 (cr16_hard_regno_nregs): New function.
13619 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
13620 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
13621 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
13622 (cris_hard_regno_nregs): New function.
13623 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
13624 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
13625 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13626 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
13627 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
13628 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
13629 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
13630 (frv_hard_regno_nregs): Make static. Take and return an
13632 (frv_class_max_nregs): Remove outdated copy of documentation.
13633 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
13634 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
13635 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
13636 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
13637 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
13638 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
13639 (TARGET_HARD_REGNO_NREGS): Redefine.
13640 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
13641 (CLASS_MAX_NREGS): Update comment.
13642 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
13643 (ia64_hard_regno_nregs): New function.
13644 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
13645 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
13646 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
13647 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
13648 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
13650 (m32c_hard_regno_nregs): Likewise. Make static.
13651 (TARGET_HARD_REGNO_NREGS): Redefine.
13652 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
13653 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
13654 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
13655 (m68k_hard_regno_nregs): New function.
13656 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
13657 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
13658 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
13659 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
13660 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
13661 Take and return an unsigned int.
13662 (TARGET_HARD_REGNO_NREGS): Redefine.
13663 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
13664 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13665 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
13666 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
13667 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
13668 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
13669 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
13670 (msp430_hard_regno_nregs): Make static. Take and return an
13672 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
13673 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
13674 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
13675 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
13676 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
13677 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
13678 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
13679 (TARGET_HARD_REGNO_NREGS): Redefine.
13680 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
13681 (PA_HARD_REGNO_NREGS): ...this.
13682 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
13683 (PA_HARD_REGNO_NREGS): ...this.
13684 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13685 (pa_hard_regno_nregs): New function.
13686 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
13687 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
13688 (pdp11_hard_regno_nregs): New function.
13689 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
13690 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
13691 (rs6000_hard_regno_nregs_hook): New function.
13692 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
13693 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
13694 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
13695 Take and return an unsigned int. Move earlier in file.
13696 (TARGET_HARD_REGNO_NREGS): Redefine.
13697 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
13698 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
13699 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
13700 (rl78_hard_regno_nregs): Make static. Take and return an
13702 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
13703 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
13704 (rs6000_hard_regno_nregs_hook): New function.
13705 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
13706 * config/rx/rx.c (rx_hard_regno_nregs): New function.
13707 (TARGET_HARD_REGNO_NREGS): Redefine.
13708 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
13709 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
13710 instead of HARD_REGNO_NREGS.
13711 (s390_hard_regno_nregs): New function.
13712 (s390_hard_regno_mode_ok): Add comment from s390.h.
13713 (TARGET_HARD_REGNO_NREGS): Redefine.
13714 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
13715 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
13716 (sh_hard_regno_nregs): New function.
13717 (sh_pass_in_reg_p): Use it.
13718 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
13719 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13720 (sparc_hard_regno_nregs): New function.
13721 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
13722 * config/spu/spu.c (spu_hard_regno_nregs): New function.
13723 (spu_function_arg_advance): Use it, supplying a valid register number.
13724 (TARGET_HARD_REGNO_NREGS): Redefine.
13725 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
13726 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
13727 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
13728 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
13729 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
13730 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
13731 (CLASS_MAX_NREGS): Remove copy of old documentation.
13732 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
13733 (visium_hard_regno_nregs): New function.
13734 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
13735 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
13736 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13737 xtensa_hard_regno_nregs): New function.
13738 * system.h (HARD_REGNO_NREGS): Poison.
13740 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13742 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
13743 hard_regno_nregs instead of HARD_REGNO_NREGS.
13744 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
13745 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
13746 (c6x_expand_epilogue): Likewise.
13747 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
13748 (frv_read_iacc_argument): Likewise.
13749 * config/sh/sh.c: Include regs.h.
13750 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13751 (regs_used): Likewise.
13752 (output_stack_adjust): Likewise.
13753 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
13754 * expmed.c: Include regs.h.
13755 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13756 * ree.c: Include regs.h.
13757 (combine_reaching_defs): Use hard_regno_nregs instead of
13759 (add_removable_extension): Likewise.
13761 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13763 * regs.h (hard_regno_nregs): Turn into a function.
13764 (end_hard_regno): Update accordingly.
13765 * caller-save.c (setup_save_areas): Likewise.
13766 (save_call_clobbered_regs): Likewise.
13767 (replace_reg_with_saved_mem): Likewise.
13768 (insert_restore): Likewise.
13769 (insert_save): Likewise.
13770 * combine.c (can_change_dest_mode): Likewise.
13771 (move_deaths): Likewise.
13772 (distribute_notes): Likewise.
13773 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
13774 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
13775 (rs6000_split_multireg_move): Likewise.
13776 (rs6000_register_move_cost): Likewise.
13777 (rs6000_memory_move_cost): Likewise.
13778 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
13779 (rs6000_split_multireg_move): Likewise.
13780 (rs6000_register_move_cost): Likewise.
13781 (rs6000_memory_move_cost): Likewise.
13782 * cselib.c (cselib_reset_table): Likewise.
13783 (cselib_lookup_1): Likewise.
13784 * emit-rtl.c (set_mode_and_regno): Likewise.
13785 * function.c (aggregate_value_p): Likewise.
13786 * ira-color.c (setup_profitable_hard_regs): Likewise.
13787 (check_hard_reg_p): Likewise.
13788 (calculate_saved_nregs): Likewise.
13789 (assign_hard_reg): Likewise.
13790 (improve_allocation): Likewise.
13791 (calculate_spill_cost): Likewise.
13792 * ira-emit.c (modify_move_list): Likewise.
13793 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
13794 (ira_hard_reg_in_set_p): Likewise.
13795 * ira.c (setup_reg_mode_hard_regset): Likewise.
13796 (clarify_prohibited_class_mode_regs): Likewise.
13797 (check_allocation): Likewise.
13798 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13799 (lra_setup_reg_renumber): Likewise.
13800 (setup_try_hard_regno_pseudos): Likewise.
13801 (spill_for): Likewise.
13802 (assign_hard_regno): Likewise.
13803 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
13804 * lra-constraints.c (in_class_p): Likewise.
13805 (lra_constraint_offset): Likewise.
13806 (simplify_operand_subreg): Likewise.
13807 (lra_constraints): Likewise.
13808 (split_reg): Likewise.
13809 (split_if_necessary): Likewise.
13810 (invariant_p): Likewise.
13811 (inherit_in_ebb): Likewise.
13812 * lra-lives.c (process_bb_lives): Likewise.
13813 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
13814 (get_hard_regs): Likewise.
13815 (do_remat): Likewise.
13816 * lra-spills.c (assign_spill_hard_regs): Likewise.
13817 * mode-switching.c (create_pre_exit): Likewise.
13818 * postreload.c (reload_combine_recognize_pattern): Likewise.
13819 * recog.c (peep2_find_free_register): Likewise.
13820 * regcprop.c (kill_value_regno): Likewise.
13821 (set_value_regno): Likewise.
13822 (copy_value): Likewise.
13823 (maybe_mode_change): Likewise.
13824 (find_oldest_value_reg): Likewise.
13825 (copyprop_hardreg_forward_1): Likewise.
13826 * regrename.c (check_new_reg_p): Likewise.
13827 (regrename_do_replace): Likewise.
13828 * reload.c (push_reload): Likewise.
13829 (combine_reloads): Likewise.
13830 (find_dummy_reload): Likewise.
13831 (operands_match_p): Likewise.
13832 (find_reloads): Likewise.
13833 (find_equiv_reg): Likewise.
13834 (reload_adjust_reg_for_mode): Likewise.
13835 * reload1.c (count_pseudo): Likewise.
13836 (count_spilled_pseudo): Likewise.
13837 (find_reg): Likewise.
13838 (clear_reload_reg_in_use): Likewise.
13839 (free_for_value_p): Likewise.
13840 (allocate_reload_reg): Likewise.
13841 (choose_reload_regs): Likewise.
13842 (reload_adjust_reg_for_temp): Likewise.
13843 (emit_reload_insns): Likewise.
13844 (delete_output_reload): Likewise.
13845 * rtlanal.c (subreg_get_info): Likewise.
13846 * sched-deps.c (sched_analyze_reg): Likewise.
13847 * sel-sched.c (init_regs_for_mode): Likewise.
13848 (mark_unavailable_hard_regs): Likewise.
13849 (choose_best_reg_1): Likewise.
13850 (verify_target_availability): Likewise.
13851 * valtrack.c (dead_debug_insert_temp): Likewise.
13852 * var-tracking.c (track_loc_p): Likewise.
13853 (emit_note_insn_var_location): Likewise.
13854 * varasm.c (make_decl_rtl): Likewise.
13855 * reginfo.c (choose_hard_reg_mode): Likewise.
13856 (init_reg_modes_target): Refer directly to
13857 this_target_regs->x_hard_regno_nregs.
13859 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13861 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
13862 instead of hard_regno_nregs.
13864 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13866 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
13867 end_hard_regno instead of hard_regno_nregs.
13868 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
13869 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
13870 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
13871 * ira-color.c (improve_allocation): Likewise.
13872 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13873 * lra-lives.c (mark_regno_live): Likewise.
13874 (mark_regno_dead): Likewise.
13875 * lra-remat.c (operand_to_remat): Likewise.
13876 * lra.c (collect_non_operand_hard_regs): Likewise.
13877 * postreload.c (reload_combine_note_store): Likewise.
13878 (move2add_valid_value_p): Likewise.
13879 * reload.c (regno_clobbered_p): Likewise.
13881 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13883 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
13885 * config/v850/v850.c (v850_reorg): Likewise.
13886 * reload.c (refers_to_regno_for_reload_p): Likewise.
13887 (find_equiv_reg): Likewise.
13888 * reload1.c (reload_reg_reaches_end_p): Likewise.
13890 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13892 * caller-save.c (add_used_regs): Use REG_NREGS instead of
13894 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
13895 * config/arm/arm.c (output_move_neon): Likewise.
13896 (arm_attr_length_move_neon): Likewise.
13897 (neon_split_vcombine): Likewise.
13898 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
13899 (c6x_mark_reg_written): Likewise.
13900 (c6x_dwarf_register_span): Likewise.
13901 * config/i386/i386.c (ix86_save_reg): Likewise.
13902 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
13903 (rws_access_reg): Likewise.
13904 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
13905 * mode-switching.c (create_pre_exit): Likewise.
13906 * ree.c (combine_reaching_defs): Likewise.
13907 (add_removable_extension): Likewise.
13908 * regcprop.c (find_oldest_value_reg): Likewise.
13909 (copyprop_hardreg_forward_1): Likewise.
13910 * reload.c (reload_inner_reg_of_subreg): Likewise.
13911 (push_reload): Likewise.
13912 (combine_reloads): Likewise.
13913 (find_dummy_reload): Likewise.
13914 (reload_adjust_reg_for_mode): Likewise.
13915 * reload1.c (find_reload_regs): Likewise.
13916 (forget_old_reloads_1): Likewise.
13917 (reload_reg_free_for_value_p): Likewise.
13918 (reload_adjust_reg_for_temp): Likewise.
13919 (emit_reload_insns): Likewise.
13920 (delete_output_reload): Likewise.
13921 * sel-sched.c (choose_best_reg_1): Likewise.
13922 (choose_best_pseudo_reg): Likewise.
13924 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13925 Alan Hayward <alan.hayward@arm.com>
13926 David Sherwood <david.sherwood@arm.com>
13928 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13929 * target.def (slow_unaligned_access): New hook.
13930 * targhooks.h (default_slow_unaligned_access): Declare.
13931 * targhooks.c (default_slow_unaligned_access): New function.
13932 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13933 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13934 * doc/tm.texi: Regenerate.
13935 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13936 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13937 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13939 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13940 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13942 (rs6000_slow_unaligned_access): New function.
13943 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13944 (expand_block_compare): Likewise.
13945 (expand_strn_compare): Likewise.
13946 (rs6000_rtx_costs): Likewise.
13947 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13948 (riscv_slow_unaligned_access): Likewise.
13949 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13950 (riscv_slow_unaligned_access_p): ...this and make static.
13951 (riscv_option_override): Update accordingly.
13952 (riscv_slow_unaligned_access): New function.
13953 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13954 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13955 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13956 (rs6000_slow_unaligned_access): New function.
13957 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13958 (rs6000_rtx_costs): Likewise.
13959 * config/rs6000/rs6000-string.c (expand_block_compare)
13960 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13961 of SLOW_UNALIGNED_ACCESS.
13962 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13963 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13964 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13965 of SLOW_UNALIGNED_ACCESS.
13966 * expmed.c (simple_mem_bitfield_p): Likewise.
13967 * expr.c (alignment_for_piecewise_move): Likewise.
13968 (emit_group_load_1): Likewise.
13969 (emit_group_store): Likewise.
13970 (copy_blkmode_from_reg): Likewise.
13971 (emit_push_insn): Likewise.
13972 (expand_assignment): Likewise.
13973 (store_field): Likewise.
13974 (expand_expr_real_1): Likewise.
13975 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13976 * lra-constraints.c (simplify_operand_subreg): Likewise.
13977 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13978 * gimple-ssa-store-merging.c: Likewise in block comment at start
13980 * tree-ssa-strlen.c: Include target.h.
13981 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13982 of SLOW_UNALIGNED_ACCESS.
13983 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13985 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13987 PR rtl-optimization/82185
13988 * expmed.c (emit_store_flag_int): Only test tem if it has been
13991 2017-09-12 Richard Biener <rguenther@suse.de>
13993 PR middle-end/82149
13994 * match.pd ((FTYPE) N CMP CST): Fix typo.
13996 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13998 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
14000 (mips_near_type_p): Add 'short_call' attribute as a synonym
14002 * doc/extend.texi (short_call): Document new function attribute.
14004 2017-09-12 Jakub Jelinek <jakub@redhat.com>
14007 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
14008 assertion check that in the condition.
14009 (get_atomic_generic_size): Likewise. Before testing if parameter
14010 has pointer type, if it has array type, call for C++
14011 default_conversion to perform array-to-pointer conversion.
14013 2017-09-12 Richard Biener <rguenther@suse.de>
14015 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
14016 for operations we cannot scalarize.
14018 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
14020 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
14021 vectors heap vectors. Clean up comments.
14022 Make visited_bbs a reference.
14023 (profitable_jump_thread_path): Make GC
14024 vectors heap vectors. Clean up comments.
14026 (convert_and_register_jump_thread_path): Make GC vectors heap
14028 (check_subpath_and_update_thread_path): Same. Clean up comments.
14029 Make visited_bbs a reference.
14030 (handle_phi): Abstract common code to to
14031 register_jump_thread_path_if_profitable.
14032 Rename VAR_BB to DEF_BB.
14034 Make GC vectors heap vectors.
14035 Make visited_bbs a reference.
14036 (handle_assignment): Same.
14037 (register_jump_thread_path_if_profitable): New.
14038 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
14040 Make GC vectors heap vectors. Clean up comments.
14041 Make visited_bbs a reference.
14042 (find_jump_threads_backwards): Make visited_bbs live in the stack.
14043 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
14046 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
14049 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
14050 words of E_DImode object are reachable by xtensa_uimm8x4 access.
14052 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
14054 Revert r251800 and r251799.
14056 2017-09-11 Martin Jambor <mjambor@suse.cz>
14059 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
14060 arguments in advance.
14061 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
14062 use it to find predecessor edges.
14063 (naive_outof_ssa): Collect vector of predecessors.
14065 2017-09-08 Jason Merrill <jason@redhat.com>
14067 PR c++/70029 - ICE with ref-qualifier and -flto
14068 * langhooks.h (struct lang_hooks_for_types): Add
14069 copy_lang_qualifiers.
14070 * attribs.c (build_type_attribute_qual_variant): Use it.
14071 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
14073 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
14074 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
14076 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
14079 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
14080 (*mulsi3_sp64): New instruction.
14081 (mulsi3): New expander.
14083 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
14085 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
14087 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14089 * sancov.c: Include memmodel.h.
14091 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
14094 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
14097 2017-09-07 Carl Love <cel@us.ibm.com>
14099 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
14100 the sldi instruction.
14102 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
14104 * sancov.c: Include tm_p.h.
14106 2017-09-07 Jakub Jelinek <jakub@redhat.com>
14109 * output.h (switch_to_other_text_partition): New declaration.
14110 * varasm.c (switch_to_other_text_partition): New function.
14111 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
14112 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
14113 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
14114 to the other text partition before emitting LCL label and switch back
14115 after emitting the word after it.
14117 2017-09-07 Richard Biener <rguenther@suse.de>
14119 * passes.def (pass_split_crit_edges): Remove instance before PRE.
14120 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
14121 critical edges here, after loop init.
14122 (pass_data_pre): Remove PROP_no_crit_edges flags.
14123 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
14124 for valueization of call args to avoid leaking VN_TOP.
14125 (visit_use): Assert we do not visit default defs.
14126 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
14127 Use error_mark_node to more easily detect leaking VN_TOP.
14128 All default-defs are varying, not VN_TOP. Mark them visited.
14129 (run_scc_vn): Make code match comment.
14131 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14133 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
14134 OPTION_MASK_FLOAT128_KEYWORD.
14135 (POWERPC_MASKS): Likewise.
14136 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
14137 support for the -mfloat128-type option, and make -mfloat128
14138 default on PowerPC Linux systems. Define or undefine
14139 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
14140 Define __float128 to be __ieee128 if IEEE 128-bit support is
14141 enabled, or undefine it.
14142 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
14143 Delete defining __FLOAT128_TYPE__.
14144 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
14145 -mfloat128-type option and make -mfloat128 default on PowerPC
14147 (TARGET_FLOAT128_TYPE): Likewise.
14148 (-mfloat128-type): Likewise.
14149 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14150 Delete the -mfloat128-type option and make -mfloat128 default on
14151 PowerPC Linux systems. Always use __ieee128 to be the keyword for
14152 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
14153 128-bit floating point is enabled. Change tests from using
14154 -mfloat128-type to -mfloat128.
14155 (rs6000_mangle_type): Use the correct mangling for the __float128
14156 type even if normal long double is restricted to 64-bits.
14157 (floatn_mode): Enable the _Float128 type by default on VSX Linux
14159 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
14160 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
14161 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
14163 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
14164 documentation for -mfloat128.
14166 2017-09-06 Olivier Hainque <hainque@adacore.com>
14168 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
14170 2017-09-06 Wish Wu <wishwu007@gmail.com>
14171 Jakub Jelinek <jakub@redhat.com>
14173 * asan.c (initialize_sanitizer_builtins): Add
14174 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
14175 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
14176 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
14177 BT_FN_VOID_UINT64_PTR variables.
14178 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
14179 (BT_FN_VOID_UINT16_UINT16): Likewise.
14180 (BT_FN_VOID_UINT32_UINT32): Likewise.
14181 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
14182 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
14183 (BT_FN_VOID_UINT64_PTR): Likewise.
14184 * common.opt (flag_sanitize_coverage): New variable.
14185 (fsanitize-coverage=trace-pc): Remove.
14186 (fsanitize-coverage=): Add.
14187 * flag-types.h (enum sanitize_coverage_code): New enum.
14188 * fold-const.c (fold_range_test): Disable non-short-circuit
14189 optimization if flag_sanitize_coverage.
14190 (fold_truth_andor): Likewise.
14191 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14192 * opts.c (COVERAGE_SANITIZER_OPT): Define.
14193 (coverage_sanitizer_opts): New array.
14194 (get_closest_sanitizer_option): Add OPTS argument, handle also
14195 OPT_fsanitize_coverage_.
14196 (parse_sanitizer_options): Adjusted to also handle
14197 OPT_fsanitize_coverage_.
14198 (common_handle_option): Add OPT_fsanitize_coverage_.
14199 * sancov.c (instrument_comparison, instrument_switch): New function.
14200 (sancov_pass): Add trace-cmp support.
14201 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
14202 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
14203 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
14204 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
14205 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
14206 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
14207 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
14208 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
14209 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
14210 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
14212 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14214 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14215 error. Only quit immediately if parsing is complete.
14216 (BEGIN): Initialize fatal_err and parse_done.
14217 (begin fpu, end fpu): Check number of arguments.
14218 (begin arch, end arch): Likewise.
14219 (begin cpu, end cpu): Likewise.
14220 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14221 (optalias): Likewise.
14223 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14225 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14226 * config/arm/arm-isa.h: Delete. Move definitions to ...
14227 * arm-cpus.in: ... here. Use new feature and fgroup values.
14228 * config/arm/arm.c (arm_option_override): Use lower case for feature
14230 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14231 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14232 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14234 (print_isa_bits_for): New function.
14235 (gen_isa): New function.
14236 (gen_comm_data): Use print_isa_bits_for.
14237 (define feature): New keyword.
14238 (define fgroup): New keyword.
14239 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
14240 (arm-isa.h): Add rule to generate file.
14241 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14242 case for feature bit names.
14244 2017-09-06 Richard Biener <rguenther@suse.de>
14246 * tree-ssa-pre.c (NECESSARY): Remove.
14247 (create_expression_by_pieces): Do not touch pass-local flags.
14248 (insert_into_preds_of_block): Likewise.
14249 (do_pre_regular_insertion): Likewise.
14250 (eliminate_insert): Likewise.
14251 (eliminate_dom_walker::before_dom_children): Likewise.
14252 (fini_eliminate): Do not look at inserted_exprs.
14253 (mark_operand_necessary): Remove.
14254 (remove_dead_inserted_code): Replace with simple work-list
14255 algorithm based on inserted_exprs and SSA uses.
14256 (pass_pre::execute): Re-order fini_eliminate and
14257 remove_dead_inserted_code.
14259 2017-09-06 Olivier Hainque <hainque@adacore.com>
14261 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
14262 for VxWorks 7. Adjust surrounding comments.
14264 2017-09-06 Richard Biener <rguenther@suse.de>
14266 * gimple-ssa-strength-reduction.c
14267 (find_candidates_dom_walker::before_dom_children): Also allow
14270 2017-09-06 Richard Biener <rguenther@suse.de>
14272 PR tree-optimization/82108
14273 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
14274 for gap in the non-permutation SLP case.
14276 2017-09-06 Martin Jambor <mjambor@suse.cz>
14278 PR tree-optimization/82078
14279 * tree-sra.c (sort_and_splice_var_accesses): Move call to
14280 add_access_to_work_queue...
14281 (build_accesses_from_assign): ...here.
14282 (propagate_all_subaccesses): Make sure racc is the group
14283 representative, if there is one.
14285 2017-09-06 Jakub Jelinek <jakub@redhat.com>
14287 PR middle-end/82095
14288 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
14291 2017-09-06 Richard Biener <rguenther@suse.de>
14293 * gimple-ssa-strength-reduction.c
14294 (find_candidates_dom_walker::before_doom_children): Use a
14295 type and not a mode check.
14297 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
14300 * config/arm/predicates.md (arm_general_adddi_operand): Create new
14302 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
14304 2017-09-05 Jeff Law <law@redhat.com>
14306 PR tree-optimization/64910
14307 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
14308 cases where we have 3 or more operands.
14310 2017-09-05 Jakub Jelinek <jakub@redhat.com>
14312 PR middle-end/81768
14313 * omp-low.c (lower_omp_for): Recompute tree invariant if
14314 gimple_omp_for_initial/final is ADDR_EXPR.
14316 PR middle-end/81768
14317 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
14318 into gimple val before gimplification fo the COND_EXPR.
14320 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
14322 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
14323 REGION_COPY argument.
14324 (thread_through_all_blocks): Remove unused argument to
14325 duplicate_thread_path.
14327 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14328 Alan Hayward <alan.hayward@arm.com>
14329 David Sherwood <david.sherwood@arm.com>
14331 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
14332 Take a scalar_mode rather than a machine_mode.
14333 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14334 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
14335 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14336 (aarch64_gen_adjusted_ldpstp): Likewise.
14337 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
14339 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14340 Alan Hayward <alan.hayward@arm.com>
14341 David Sherwood <david.sherwood@arm.com>
14343 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
14344 Take a scalar_int_mode instead of a machine_mode.
14345 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14346 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14347 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14348 (aarch64_simd_attr_length_rglist): Delete.
14349 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
14350 a scalar_int_mode instead of a machine_mode.
14351 (aarch64_add_offset): Likewise.
14352 (aarch64_internal_mov_immediate): Likewise
14353 (aarch64_add_constant_internal): Likewise.
14354 (aarch64_add_constant): Likewise.
14355 (aarch64_movw_imm): Likewise.
14356 (aarch64_rtx_arith_op_extract_p): Likewise.
14357 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14358 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14359 Remove assert that the mode isn't a vector.
14360 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14361 (aarch64_expand_mov_immediate): Update calls after above changes.
14362 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
14363 (aarch64_and_bitmask_imm): Check for scalar integer modes.
14364 (aarch64_move_imm): Likewise.
14365 (aarch64_can_const_movi_rtx_p): Likewise.
14366 (aarch64_strip_extend): Likewise.
14367 (aarch64_extr_rtx_p): Likewise.
14368 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
14369 a CONST_INT when the mode parameter is VOIDmode.
14370 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
14372 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14374 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
14375 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14376 (bitwise_type_for_mode): Update accordingly.
14378 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14380 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
14381 * stor-layout.c (mode_for_size_tree): Likewise.
14382 (mode_for_array): Update accordingly.
14383 (layout_decl): Likewise.
14384 (compute_record_mode): Likewise. Only set the mode once.
14386 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14388 * target.def (get_mask_mode): Change return type to opt_mode.
14390 * doc/tm.texi: Regenerate.
14391 * targhooks.h (default_get_mask_mode): Return an opt_mode.
14392 * targhooks.c (default_get_mask_mode): Likewise.
14393 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14394 * optabs-query.c (can_vec_mask_load_store_p): Update use of
14395 targetm.get_mask_mode.
14396 * tree.c (build_truth_vector_type): Likewise.
14398 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14400 * machmode.h (mode_for_vector): Return an opt_mode.
14401 * stor-layout.c (mode_for_vector): Likewise.
14402 (mode_for_int_vector): Update accordingly.
14403 (layout_type): Likewise.
14404 * config/i386/i386.c (emit_memmov): Likewise.
14405 (ix86_expand_set_or_movmem): Likewise.
14406 (ix86_expand_vector_init): Likewise.
14407 (ix86_get_mask_mode): Likewise.
14408 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
14410 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
14411 * expmed.c (extract_bit_field_1): Likewise.
14412 * expr.c (expand_expr_real_2): Likewise.
14413 * optabs-query.c (can_vec_perm_p): Likewise.
14414 (can_vec_mask_load_store_p): Likewise.
14415 * optabs.c (expand_vec_perm): Likewise.
14416 * targhooks.c (default_get_mask_mode): Likewise.
14417 * tree-vect-stmts.c (vectorizable_store): Likewise.
14418 (vectorizable_load): Likewise.
14419 (get_vectype_for_scalar_type_and_size): Likewise.
14421 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14423 * machmode.h (mode_for_int_vector): New function.
14424 * stor-layout.c (mode_for_int_vector): Likewise.
14425 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
14426 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
14427 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14428 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14429 (s390_expand_vcond): Likewise.
14431 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14433 * machmode.h (opt_machine_mode): New type.
14434 (opt_mode<T>): Allow construction from anything that can be
14436 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
14437 (mode_for_size): Return an opt_machine_mode.
14438 * stor-layout.c (mode_for_size): Likewise.
14439 (mode_for_size_tree): Update call accordingly.
14440 (bitwise_mode_for_mode): Likewise.
14441 (make_fract_type): Likewise.
14442 (make_accum_type): Likewise.
14443 * caller-save.c (replace_reg_with_saved_mem): Update call
14445 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14446 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14447 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14448 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14449 * expmed.c (extract_bit_field_1): Likewise.
14450 * reload.c (get_secondary_mem): Likewise.
14451 * varasm.c (assemble_integer): Likewise.
14452 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
14455 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14457 * machmode.h (decimal_float_mode_for_size): New function.
14458 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
14459 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
14460 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
14461 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
14462 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
14463 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
14465 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14467 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
14468 (get_builtin_sync_mode): Likewise.
14469 (expand_ifn_atomic_compare_exchange): Likewise.
14470 (expand_builtin_atomic_clear): Likewise.
14471 (expand_builtin_atomic_test_and_set): Likewise.
14472 (fold_builtin_atomic_always_lock_free): Likewise.
14473 * calls.c (compute_argument_addresses): Likewise.
14474 (emit_library_call_value_1): Likewise.
14475 (store_one_arg): Likewise.
14476 * combine.c (combine_instructions): Likewise.
14477 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
14478 * config/arm/arm.c (arm_function_value): Likewise.
14479 (aapcs_allocate_return_reg): Likewise.
14480 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
14481 * config/i386/i386.c (construct_container): Likewise.
14482 (ix86_gimplify_va_arg): Likewise.
14483 (ix86_expand_sse_cmp): Likewise.
14484 (emit_memmov): Likewise.
14485 (emit_memset): Likewise.
14486 (expand_small_movmem_or_setmem): Likewise.
14487 (ix86_expand_pextr): Likewise.
14488 (ix86_expand_pinsr): Likewise.
14489 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
14490 * config/microblaze/microblaze.c (microblaze_block_move_straight):
14492 * config/mips/mips.c (mips_function_value_1) Likewise.
14493 (mips_block_move_straight): Likewise.
14494 (mips_expand_ins_as_unaligned_store): Likewise.
14495 * config/powerpcspe/powerpcspe.c
14496 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14497 (rs6000_darwin64_record_arg_flush): Likewise.
14498 * config/rs6000/rs6000.c
14499 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14500 (rs6000_darwin64_record_arg_flush): Likewise.
14501 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
14502 (sparc_function_value_1): Likewise.
14503 * config/spu/spu.c (adjust_operand): Likewise.
14504 (spu_emit_branch_or_set): Likewise.
14505 (arith_immediate_p): Likewise.
14506 * emit-rtl.c (gen_lowpart_common): Likewise.
14507 * expr.c (expand_expr_real_1): Likewise.
14508 * function.c (assign_parm_setup_block): Likewise.
14509 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
14510 * reload1.c (alter_reg): Likewise.
14511 * stor-layout.c (mode_for_vector): Likewise.
14512 (layout_type): Likewise.
14514 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14516 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
14517 (spu_convert_move): Likewise.
14518 * lower-subreg.c (resolve_simple_move): Likewise.
14520 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14523 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
14524 define_insn to a define_expand.
14525 (altivec_vsum2sws_direct): New define_insn.
14526 (altivec_vsumsws): Convert from a define_insn to a define_expand.
14528 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14530 * config/arm/arm.c (arm_option_params_internal): Improve setting of
14533 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
14538 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
14539 realign stack if stack alignment needed is less than incoming
14542 2017-09-05 Marek Polacek <polacek@redhat.com>
14545 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
14548 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14550 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
14552 2017-09-05 Richard Biener <rguenther@suse.de>
14554 PR tree-optimization/82084
14555 * fold-const.c (can_native_encode_string_p): Handle wide characters.
14557 2017-09-05 Richard Biener <rguenther@suse.de>
14559 PR tree-optimization/82102
14560 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
14562 2017-09-05 Martin Liska <mliska@suse.cz>
14564 PR tree-optimization/82032
14565 * tree-cfg.c (generate_range_test): New function.
14566 * tree-cfg.h (generate_range_test): Declared here.
14567 * tree-cfgcleanup.c (convert_single_case_switch): New function.
14568 (cleanup_control_expr_graph): Use it.
14569 * tree-switch-conversion.c (try_switch_expansion): Remove
14571 (emit_case_nodes): Use generate_range_test.
14573 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14576 * config/i386/i386.md (*<btsc><mode>_mask): Add
14577 TARGET_USE_BT to insn constraint.
14578 (*btr<mode>_mask): Ditto.
14580 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
14582 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
14583 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
14585 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
14588 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
14589 TARGET_NEON and TARGET_IWMMXT.
14590 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
14591 TARGET_NEON and TARGET_IWMMXT.
14592 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
14594 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14596 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
14597 (ix86_rewrite_tls_address): Ditto.
14598 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
14599 (ix86_rewrite_tls_address_1): Ditto.
14600 (ix86_rewrite_tls_address): Ditto.
14601 * config/i386/predicates.md (tls_address_pattern): New predicate.
14602 * config/i386/i386.md (TLS address splitter): New splitter.
14604 2017-09-04 Richard Biener <rguenther@suse.de>
14606 PR tree-optimization/82084
14607 * fold-const.h (can_native_encode_string_p): Declare.
14608 * fold-const.c (can_native_encode_string_p): Factor out from ...
14609 (native_encode_string): ... here.
14610 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
14611 vectorizing stores from constants we later cannot handle.
14613 2017-09-04 Marek Polacek <polacek@redhat.com>
14616 * doc/invoke.texi: Update -Wtautological-compare documentation.
14618 2017-09-04 Jeff Law <law@redhat.com>
14620 PR tree-optimization/64910
14621 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
14622 swap the first and last operand if the last is a constant.
14624 2017-09-04 Marek Polacek <polacek@redhat.com>
14627 * convert.c (do_narrow): When sanitizing signed integer overflows,
14628 bail out for signed types.
14629 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
14631 2017-09-04 Richard Biener <rguenther@suse.de>
14633 PR tree-optimization/82060
14634 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14635 Move devirtualization after stmt folding and before EH/AB/noreturn
14636 cleanup to get the stmt refs canonicalized. Use a bool instead
14637 of gimple_modified_p since that doesn't work for NOPs. Schedule
14638 NOPs generated by folding for removal.
14640 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14641 Alan Hayward <alan.hayward@arm.com>
14642 David Sherwood <david.sherwood@arm.com>
14644 * coretypes.h (pad_direction): New enum.
14645 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
14646 (FUNCTION_ARG_PADDING): Likewise.
14647 * target.def (function_arg_padding): New hook.
14648 * targhooks.h (default_function_arg_padding): Declare.
14649 * targhooks.c (default_function_arg_padding): New function.
14650 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
14651 (TARGET_FUNCTION_ARG_PADDING): ...this.
14652 * doc/tm.texi: Regenerate.
14653 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
14654 instead of direction.
14655 (compute_argument_addresses): Likewise.
14656 (load_register_parameters): Likewise.
14657 (emit_library_call_value_1): Likewise.
14658 (store_one_arg): Use targetm.calls.function_arg_padding instead
14659 of FUNCTION_ARG_PADDING.
14660 (must_pass_in_stack_var_size_or_pad): Likewise.
14661 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
14662 (emit_group_store): Likewise.
14663 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
14664 instead of FUNCTION_ARG_PADDING.
14665 (emit_push_insn): Likewise, and propagate enum change throughout
14667 * function.h (direction): Delete.
14668 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
14670 * function.c (assign_parm_find_stack_rtl): Likewise.
14671 (assign_parm_setup_block_p): Likewise.
14672 (assign_parm_setup_block): Likewise.
14673 (gimplify_parameters): Likewise.
14674 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
14675 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
14677 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
14678 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14679 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
14680 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
14681 (aarch64_function_arg_padding): ...this new function.
14682 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
14683 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14684 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
14685 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14686 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
14687 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14688 (arm_pad_arg_upward): Replace with...
14689 (arm_function_arg_padding): ...this new function.
14690 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
14692 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
14693 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
14694 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14695 (ia64_hpux_function_arg_padding): Replace with...
14696 (ia64_function_arg_padding): ...this new function. Use pad_direction
14697 instead of direction. Check for TARGET_HPUX.
14698 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
14699 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14700 (iq2000_function_arg_padding): New function.
14701 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
14702 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
14703 (mips_function_arg_padding): ...this new function.
14704 (mips_pad_reg_upward): Update accordingly.
14705 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14706 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
14707 targetm.calls.function_arg_padding.
14708 (FUNCTION_ARG_PADDING): Delete.
14709 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14710 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
14711 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14712 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
14713 (nios2_block_reg_padding): Return pad_direction instead of direction.
14714 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
14715 instead of direction.
14716 (nios2_function_arg_padding): Likewise. Make static.
14717 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14718 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
14719 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
14720 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
14721 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14722 (pa_function_arg_padding): Make static. Return pad_direction instead
14724 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
14725 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14726 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
14727 instead of direction. Use targetm.calls.function_arg_padding.
14728 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
14729 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14730 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
14731 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
14732 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
14734 (function_arg_padding): Rename to...
14735 (rs6000_function_arg_padding): ...this. Make static. Return
14736 pad_direction instead of direction.
14737 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14738 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
14739 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14740 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
14741 instead of direction. Use targetm.calls.function_arg_padding.
14742 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
14743 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14744 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
14745 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
14746 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14747 (function_arg_padding): Rename to...
14748 (rs6000_function_arg_padding): ...this. Make static. Return
14749 pad_direction instead of direction.
14750 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14751 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
14752 * config/s390/s390.c (s390_function_arg_padding): New function.
14753 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14754 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
14755 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
14756 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14757 (function_arg_padding): Rename to...
14758 (sparc_function_arg_padding): ...this. Make static. Return
14759 pad_direction instead of direction.
14760 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
14761 * config/spu/spu.c (spu_function_arg_padding): New function.
14762 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14763 * system.h (FUNCTION_ARG_PADDING): Poison.
14765 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14766 Alan Hayward <alan.hayward@arm.com>
14767 David Sherwood <david.sherwood@arm.com>
14769 * target.def (modes_tieable_p): New hook.
14770 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
14771 (TARGET_MODES_TIEABLE_P): ...this.
14772 * doc/tm.texi.in: Regenerate.
14773 * hooks.h (hook_bool_mode_mode_true): Declare.
14774 * hooks.c (hook_bool_mode_mode_true): New function.
14775 * combine.c (subst): Use targetm.modes_tieable_p instead of
14777 * dse.c (find_shift_sequence): Likewise.
14778 * expmed.c (extract_low_bits): Likewise.
14779 * lower-subreg.c: Include target.h.
14780 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
14782 * rtlanal.c (rtx_cost): Likewise.
14783 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
14784 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
14785 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
14786 (TARGET_MODES_TIEABLE_P): Redefine.
14787 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
14788 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
14789 (TARGET_MODES_TIEABLE_P): Redefine.
14790 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
14791 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
14792 (arc_modes_tieable_p): New function.
14793 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
14794 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
14795 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
14796 (arm_modes_tieable_p): Make static.
14797 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
14798 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
14799 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
14800 (TARGET_MODES_TIEABLE_P): Redefine.
14801 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
14802 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
14803 (TARGET_MODES_TIEABLE_P): Redefine.
14804 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
14805 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
14806 (cr16_modes_tieable_p): New function.
14807 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
14808 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
14809 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
14810 (TRULY_NOOP_TRUNCATION): Update comment.
14811 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
14812 (TRULY_NOOP_TRUNCATION): Update comment.
14813 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
14814 (frv_modes_tieable_p): New function.
14815 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
14816 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
14817 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
14818 (TARGET_MODES_TIEABLE_P): Redefine.
14819 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
14820 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
14821 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
14822 (TARGET_MODES_TIEABLE_P): Redefine.
14823 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
14824 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
14825 (ia64_modes_tieable_p): New function.
14826 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
14827 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
14828 (iq2000_modes_tieable_p): New function.
14829 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
14830 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
14831 (lm32_modes_tieable_p): New function.
14832 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
14833 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
14834 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
14835 (TARGET_MODES_TIEABLE_P): Redefine.
14836 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
14837 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
14838 (m32r_modes_tieable_p): New function.
14839 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
14840 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
14841 (m68k_modes_tieable_p): New function.
14842 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
14843 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
14844 (mcore_modes_tieable_p): New function.
14845 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
14846 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
14848 (TARGET_MODES_TIEABLE_P): Redefine.
14849 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
14850 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
14851 * config/mips/mips.c (mips_modes_tieable_p): Make static.
14852 (TARGET_MODES_TIEABLE_P): Redefine.
14853 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
14854 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
14855 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
14856 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
14857 (mn10300_modes_tieable_p): ...this and make static.
14858 (TARGET_MODES_TIEABLE_P): Redefine.
14859 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
14860 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
14861 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
14862 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
14863 (msp430_modes_tieable_p): Make static.
14864 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
14865 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
14866 (TARGET_MODES_TIEABLE_P): Redefine.
14867 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
14868 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
14869 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
14870 (TARGET_MODES_TIEABLE_P): Redefine.
14871 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
14872 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
14873 * config/pa/pa.c (pa_modes_tieable_p): Make static.
14874 (TARGET_MODES_TIEABLE_P): Redefine.
14875 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
14876 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
14877 (pdp11_modes_tieable_p): New function.
14878 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
14879 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
14880 (rs6000_modes_tieable_p): New function.
14881 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14882 * config/powerpcspe/powerpcspe.md: Update comment.
14883 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
14884 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
14885 (TARGET_MODES_TIEABLE_P): Redefine.
14886 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
14887 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
14888 (rl78_modes_tieable_p): New function.
14889 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
14890 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
14891 (rs6000_modes_tieable_p): New function.
14892 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14893 * config/rs6000/rs6000.md: Update comment.
14894 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
14895 * config/rx/rx.c (rx_modes_tieable_p): New function.
14896 (TARGET_MODES_TIEABLE_P): Redefine.
14897 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
14898 * config/s390/s390.c (s390_modes_tieable_p): New function.
14899 (TARGET_MODES_TIEABLE_P): Redefine.
14900 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
14901 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
14902 (sh_modes_tieable_p): New function.
14903 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
14904 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
14905 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
14906 (sparc_modes_tieable_p): Make static.
14907 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
14908 * config/spu/spu.c (spu_modes_tieable_p): New function.
14909 (TARGET_MODES_TIEABLE_P): Redefine.
14910 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
14911 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
14912 (TARGET_MODES_TIEABLE_P): Redefine.
14913 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
14914 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
14915 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
14916 * config/v850/v850.c (v850_modes_tieable_p): New function.
14917 (TARGET_MODES_TIEABLE_P): Redefine.
14918 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
14919 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
14920 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
14921 (visium_modes_tieable_p): New function.
14922 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
14923 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
14924 (xtensa_modes_tieable_p): New function.
14925 * system.h (MODES_TIEABLE_P): Poison.
14927 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14928 Alan Hayward <alan.hayward@arm.com>
14929 David Sherwood <david.sherwood@arm.com>
14931 * target.def (hard_regno_mode_ok): New hook.
14932 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14933 (TARGET_HARD_REGNO_MODE_OK): ...this.
14934 * doc/tm.texi.in: Regenerate.
14935 * hooks.h (hook_bool_uint_mode_true): Declare.
14936 * hooks.c (hook_bool_uint_mode_true): New function.
14937 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14938 HARD_REGNO_MODE_OK.
14939 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14940 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14941 instead of HARD_REGNO_MODE_OK.
14942 * caller-save.c: Include target.h.
14943 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14944 HARD_REGNO_MODE_OK.
14945 * combine.c (can_combine_p): Likewise.
14946 (combinable_i3pat): Likewise.
14947 (can_change_dest_mode): Likewise.
14948 * expr.c (init_expr_target): Likewise.
14949 (convert_move): Likewise.
14950 (convert_modes): Likewise.
14951 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14952 (setup_prohibited_mode_move_regs): Likewise.
14953 * ira.h (target_ira): Likewise.
14954 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14955 * lra-constraints.c (process_alt_operands): Likewise.
14956 (split_reg): Likewise.
14957 * recog.c (peep2_find_free_register): Likewise.
14958 * ree.c (combine_reaching_defs): Likewise.
14959 * regcprop.c (maybe_mode_change): Likewise.
14960 * reginfo.c (init_reg_sets_1): Likewise.
14961 (choose_hard_reg_mode): Likewise.
14962 (simplifiable_subregs): Likewise.
14963 * regrename.c (check_new_reg_p): Likewise.
14964 * reload.c (find_valid_class): Likewise.
14965 (find_valid_class_1): Likewise.
14966 (reload_inner_reg_of_subreg): Likewise.
14967 (push_reload): Likewise.
14968 (combine_reloads): Likewise.
14969 (find_dummy_reload): Likewise.
14970 (find_reloads): Likewise.
14971 * reload1.c (find_reg): Likewise.
14972 (set_reload_reg): Likewise.
14973 (allocate_reload_reg): Likewise.
14974 (choose_reload_regs): Likewise.
14975 (reload_adjust_reg_for_temp): Likewise.
14976 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14977 (simplify_subreg_regno): Likewise.
14978 * sel-sched.c (init_regs_for_mode): Likewise.
14979 * varasm.c (make_decl_rtl): Likewise.
14980 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14981 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14982 HARD_REGNO_MODE_OK.
14983 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14984 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14985 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14986 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14987 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14988 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14989 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14990 (arc_mode_class): Delete.
14991 (HARD_REGNO_MODE_OK): Delete.
14992 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14993 (arc_hard_regno_mode_ok): Rename old array to...
14994 (arc_hard_regno_mode_ok_modes): ...this.
14995 (arc_conditional_register_usage): Update accordingly.
14996 (arc_mode_class): Make static.
14997 (arc_hard_regno_mode_ok): New function.
14998 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14999 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
15000 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15001 (arm_hard_regno_mode_ok): Make static.
15002 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
15003 HARD_REGNO_MODE_OK.
15004 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
15005 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
15006 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
15008 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15009 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
15010 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
15011 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
15012 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
15013 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15014 * config/bfin/predicates.md (valid_reg_operand): Use
15015 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
15016 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
15017 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
15018 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15019 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
15020 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
15021 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15022 (cr16_hard_regno_mode_ok): Make static and return a bool.
15023 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
15024 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15025 (cris_hard_regno_mode_ok): New function.
15026 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
15027 (epiphany_mode_class): Delete.
15028 (HARD_REGNO_MODE_OK): Delete.
15029 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
15030 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15031 (hard_regno_mode_ok): Rename to...
15032 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
15033 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
15034 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
15035 HARD_REGNO_MODE_OK.
15036 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
15037 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
15038 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15039 (frv_hard_regno_mode_ok): Make static and return a bool.
15040 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
15041 HARD_REGNO_MODE_OK.
15042 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
15043 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
15044 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
15045 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
15047 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15048 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
15049 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
15050 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
15052 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15053 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
15054 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15055 (ia64_hard_regno_mode_ok): New function.
15056 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
15057 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15058 (iq2000_hard_regno_mode_ok): New function.
15059 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
15060 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15061 (lm32_hard_regno_mode_ok): New function.
15062 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
15063 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
15064 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
15065 instead of HARD_REGNO_MODE_OK.
15066 (m32c_hard_regno_ok): Rename to...
15067 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
15068 (m32c_cannot_change_mode_class): Update accordingly.
15069 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15070 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
15071 (m32r_mode_class): Delete.
15072 (HARD_REGNO_MODE_OK): Delete.
15073 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15074 (m32r_hard_regno_mode_ok): Rename to...
15075 (m32r_hard_regno_modes): ...this.
15076 (m32r_mode_class): Make static.
15077 (m32r_hard_regno_mode_ok): New function.
15078 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
15079 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
15080 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15081 (m68k_hard_regno_mode_ok): Make static.
15082 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
15083 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15084 (mcore_hard_regno_mode_ok): New function.
15085 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
15086 (HARD_REGNO_MODE_OK): Delete.
15087 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
15089 (microblaze_hard_regno_mode_ok_p): ...this and make static.
15090 (microblaze_hard_regno_mode_ok): New function.
15091 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15092 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
15093 (mips_hard_regno_mode_ok): Delete.
15094 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
15095 (mips_hard_regno_mode_ok_p): ...this and make static.
15096 (mips_hard_regno_mode_ok_p): Rename to...
15097 (mips_hard_regno_mode_ok_uncached): ...this.
15098 (mips_hard_regno_mode_ok): New function.
15099 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
15100 of HARD_REGNO_MODE_OK.
15101 (mips_option_override): Update after above name changes.
15102 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15103 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
15104 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
15105 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
15106 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
15107 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15108 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
15109 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
15110 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
15111 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15112 (msp430_hard_regno_mode_ok): Make static and return a bool.
15113 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
15114 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
15115 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
15117 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15118 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
15119 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
15120 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
15121 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
15122 (PA_HARD_REGNO_MODE_OK): ...this
15123 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
15124 (PA_HARD_REGNO_MODE_OK): ...this.
15125 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15126 (pa_hard_regno_mode_ok): New function.
15127 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
15128 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15129 (pdp11_hard_regno_mode_ok): New function.
15130 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
15131 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
15133 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
15135 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15136 (rs6000_hard_regno_mode_ok): Rename to...
15137 (rs6000_hard_regno_mode_ok_uncached): ...this.
15138 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15139 (rs6000_hard_regno_mode_ok): New function.
15140 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
15141 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
15142 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
15143 (riscv_hard_regno_mode_ok): ...this and make static.
15144 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15145 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
15146 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
15147 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15148 (rl78_hard_regno_mode_ok): Make static and return bool.
15149 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
15150 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
15152 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
15153 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15154 (rs6000_hard_regno_mode_ok): Rename to...
15155 (rs6000_hard_regno_mode_ok_uncached): ...this.
15156 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15157 (rs6000_hard_regno_mode_ok): New function.
15158 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
15159 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
15160 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15161 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
15162 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
15163 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
15164 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15165 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
15166 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
15167 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15168 (sh_hard_regno_mode_ok): Make static.
15169 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
15170 instead of HARD_REGNO_MODE_OK.
15171 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
15172 (sparc_mode_class): Delete.
15173 (HARD_REGNO_MODE_OK): Delete.
15174 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15175 (hard_regno_mode_classes): Make static.
15176 (sparc_mode_class): Likewise.
15177 (sparc_hard_regno_mode_ok): New function.
15178 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
15179 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
15180 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
15182 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15183 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
15184 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
15185 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
15186 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
15187 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15188 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
15189 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
15190 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15191 (visium_hard_regno_mode_ok): New function.
15192 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
15193 instead of HARD_REGNO_MODE_OK.
15194 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
15195 (HARD_REGNO_MODE_OK): Delete.
15196 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
15197 (xtensa_hard_regno_mode_ok_p): ...this and make static.
15198 (xtensa_option_override): Update accordingly.
15199 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15200 (xtensa_hard_regno_mode_ok): New function.
15201 * system.h (HARD_REGNO_MODE_OK): Poison.
15203 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15204 Alan Hayward <alan.hayward@arm.com>
15205 David Sherwood <david.sherwood@arm.com>
15207 * target.def (hard_regno_call_part_clobbered): New hook.
15208 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
15209 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
15210 * doc/tm.texi: Regenerate.
15211 * hooks.h (hook_bool_uint_mode_false): Declare.
15212 * hooks.c (hook_bool_uint_mode_false): New function.
15213 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15214 * cselib.c (cselib_process_insn): Use
15215 targetm.hard_regno_call_part_clobbered instead of
15216 HARD_REGNO_CALL_PART_CLOBBERED.
15217 * ira-conflicts.c (ira_build_conflicts): Likewise.
15218 * ira-costs.c (ira_tune_allocno_costs): Likewise.
15219 * lra-constraints.c (need_for_call_save_p): Likewise.
15220 * lra-lives.c: Include target.h.
15221 (check_pseudos_live_through_calls): Use
15222 targetm.hard_regno_call_part_clobbered instead of
15223 HARD_REGNO_CALL_PART_CLOBBERED.
15224 * regcprop.c: Include target.h.
15225 (copyprop_hardreg_forward_1): Use
15226 targetm.hard_regno_call_part_clobbered instead of
15227 HARD_REGNO_CALL_PART_CLOBBERED.
15228 * reginfo.c (choose_hard_reg_mode): Likewise.
15229 * regrename.c (check_new_reg_p): Likewise.
15230 * reload.c (find_equiv_reg): Likewise.
15231 * reload1.c (emit_reload_insns): Likewise.
15232 * sched-deps.c (deps_analyze_insn): Likewise.
15233 * sel-sched.c (init_regs_for_mode): Likewise.
15234 (mark_unavailable_hard_regs): Likewise.
15235 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
15236 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15237 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
15239 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15240 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15241 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
15243 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
15245 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15246 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15247 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
15249 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15250 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15251 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
15253 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15254 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
15256 * config/powerpcspe/powerpcspe.c
15257 (rs6000_hard_regno_call_part_clobbered): New function.
15258 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15259 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15260 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
15262 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15263 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15264 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
15266 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15267 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15268 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
15270 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15271 Alan Hayward <alan.hayward@arm.com>
15272 David Sherwood <david.sherwood@arm.com>
15274 * rtl.h (subreg_memory_offset): Declare.
15275 * emit-rtl.c (subreg_memory_offset): New function.
15276 * expmed.c (store_bit_field_1): Use it.
15277 * expr.c (undefined_operand_subword_p): Likewise.
15278 * simplify-rtx.c (simplify_subreg): Likewise.
15280 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
15282 PR rtl-optimization/57448
15285 * optabs.c (expand_atomic_load): Place compiler memory barriers if
15286 using atomic_load pattern.
15287 (expand_atomic_store): Likewise.
15289 2017-09-04 Jakub Jelinek <jakub@redhat.com>
15292 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
15293 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
15294 handling. Use replace_call_with_value with NULL instead of
15295 gsi_replace, unlink_stmt_vdef and release_defs.
15297 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
15300 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
15302 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15305 * rtl.h (emit_library_call_value_1): Declare.
15306 (emit_library_call): Replace declaration with a series of overloads.
15307 Remove the parameter count argument.
15308 (emit_library_call_value): Likewise.
15309 * calls.c (emit_library_call_value_1): Make global. Replace varargs
15310 with an "rtx_mode_t *".
15311 (emit_library_call_value): Delete.
15312 (emit_library_call): Likewise.
15313 * asan.c (asan_emit_stack_protection): Update calls accordingly.
15314 (asan_emit_allocas_unpoison): Likewise.
15315 * builtins.c (expand_builtin_powi): Likewise.
15316 (expand_asan_emit_allocas_unpoison): Likewise.
15317 * cfgexpand.c (expand_main_function): Likewise.
15318 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
15319 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
15320 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
15321 * config/arm/arm.c (arm_trampoline_init): Likewise.
15322 (arm_call_tls_get_addr): Likewise.
15323 (arm_expand_divmod_libfunc): Likewise.
15324 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
15325 (smulsi3_highpart): Likewise.
15326 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
15327 (c6x_expand_compare): Likewise.
15328 (c6x_expand_movmem): Likewise.
15329 * config/frv/frv.c (frv_trampoline_init): Likewise.
15330 * config/i386/i386.c (ix86_trampoline_init): Likewise.
15331 (ix86_expand_divmod_libfunc): Likewise.
15332 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
15333 (ia64_expand_compare): Likewise.
15334 (ia64_profile_hook): Likewise.
15335 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
15336 (nonlocal_goto): Likewise.
15337 (restore_stack_nonlocal): Likewise.
15338 * config/m32r/m32r.c (block_move_call): Likewise.
15339 (m32r_trampoline_init): Likewise.
15340 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
15341 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
15342 (m68k_call_m68k_read_tp): Likewise.
15343 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
15344 (microblaze_expand_divide): Likewise.
15345 * config/mips/mips.h (mips_args): Likewise.
15346 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
15347 (MIPS_ICACHE_SYNC): Likewise.
15348 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
15349 (nios2_trampoline_init): Likewise.
15350 * config/pa/pa.c (hppa_tls_call): Likewise.
15351 (pa_trampoline_init): Likewise.
15352 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
15353 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
15354 (expand_strn_compare): Likewise.
15355 (rs6000_generate_compare): Likewise.
15356 (rs6000_expand_float128_convert): Likewise.
15357 (output_profile_hook): Likewise.
15358 (rs6000_trampoline_init): Likewise.
15359 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
15360 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
15361 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
15362 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
15363 (rs6000_generate_compare): Likewise.
15364 (rs6000_expand_float128_convert): Likewise.
15365 (output_profile_hook): Likewise.
15366 (rs6000_trampoline_init): Likewise.
15367 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
15368 * config/sh/sh.c (sh_trampoline_init): Likewise.
15369 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
15370 (sparc_emit_float_lib_cmp): Likewise.
15371 (sparc32_initialize_trampoline): Likewise.
15372 (sparc64_initialize_trampoline): Likewise.
15373 (sparc_profile_hook): Likewise.
15374 * config/spu/spu.c (ea_load_store): Likewise.
15375 * config/spu/spu.md (floatunssidf2): Likewise.
15376 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
15377 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
15378 * config/visium/visium.c (expand_block_move_4): Likewise.
15379 (expand_block_move_2): Likewise.
15380 (expand_block_move_1): Likewise.
15381 (expand_block_set_4): Likewise.
15382 (expand_block_set_2): Likewise.
15383 (expand_block_set_1): Likewise.
15384 (visium_trampoline_init): Likewise.
15385 (visium_profile_hook): Likewise.
15386 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
15387 (xtensa_setup_frame_addresses): Likewise.
15388 (xtensa_trampoline_init): Likewise.
15389 * except.c (sjlj_emit_function_enter): Likewise.
15390 (sjlj_emit_function_exit): Likewise.
15391 * explow.c (allocate_dynamic_stack_space): Likewise.
15392 (probe_stack_range): Likewise.
15393 * expr.c (convert_mode_scalar): Likewise.
15394 * optabs.c (expand_binop): Likewise.
15395 (expand_twoval_binop_libfunc): Likewise.
15396 (expand_unop): Likewise.
15397 (prepare_cmp_insn): Likewise.
15398 (prepare_float_lib_cmp): Likewise.
15399 (expand_float): Likewise.
15400 (expand_fix): Likewise.
15401 (expand_fixed_convert): Likewise.
15402 (maybe_emit_sync_lock_test_and_set): Likewise.
15403 (expand_atomic_compare_and_swap): Likewise.
15404 (expand_mem_thread_fence): Likewise.
15405 (expand_atomic_fetch_op): Likewise.
15407 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
15409 * doc/generic.texi (OpenACC): Adjust URL.
15410 * doc/invoke.texi (C Dialect Options): Ditto.
15412 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
15414 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
15415 predicate for operand 1. Add (m,<S>) constraint.
15416 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
15417 Prevent memory operand 1 with register operand 2.
15419 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
15421 PR rtl-optimization/82024
15422 * combine.c (try_combine): If the combination result is a PARALLEL,
15423 and we only need to retain the SET in there that would be placed
15424 at I2, check that we can place that at I3 instead, before doing so.
15426 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15429 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
15431 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
15432 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
15435 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
15436 Jeff Law <law@redhat.com>
15438 * varasm.c (bss_initializer_p): Do not put constants into .bss
15439 (categorize_decl_for_section): Handle bss_initializer_p returning
15440 false when DECL_INITIAL is NULL.
15442 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15445 * config/s390/s390.c (s390_can_inline_p): New function.
15447 2017-09-01 Jeff Law <law@redhat.com>
15449 PR tree-optimization/82052
15450 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15451 Always initialize the returned slot after a hash table miss
15452 when INSERT is true.
15454 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
15456 * config/s390/s390.md (mem_signal_fence): Remove.
15457 * doc/md.texi (mem_signal_fence): Remove.
15458 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
15460 * target-insns.def (mem_signal_fence): Remove.
15462 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15465 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
15468 * asan.c (create_odr_indicator): Strip name encoding from assembler
15469 name before appending it after __odr_asan_.
15471 2017-09-01 Martin Liska <mliska@suse.cz>
15473 PR tree-optimization/82059
15474 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
15475 frequency only when an edge is redirected.
15477 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15479 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
15480 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
15481 (arc_conditional_register_usage): Remove ARC600 lp_count
15483 (arc_file_start): Emit Tag_ARC_CPU_variation.
15484 (arc_can_use_doloop_p): New conditions to use ZOLs.
15485 (hwloop_fail): New function.
15486 (hwloop_optimize): Likewise.
15487 (hwloop_pattern_reg): Likewise.
15488 (arc_doloop_hooks): New struct, to be used with reorg_loops.
15489 (arc_reorg_loops): New function, calls reorg_loops.
15490 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
15491 (arc600_corereg_hazard): Remove ZOL checking, case handled by
15493 (arc_loop_hazard): Remove function, functionality moved into
15495 (arc_hazard): Remove arc_loop_hazard call.
15496 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
15497 into hwloop_optimize.
15498 (arc_label_align): Remove ZOL handling.
15499 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
15500 * config/arc/arc.md (doloop_begin): Remove pattern.
15501 (doloop_begin_i): Likewise.
15502 (doloop_end_i): Likewise.
15503 (doloop_fallback): Likewise.
15504 (doloop_fallback_m): Likewise.
15505 (doloop_end): Reimplement expand.
15506 (arc_lp): New pattern for LP instruction.
15507 (loop_end): New pattern.
15508 (loop_fail): Likewise.
15509 (decrement_and_branch_until_zero): Likewise.
15510 * config/arc/arc.opt (mlpc-width): New option.
15511 * doc/invoke.texi (mlpc-width): Document option.
15513 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15515 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
15516 (arc_ccfsm_advance): Fix checking for delay slots.
15517 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
15519 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15521 * config/arc/arc.md (movqi_insn): Add stores to save constant long
15523 (movhi_insn): Update store instruction constraint which are saving
15524 6-bit short immediates.
15525 (movsi_insn): Consider also short scaled load operations.
15526 (zero_extendhisi2_i): Use Usd constraint instead of T.
15527 (extendhisi2_i): Add q constraint.
15528 (arc_clzsi2): Add type and length attributes.
15529 (arc_ctzsi2): Likewise.
15530 * config/arc/constraints.md (Usc): Update constraint, the
15531 assembler can parse two relocations for a single instruction.
15533 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15535 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
15536 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
15538 2017-08-31 Olivier Hainque <hainque@adacore.com>
15540 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
15541 match as powerpc-wrs-vxworks*.
15543 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
15545 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
15546 register constraint for by-element operand.
15547 (aarch64_mls_elt_merge<mode>): Likewise.
15549 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15551 * config/arc/arc.c (arc_can_follow_jump): Check for short
15554 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15556 * config.gcc: Use g.opt for arc.
15557 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
15558 functionality moved to ...
15559 (legitimate_scaled_address_p): New function, ...here.
15560 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
15561 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
15562 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
15564 (arc_override_options): Handle G option.
15565 (arc_output_pic_addr_const): Correct function definition.
15566 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
15567 (arc_decl_anon_ns_mem_p): Delete.
15568 (arc_in_small_data_p): Overhaul this function to take into
15569 consideration the value given via G option.
15570 (arc_rewrite_small_data_1): Renamed and corrected old
15571 arc_rewrite_small_data function.
15572 (arc_rewrite_small_data): New function.
15573 (small_data_pattern): Don't use pic_offset_table_rtx.
15574 * config/arc/arc.h (CC1_SPEC): Recognize G option.
15575 * config/arc/simdext.md (movmisalignv2hi): Use
15576 prepare_move_operands function.
15578 (movmisalign*): Likewise.
15579 * doc/invoke.texi (ARC options): Document -G option.
15581 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15583 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
15585 * config/arc/arc.c (arc_print_operand): Output scalled address for
15586 sdata whenever is possible.
15587 (arc_in_small_data_p): Allow sdata for 64bit datum when double
15588 load/stores are available.
15589 (compact_sda_memory_operand): Check for the alignment required by
15590 code density instructions.
15591 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
15593 * config/arc/constraints.md (Usd): Update constraint.
15594 (Us0): New constraint.
15595 (Usc): Update constraint.
15597 2017-08-31 Richard Biener <rguenther@suse.de>
15599 PR middle-end/82054
15600 * dwarf2out.c (dwarf2out_early_global_decl): Process each
15601 function only once.
15603 2017-08-31 Tamar Christina <tamar.christina@arm.com>
15605 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
15606 Resize type_signature.
15608 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15609 Alan Hayward <alan.hayward@arm.com>
15610 David Sherwood <david.sherwood@arm.com>
15612 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
15613 subregs whose inner modes can be stored in GPRs.
15614 (aarch64_classify_index): Likewise.
15616 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15617 Alan Hayward <alan.hayward@arm.com>
15618 David Sherwood <david.sherwood@arm.com>
15620 * config/aarch64/iterators.md (V_cmp_result): Rename to...
15621 (V_INT_EQUIV): ...this.
15622 (v_cmp_result): Rename to...
15623 (v_int_equiv): ...this.
15624 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
15625 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
15626 (copysign<mode>3): Likewise.
15627 (aarch64_simd_bsl<mode>_internal): Likewise.
15628 (aarch64_simd_bsl<mode>): Likewise.
15629 (vec_cmp<mode><mode>): Likewise.
15630 (vcond<mode><mode>): Likewise.
15631 (vcond<v_cmp_mixed><mode>): Likewise.
15632 (vcondu<mode><v_cmp_mixed>): Likewise.
15633 (aarch64_cm<optab><mode>): Likewise.
15634 (aarch64_cmtst<mode>): Likewise.
15635 (aarch64_fac<optab><mode>): Likewise.
15636 (vec_perm_const<mode>): Likewise.
15637 (vcond_mask_<mode><v_cmp_result>): Rename to...
15638 (vcond_mask_<mode><v_int_equiv>): ...this.
15639 (vec_cmp<mode><v_cmp_result>): Rename to...
15640 (vec_cmp<mode><v_int_equiv>): ...this.
15642 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15643 Alan Hayward <alan.hayward@arm.com>
15644 David Sherwood <david.sherwood@arm.com>
15646 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
15648 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
15649 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
15650 (UNSPEC_LD4_DREG): New unspecs.
15651 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
15652 (aarch64_ld2<mode>_dreg_be): Replace with...
15653 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
15655 (aarch64_ld3<mode>_dreg_le)
15656 (aarch64_ld3<mode>_dreg_be): Replace with...
15657 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
15659 (aarch64_ld4<mode>_dreg_le)
15660 (aarch64_ld4<mode>_dreg_be): Replace with...
15661 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
15664 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15666 PR tree-optimization/81987
15667 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
15668 insert an initializer in a location not dominated by the stride
15671 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15673 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
15674 on the entire header of the finally block in the fallthru case.
15676 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15678 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
15680 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
15682 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
15683 rs6000_emit_move_from_cr and call renamed function.
15684 (rs6000_emit_prologue): Call renamed functions.
15685 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
15686 movesi_from_cr, remove volatile CRs.
15688 2017-08-30 Jon Beniston <jon@beniston.com>
15689 Richard Biener <rguenther@suse.de>
15691 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
15692 of VECTOR_MODE_P check.
15693 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
15694 element vector types.
15696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15698 * df.h (df_read_modify_subreg_p): Remove in favor of...
15699 * rtl.h (read_modify_subreg_p): ...this new function. Take a
15700 const_rtx instead of an rtx.
15701 * cprop.c (local_cprop_find_used_regs): Update accordingly.
15702 * df-problems.c (df_word_lr_mark_ref): Likewise.
15703 * ira-lives.c (mark_pseudo_reg_live): Likewise.
15704 (mark_pseudo_reg_dead): Likewise.
15705 (mark_ref_dead): Likewise.
15706 * reginfo.c (init_subregs_of_mode): Likewise.
15707 * sched-deps.c (sched_analyze_1): Likewise.
15708 * df-scan.c (df_def_record_1): Likewise.
15709 (df_uses_record): Likewise.
15710 (df_read_modify_subreg_p): Remove in favor of...
15711 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
15712 const_rtx instead of an rtx.
15714 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15715 Alan Hayward <alan.hayward@arm.com>
15716 David Sherwood <david.sherwood@arm.com>
15718 * rtl.h (partial_subreg_p): New function.
15719 * caller-save.c (save_call_clobbered_regs): Use it.
15720 * calls.c (expand_call): Likewise.
15721 * combine.c (combinable_i3pat): Likewise.
15722 (simplify_set): Likewise.
15723 (make_extraction): Likewise.
15724 (make_compound_operation_int): Likewise.
15725 (gen_lowpart_or_truncate): Likewise.
15726 (force_to_mode): Likewise.
15727 (make_field_assignment): Likewise.
15728 (reg_truncated_to_mode): Likewise.
15729 (record_truncated_value): Likewise.
15730 (move_deaths): Likewise.
15731 * cse.c (record_jump_cond): Likewise.
15732 (cse_insn): Likewise.
15733 * cselib.c (cselib_lookup_1): Likewise.
15734 * expmed.c (extract_bit_field_using_extv): Likewise.
15735 * function.c (assign_parm_setup_reg): Likewise.
15736 * ifcvt.c (noce_convert_multiple_sets): Likewise.
15737 * ira-build.c (create_insn_allocnos): Likewise.
15738 * lra-coalesce.c (merge_pseudos): Likewise.
15739 * lra-constraints.c (match_reload): Likewise.
15740 (simplify_operand_subreg): Likewise.
15741 (curr_insn_transform): Likewise.
15742 * lra-lives.c (process_bb_lives): Likewise.
15743 * lra.c (new_insn_reg): Likewise.
15744 (lra_substitute_pseudo): Likewise.
15745 * regcprop.c (mode_change_ok): Likewise.
15746 (maybe_mode_change): Likewise.
15747 (copyprop_hardreg_forward_1): Likewise.
15748 * reload.c (push_reload): Likewise.
15749 (find_reloads): Likewise.
15750 (find_reloads_subreg_address): Likewise.
15751 * reload1.c (alter_reg): Likewise.
15752 (eliminate_regs_1): Likewise.
15753 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15755 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
15757 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
15758 back to if statements, including unpack.
15760 2017-08-30 Martin Liska <mliska@suse.cz>
15762 PR inline-asm/82001
15763 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
15765 (func_checker::compare_asm_inputs_outputs): ... this function.
15766 (func_checker::compare_gimple_asm): Use the function to compare
15767 also ASM constrains.
15768 * ipa-icf-gimple.h: Rename the function.
15770 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15771 Alan Hayward <alan.hayward@arm.com>
15772 David Sherwood <david.sherwood@arm.com>
15774 * coretypes.h (complex_mode): New type.
15775 * gdbhooks.py (build_pretty_printer): Handle it.
15776 * machmode.h (complex_mode): New class.
15777 (complex_mode::includes_p): New function.
15778 (is_complex_int_mode): Likewise.
15779 (is_complex_float_mode): Likewise.
15780 * genmodes.c (get_mode_class): Handle complex mode classes.
15781 * function.c (expand_function_end): Use is_complex_int_mode.
15783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15784 Alan Hayward <alan.hayward@arm.com>
15785 David Sherwood <david.sherwood@arm.com>
15787 * coretypes.h (scalar_mode_pod): New typedef.
15788 * gdbhooks.py (build_pretty_printer): Handle it.
15789 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
15790 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
15791 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
15792 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
15793 as_a <scalar_mode>.
15795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15796 Alan Hayward <alan.hayward@arm.com>
15797 David Sherwood <david.sherwood@arm.com>
15799 * machmode.h (mode_for_vector): Take a scalar_mode instead
15801 * stor-layout.c (mode_for_vector): Likewise.
15802 * explow.c (promote_mode): Use as_a <scalar_mode>.
15803 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
15805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15806 Alan Hayward <alan.hayward@arm.com>
15807 David Sherwood <david.sherwood@arm.com>
15809 * target.def (preferred_simd_mode): Take a scalar_mode
15810 instead of a machine_mode.
15811 * targhooks.h (default_preferred_simd_mode): Likewise.
15812 * targhooks.c (default_preferred_simd_mode): Likewise.
15813 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
15814 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
15815 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
15816 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
15817 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
15818 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
15819 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
15820 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
15822 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
15823 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
15824 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
15825 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
15826 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
15827 * doc/tm.texi: Regenerate.
15828 * optabs-query.c (can_vec_mask_load_store_p): Return false for
15831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15832 Alan Hayward <alan.hayward@arm.com>
15833 David Sherwood <david.sherwood@arm.com>
15835 * target.def (scalar_mode_supported_p): Take a scalar_mode
15836 instead of a machine_mode.
15837 * targhooks.h (default_scalar_mode_supported_p): Likewise.
15838 * targhooks.c (default_scalar_mode_supported_p): Likewise.
15839 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
15840 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15841 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
15842 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
15843 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
15844 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
15845 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
15846 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
15847 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
15848 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
15849 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
15850 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
15852 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
15853 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
15854 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15855 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15856 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
15858 * doc/tm.texi: Regenerate.
15860 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15861 Alan Hayward <alan.hayward@arm.com>
15862 David Sherwood <david.sherwood@arm.com>
15864 * coretypes.h (opt_scalar_mode): New typedef.
15865 * gdbhooks.py (build_pretty_printers): Handle it.
15866 * machmode.h (mode_iterator::get_2xwider): Add overload for
15868 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
15870 * expr.c (convert_mode_scalar): Likewise.
15871 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15872 * optabs.c (expand_float): Likewise.
15873 (expand_fix): Likewise.
15874 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15877 Alan Hayward <alan.hayward@arm.com>
15878 David Sherwood <david.sherwood@arm.com>
15880 * optabs.c (expand_float): Explicitly check for scalars before
15881 using a branching expansion.
15882 (expand_fix): Likewise.
15884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15885 Alan Hayward <alan.hayward@arm.com>
15886 David Sherwood <david.sherwood@arm.com>
15888 * expr.c (convert_mode): Split scalar handling out into...
15889 (convert_mode_scalar): ...this new function. Treat the modes
15892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15893 Alan Hayward <alan.hayward@arm.com>
15894 David Sherwood <david.sherwood@arm.com>
15896 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
15898 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
15900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15901 Alan Hayward <alan.hayward@arm.com>
15902 David Sherwood <david.sherwood@arm.com>
15904 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
15905 rather than a machine_mode.
15906 (fixed_from_string): Likewise.
15907 (fixed_convert): Likewise.
15908 (fixed_convert_from_int): Likewise.
15909 (fixed_convert_from_real): Likewise.
15910 (real_convert_from_fixed): Likewise.
15911 * fixed-value.c (fixed_from_double_int): Likewise.
15912 (fixed_from_string): Likewise.
15913 (fixed_convert): Likewise.
15914 (fixed_convert_from_int): Likewise.
15915 (fixed_convert_from_real): Likewise.
15916 (real_convert_from_fixed): Likewise.
15917 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
15919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15920 Alan Hayward <alan.hayward@arm.com>
15921 David Sherwood <david.sherwood@arm.com>
15923 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
15924 of separate mode class checks. Do not allow vector modes here.
15925 (immed_wide_int_const): Use as_a <scalar_mode>.
15926 * explow.c (trunc_int_for_mode): Likewise.
15927 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15928 (wi::shwi): Likewise.
15929 (wi::min_value): Likewise.
15930 (wi::max_value): Likewise.
15931 * dwarf2out.c (loc_descriptor): Likewise.
15932 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15933 for CONST_WIDE_INT.
15935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15936 Alan Hayward <alan.hayward@arm.com>
15937 David Sherwood <david.sherwood@arm.com>
15939 * tree.h (SCALAR_TYPE_MODE): New macro.
15940 * expr.c (expand_expr_addr_expr_1): Use it.
15941 (expand_expr_real_2): Likewise.
15942 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15943 (fold_convert_const_fixed_from_int): Likewise.
15944 (fold_convert_const_fixed_from_real): Likewise.
15945 (native_encode_fixed): Likewise
15946 (native_encode_complex): Likewise
15947 (native_encode_vector): Likewise.
15948 (native_interpret_fixed): Likewise.
15949 (native_interpret_real): Likewise.
15950 (native_interpret_complex): Likewise.
15951 (native_interpret_vector): Likewise.
15952 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15953 (simd_clone_adjust_argument_types): Likewise.
15954 (simd_clone_init_simd_arrays): Likewise.
15955 (simd_clone_adjust): Likewise.
15956 * stor-layout.c (layout_type): Likewise.
15957 * tree.c (build_minus_one_cst): Likewise.
15958 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15959 * tree-inline.c (estimate_move_cost): Likewise.
15960 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15961 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15962 (vectorizable_reduction): Likewise.
15963 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15964 (vect_recog_mixed_size_cond_pattern): Likewise.
15965 (check_bool_pattern): Likewise.
15966 (adjust_bool_pattern): Likewise.
15967 (search_type_for_mask_1): Likewise.
15968 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15969 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15970 (vectorizable_load): Likewise.
15971 (vectorizable_store): Likewise.
15972 * ubsan.c (ubsan_encode_value): Likewise.
15973 * varasm.c (output_constant): Likewise.
15975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15976 Alan Hayward <alan.hayward@arm.com>
15977 David Sherwood <david.sherwood@arm.com>
15979 * coretypes.h (scalar_mode): New class.
15980 * machmode.h (scalar_mode): Likewise.
15981 (scalar_mode::includes_p): New function.
15982 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15983 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15984 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15985 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15986 * expmed.c (store_bit_field_1): Likewise.
15987 (extract_bit_field_1): Likewise.
15988 * expr.c (write_complex_part): Likewise.
15989 (read_complex_part): Likewise.
15990 (emit_move_complex_push): Likewise.
15991 (expand_expr_real_2): Likewise.
15992 * function.c (assign_parm_setup_reg): Likewise.
15993 (assign_parms_unsplit_complex): Likewise.
15994 * optabs.c (expand_binop): Likewise.
15995 * rtlanal.c (subreg_get_info): Likewise.
15996 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15997 * varasm.c (output_constant_pool_2): Likewise.
15999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16000 Alan Hayward <alan.hayward@arm.com>
16001 David Sherwood <david.sherwood@arm.com>
16003 * expmed.c (extract_high_half): Use scalar_int_mode and remove
16005 (expmed_mult_highpart_optab): Likewise.
16006 (expmed_mult_highpart): Likewise.
16008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16009 Alan Hayward <alan.hayward@arm.com>
16010 David Sherwood <david.sherwood@arm.com>
16012 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
16013 instead of a machine_mode.
16014 (builtin_memset_read_str): Likewise.
16015 * builtins.c (c_readstr): Likewise.
16016 (builtin_memcpy_read_str): Likewise.
16017 (builtin_strncpy_read_str): Likewise.
16018 (builtin_memset_read_str): Likewise.
16019 (builtin_memset_gen_str): Likewise.
16020 (expand_builtin_signbit): Use scalar_int_mode for local variables.
16021 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
16022 instead of a machine_mode.
16023 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
16025 (make_extraction): Likewise.
16026 (try_widen_shift_mode): Take and return scalar_int_modes instead
16028 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
16029 a scalar_int_mode instead of a machine_mode.
16030 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
16031 (avr_addr_space_pointer_mode): Likewise.
16032 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
16033 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
16034 (msp430_unwind_word_mode): Likewise.
16035 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
16036 (spu_addr_space_pointer_mode): Likewise.
16037 (spu_addr_space_address_mode): Likewise.
16038 (spu_libgcc_cmp_return_mode): Likewise.
16039 (spu_libgcc_shift_count_mode): Likewise.
16040 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
16041 (rl78_addr_space_pointer_mode): Likewise.
16042 (fl78_unwind_word_mode): Likewise.
16043 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
16045 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
16046 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
16047 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
16048 (mips_valid_pointer_mode): Likewise.
16049 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
16050 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
16051 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
16053 (ft32_addr_space_address_mode): Likewise.
16054 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
16055 scalar_int_mode instead of a machine_mode.
16056 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
16058 (m32c_addr_space_address_mode): Likewise.
16059 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
16060 (rs6000_eh_return_filter_mode): Likewise.
16061 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
16062 (rs6000_eh_return_filter_mode): Likewise.
16063 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
16064 (s390_libgcc_shift_count_mode): Likewise.
16065 (s390_unwind_word_mode): Likewise.
16066 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
16068 * target.def (mode_rep_extended): Likewise.
16069 (valid_pointer_mode): Likewise.
16070 (addr_space.valid_pointer_mode): Likewise.
16071 (eh_return_filter_mode): Return a scalar_int_mode rather than
16073 (libgcc_cmp_return_mode): Likewise.
16074 (libgcc_shift_count_mode): Likewise.
16075 (unwind_word_mode): Likewise.
16076 (addr_space.pointer_mode): Likewise.
16077 (addr_space.address_mode): Likewise.
16078 * doc/tm.texi: Regenerate.
16079 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
16081 (do_jump): Use scalar_int_mode for local variables.
16082 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
16083 rather than a machine_mode.
16084 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
16085 (scompare_loc_descriptor_wide): Likewise.
16086 (scompare_loc_descriptor_narrow): Likewise.
16087 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
16089 * except.c (sjlj_emit_dispatch_table): Likewise.
16090 (expand_builtin_eh_copy_values): Likewise.
16091 * explow.c (convert_memory_address_addr_space_1): Likewise.
16092 Take a scalar_int_mode rather than a machine_mode.
16093 (convert_memory_address_addr_space): Take a scalar_int_mode rather
16094 than a machine_mode.
16095 (memory_address_addr_space): Use scalar_int_mode for local variables.
16096 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
16097 rather than a machine_mode.
16098 * expmed.c (mask_rtx): Likewise.
16099 (init_expmed_one_conv): Likewise.
16100 (expand_mult_highpart_adjust): Likewise.
16101 (extract_high_half): Likewise.
16102 (expmed_mult_highpart_optab): Likewise.
16103 (expmed_mult_highpart): Likewise.
16104 (expand_smod_pow2): Likewise.
16105 (expand_sdiv_pow2): Likewise.
16106 (emit_store_flag_int): Likewise.
16107 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
16109 (extract_low_bits): Likewise.
16110 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
16112 * expr.c (pieces_addr::adjust): Likewise.
16113 (can_store_by_pieces): Likewise.
16114 (store_by_pieces): Likewise.
16115 (clear_by_pieces_1): Likewise.
16116 (expand_expr_addr_expr_1): Likewise.
16117 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
16118 (expand_expr_real_1): Likewise.
16119 (try_casesi): Likewise.
16120 * final.c (shorten_branches): Likewise.
16121 * fold-const.c (fold_convert_const_int_from_fixed): Change the
16122 type of "mode" to machine_mode.
16123 * internal-fn.c (expand_arith_overflow_result_store): Take a
16124 scalar_int_mode rather than a machine_mode.
16125 (expand_mul_overflow): Use scalar_int_mode for local variables.
16126 * loop-doloop.c (doloop_modify): Likewise.
16127 (doloop_optimize): Likewise.
16128 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
16129 than a machine_mode.
16130 (expand_doubleword_shift_condmove): Likewise.
16131 (expand_doubleword_shift): Likewise.
16132 (expand_doubleword_clz): Likewise.
16133 (expand_doubleword_popcount): Likewise.
16134 (expand_doubleword_parity): Likewise.
16135 (expand_absneg_bit): Use scalar_int_mode for local variables.
16136 (prepare_float_lib_cmp): Likewise.
16137 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
16138 rather than a machine_mode.
16139 (convert_memory_address_addr_space): Likewise.
16140 (get_mode_bounds): Likewise.
16141 (get_address_mode): Return a scalar_int_mode rather than a
16143 * rtlanal.c (get_address_mode): Likewise.
16144 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
16145 than a machine_mode.
16146 * targhooks.c (default_mode_rep_extended): Likewise.
16147 (default_valid_pointer_mode): Likewise.
16148 (default_addr_space_valid_pointer_mode): Likewise.
16149 (default_eh_return_filter_mode): Return a scalar_int_mode rather
16150 than a machine_mode.
16151 (default_libgcc_cmp_return_mode): Likewise.
16152 (default_libgcc_shift_count_mode): Likewise.
16153 (default_unwind_word_mode): Likewise.
16154 (default_addr_space_pointer_mode): Likewise.
16155 (default_addr_space_address_mode): Likewise.
16156 * targhooks.h (default_eh_return_filter_mode): Likewise.
16157 (default_libgcc_cmp_return_mode): Likewise.
16158 (default_libgcc_shift_count_mode): Likewise.
16159 (default_unwind_word_mode): Likewise.
16160 (default_addr_space_pointer_mode): Likewise.
16161 (default_addr_space_address_mode): Likewise.
16162 (default_mode_rep_extended): Take a scalar_int_mode rather than
16164 (default_valid_pointer_mode): Likewise.
16165 (default_addr_space_valid_pointer_mode): Likewise.
16166 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
16168 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
16169 rather than a machine_mode.
16170 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
16171 for local variables.
16172 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16173 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
16174 than a machine_mode.
16176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16177 Alan Hayward <alan.hayward@arm.com>
16178 David Sherwood <david.sherwood@arm.com>
16180 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
16181 the mode argument to scalar_int_mode.
16182 (do_jump_by_parts_zero_rtx): Likewise.
16183 (do_jump_by_parts_equality_rtx): Likewise.
16184 (do_jump_by_parts_greater): Take a mode argument.
16185 (do_jump_by_parts_equality): Likewise.
16186 (do_jump_1): Update calls accordingly.
16188 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16189 Alan Hayward <alan.hayward@arm.com>
16190 David Sherwood <david.sherwood@arm.com>
16192 * is-a.h (safe_dyn_cast): New function.
16193 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
16194 (jump_table_for_label): Likewise.
16195 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
16196 instead of an rtx_insn *.
16197 (shorten_branches): Use dyn_cast instead of LABEL_P and
16198 JUMP_TABLE_DATA_P. Use jump_table_for_label and
16199 rtx_jump_table_data::get_data_mode.
16200 (final_scan_insn): Likewise.
16202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16203 Alan Hayward <alan.hayward@arm.com>
16204 David Sherwood <david.sherwood@arm.com>
16206 * combine.c (try_combine): Use is_a <scalar_int_mode> when
16207 trying to combine a full-register integer set with a subreg
16210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16211 Alan Hayward <alan.hayward@arm.com>
16212 David Sherwood <david.sherwood@arm.com>
16214 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
16215 that is always either address_mode or pointer_mode.
16217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16218 Alan Hayward <alan.hayward@arm.com>
16219 David Sherwood <david.sherwood@arm.com>
16221 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
16222 when the two are known to be equal.
16224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16225 Alan Hayward <alan.hayward@arm.com>
16226 David Sherwood <david.sherwood@arm.com>
16228 * simplify-rtx.c (simplify_const_unary_operation): Use
16229 is_a <scalar_int_mode> instead of checking for a nonzero
16230 precision. Forcibly convert op_mode to a scalar_int_mode
16231 in that case. More clearly differentiate the operand and
16232 result modes and use the former when deciding what the value
16233 of a count-bits operation should be. Use is_int_mode instead
16234 of checking for a MODE_INT. Remove redundant check for whether
16235 this mode has a zero precision.
16237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16238 Alan Hayward <alan.hayward@arm.com>
16239 David Sherwood <david.sherwood@arm.com>
16241 * optabs.c (widen_leading): Change the type of the mode argument
16242 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
16243 (widen_bswap): Likewise.
16244 (expand_parity): Likewise.
16245 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
16246 (expand_ffs): Likewise.
16247 (epand_unop): Check for scalar integer modes before calling the
16250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16251 Alan Hayward <alan.hayward@arm.com>
16252 David Sherwood <david.sherwood@arm.com>
16254 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
16256 (expand_expr_real_1): Update call accordingly.
16258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16259 Alan Hayward <alan.hayward@arm.com>
16260 David Sherwood <david.sherwood@arm.com>
16262 * expmed.c (store_bit_field_using_insv): Add op0_mode and
16263 value_mode arguments. Use scalar_int_mode internally.
16264 (store_bit_field_1): Rename the new integer mode from imode
16265 to op0_mode and use it instead of GET_MODE (op0). Update calls
16266 to store_split_bit_field, store_bit_field_using_insv and
16267 store_fixed_bit_field.
16268 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
16269 Use scalar_int_mode internally. Use a bit count rather than a mode
16270 when calculating the largest bit size for get_best_mode.
16271 Update calls to store_split_bit_field and store_fixed_bit_field_1.
16272 (store_fixed_bit_field_1): Add mode and value_mode arguments.
16273 Remove assertion that OP0 has a scalar integer mode.
16274 (store_split_bit_field): Add op0_mode and value_mode arguments.
16275 Update calls to extract_fixed_bit_field.
16276 (extract_bit_field_using_extv): Add an op0_mode argument.
16277 Use scalar_int_mode internally.
16278 (extract_bit_field_1): Rename the new integer mode from imode to
16279 op0_mode and use it instead of GET_MODE (op0). Update calls to
16280 extract_split_bit_field, extract_bit_field_using_extv and
16281 extract_fixed_bit_field.
16282 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
16283 to extract_split_bit_field and extract_fixed_bit_field_1.
16284 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
16285 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
16286 on the target mode.
16287 (extract_split_bit_field): Add an op0_mode argument. Update call
16288 to extract_fixed_bit_field.
16290 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16291 Alan Hayward <alan.hayward@arm.com>
16292 David Sherwood <david.sherwood@arm.com>
16294 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
16295 * explow.c (hard_function_value): Likewise.
16296 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
16297 convert_to_mode call outside the loop.
16298 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
16299 for the mode iterator. Require the mode specified by max_pieces
16301 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
16303 (copy_blkmode_to_reg): Likewise.
16304 (set_storage_via_setmem): Likewise.
16305 * optabs.c (prepare_cmp_insn): Likewise.
16306 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16307 * stor-layout.c (finish_bitfield_representative): Likewise.
16309 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16310 Alan Hayward <alan.hayward@arm.com>
16311 David Sherwood <david.sherwood@arm.com>
16313 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
16314 * expr.c (convert_move): Use them.
16315 (convert_modes): Likewise.
16316 (store_expr_with_bounds): Likewise.
16318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16319 Alan Hayward <alan.hayward@arm.com>
16320 David Sherwood <david.sherwood@arm.com>
16322 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
16323 parameter for the mode of "x". Remove the "known_x", "known_mode"
16324 and "known_ret" arguments. Change the type of the mode argument
16325 to scalar_int_mode.
16326 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
16327 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
16328 (reg_num_sign_bit_copies_for_combine): Likewise.
16329 * rtlanal.c (nonzero_bits1): Likewise.
16330 (num_sign_bit_copies1): Likewise.
16331 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
16332 (reg_num_sign_bit_copies_general): Likewise.
16333 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
16334 (reg_nonzero_bits_general): Likewise.
16336 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16337 Alan Hayward <alan.hayward@arm.com>
16338 David Sherwood <david.sherwood@arm.com>
16340 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
16341 than in subroutines. Return 1 for non-integer modes.
16342 (cached_num_sign_bit_copies): Change the type of the mode parameter
16343 to scalar_int_mode.
16344 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
16345 classes. Handle CONST_INT_P first and then check whether X also
16346 has a scalar integer mode. Check the same thing for inner registers
16347 of a SUBREG and for values that are being extended or truncated.
16349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16350 Alan Hayward <alan.hayward@arm.com>
16351 David Sherwood <david.sherwood@arm.com>
16353 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
16354 in subroutines. Return the mode mask for non-integer modes.
16355 (cached_nonzero_bits): Change the type of the mode parameter
16356 to scalar_int_mode.
16357 (nonzero_bits1): Likewise. Remove early exit for other mode
16358 classes. Handle CONST_INT_P first and then check whether X
16359 also has a scalar integer mode.
16361 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16362 Alan Hayward <alan.hayward@arm.com>
16363 David Sherwood <david.sherwood@arm.com>
16365 * expr.c (widest_int_mode_for_size): Make the comment match the code.
16366 Return a scalar_int_mode and assert that the size is greater than
16368 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
16369 (op_by_pieces_d::op_by_pieces_d): Likewise.
16370 (op_by_pieces_d::run): Likewise.
16371 (can_store_by_pieces): Likewise.
16373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16374 Alan Hayward <alan.hayward@arm.com>
16375 David Sherwood <david.sherwood@arm.com>
16377 * combine.c (extract_left_shift): Add a mode argument and update
16379 (make_compound_operation_int): Change the type of the mode parameter
16380 to scalar_int_mode and update the call to extract_left_shift.
16382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16383 Alan Hayward <alan.hayward@arm.com>
16384 David Sherwood <david.sherwood@arm.com>
16386 * combine.c (simplify_and_const_int): Change the type of the mode
16387 parameter to scalar_int_mode.
16388 (simplify_and_const_int_1): Likewise. Update recursive call.
16390 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16391 Alan Hayward <alan.hayward@arm.com>
16392 David Sherwood <david.sherwood@arm.com>
16394 * combine.c (simplify_compare_const): Check that the mode is a
16395 scalar_int_mode (rather than VOIDmode) before testing its
16397 (simplify_comparison): Move COMPARISON_P handling out of the
16398 loop and restrict the latter part of the loop to scalar_int_modes.
16399 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
16400 and when considering SUBREG_REGs. Use is_int_mode instead of
16401 checking GET_MODE_CLASS against MODE_INT.
16403 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16404 Alan Hayward <alan.hayward@arm.com>
16405 David Sherwood <david.sherwood@arm.com>
16407 * combine.c (try_widen_shift_mode): Move check for equal modes to...
16408 (simplify_shift_const_1): ...here. Use scalar_int_mode for
16409 shift_unit_mode and for modes involved in scalar shifts.
16411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16412 Alan Hayward <alan.hayward@arm.com>
16413 David Sherwood <david.sherwood@arm.com>
16415 * combine.c (force_int_to_mode): New function, split out from...
16416 (force_to_mode): ...here. Keep xmode up-to-date and use it
16417 instead of GET_MODE (x).
16419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16420 Alan Hayward <alan.hayward@arm.com>
16421 David Sherwood <david.sherwood@arm.com>
16423 * optabs-query.h (extraction_insn::struct_mode): Change type to
16424 opt_scalar_int_mode and update comment.
16425 (extraction_insn::field_mode): Change type to scalar_int_mode.
16426 (extraction_insn::pos_mode): Likewise.
16427 * combine.c (make_extraction): Update accordingly.
16428 * optabs-query.c (get_traditional_extraction_insn): Likewise.
16429 (get_optab_extraction_insn): Likewise.
16430 * recog.c (simplify_while_replacing): Likewise.
16431 * expmed.c (narrow_bit_field_mem): Change the type of the mode
16432 parameter to opt_scalar_int_mode.
16434 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16435 Alan Hayward <alan.hayward@arm.com>
16436 David Sherwood <david.sherwood@arm.com>
16438 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
16439 to a scalar_int_mode instead of a machine_mode.
16440 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
16441 (get_best_mode): Return a boolean and use a pointer argument to store
16442 the selected mode. Replace the limit mode parameter with a bit limit.
16443 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
16444 for the values returned by bit_field_mode_iterator::next_mode.
16445 (store_bit_field): Update call to get_best_mode.
16446 (store_fixed_bit_field): Likewise.
16447 (extract_fixed_bit_field): Likewise.
16448 * expr.c (optimize_bitfield_assignment_op): Likewise.
16449 * fold-const.c (optimize_bit_field_compare): Likewise.
16450 (fold_truth_andor_1): Likewise.
16451 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
16452 Update for new type of m_mode.
16453 (get_best_mode): As above.
16455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16456 Alan Hayward <alan.hayward@arm.com>
16457 David Sherwood <david.sherwood@arm.com>
16459 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
16460 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
16461 (store_bit_field): Check is_a <scalar_int_mode> before calling
16462 strict_volatile_bitfield_p.
16463 (extract_bit_field): Likewise.
16465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16466 Alan Hayward <alan.hayward@arm.com>
16467 David Sherwood <david.sherwood@arm.com>
16469 * target.def (cstore_mode): Return a scalar_int_mode.
16470 * doc/tm.texi: Regenerate.
16471 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
16472 * targhooks.h (default_cstore_mode): Likewise.
16473 * targhooks.c (default_cstore_mode): Likewise, using a forced
16475 * expmed.c (emit_cstore): Expect the target of the cstore to be
16478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16479 Alan Hayward <alan.hayward@arm.com>
16480 David Sherwood <david.sherwood@arm.com>
16482 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
16484 (niter_desc): Likewise mode.
16485 (iv_analyze): Add a mode parameter.
16487 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
16488 and change its type to scalar_int_mode.
16489 * loop-iv.c: Update commentary at head of file.
16490 (iv_constant): Pass the mode paraeter before the rtx it describes
16491 and change its type to scalar_int_mode. Remove VOIDmode handling.
16492 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
16493 (iv_extend): Likewise.
16494 (shorten_into_mode): Likewise.
16495 (iv_add): Use scalar_int_mode.
16496 (iv_mult): Likewise.
16497 (iv_shift): Likewise.
16498 (canonicalize_iv_subregs): Likewise.
16499 (get_biv_step_1): Pass the outer_mode parameter before the rtx
16500 it describes and change its mode to scalar_int_mode. Also change
16501 the type of the returned inner_mode to scalar_int_mode.
16502 (get_biv_step): Likewise, turning outer_mode from a pointer
16503 into a direct parameter. Update call to get_biv_step_1.
16504 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
16505 iv_constant and get_biv_step.
16506 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
16507 and change its type to scalar_int_mode. Don't initialise iv->mode
16508 to VOIDmode and remove later checks for its still being VOIDmode.
16509 Update calls to iv_analyze_op and iv_analyze_expr. Check
16510 is_a <scalar_int_mode> when changing the mode under consideration.
16511 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
16512 Update call to iv_analyze_expr.
16513 (iv_analyze_op): Add a mode parameter. Reject subregs whose
16514 inner register is not also a scalar_int_mode. Update call to
16516 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
16517 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
16518 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
16519 separate mode class checks. Update calls to iv_analyze. Remove
16520 fix-up of VOIDmodes after iv_analyze_biv.
16521 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
16522 don't have a scalar_int_mode. Update call to biv_p.
16524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16525 Alan Hayward <alan.hayward@arm.com>
16526 David Sherwood <david.sherwood@arm.com>
16528 * cfgexpand.c (convert_debug_memory_address): Use
16529 as_a <scalar_int_mode>.
16530 * combine.c (expand_compound_operation): Likewise.
16531 (make_extraction): Likewise.
16532 (change_zero_ext): Likewise.
16533 (simplify_comparison): Likewise.
16534 * cse.c (cse_insn): Likewise.
16535 * dwarf2out.c (minmax_loc_descriptor): Likewise.
16536 (mem_loc_descriptor): Likewise.
16537 (loc_descriptor): Likewise.
16538 * expmed.c (init_expmed_one_mode): Likewise.
16539 (synth_mult): Likewise.
16540 (emit_store_flag_1): Likewise.
16541 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
16542 of a comparison with size.
16543 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
16544 (reduce_to_bit_field_precision): Likewise.
16545 * function.c (expand_function_end): Likewise.
16546 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
16547 * loop-doloop.c (doloop_modify): Likewise.
16548 * optabs.c (expand_binop): Likewise.
16549 (expand_unop): Likewise.
16550 (expand_copysign_absneg): Likewise.
16551 (prepare_cmp_insn): Likewise.
16552 (maybe_legitimize_operand): Likewise.
16553 * recog.c (const_scalar_int_operand): Likewise.
16554 * rtlanal.c (get_address_mode): Likewise.
16555 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16556 (simplify_cond_clz_ctz): Likewise.
16557 * tree-nested.c (get_nl_goto_field): Likewise.
16558 * tree.c (build_vector_type_for_mode): Likewise.
16559 * var-tracking.c (use_narrower_mode): Likewise.
16561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16562 Alan Hayward <alan.hayward@arm.com>
16563 David Sherwood <david.sherwood@arm.com>
16565 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
16566 * builtins.c (expand_builtin_signbit): Use it.
16567 * cfgexpand.c (expand_debug_expr): Likewise.
16568 * dojump.c (do_jump): Likewise.
16569 (do_compare_and_jump): Likewise.
16570 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
16571 * expmed.c (make_tree): Likewise.
16572 * expr.c (expand_expr_real_2): Likewise.
16573 (expand_expr_real_1): Likewise.
16574 (try_casesi): Likewise.
16575 * fold-const-call.c (fold_const_call_ss): Likewise.
16576 * fold-const.c (unextend): Likewise.
16577 (extract_muldiv_1): Likewise.
16578 (fold_single_bit_test): Likewise.
16579 (native_encode_int): Likewise.
16580 (native_encode_string): Likewise.
16581 (native_interpret_int): Likewise.
16582 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
16583 * internal-fn.c (expand_addsub_overflow): Likewise.
16584 (expand_neg_overflow): Likewise.
16585 (expand_mul_overflow): Likewise.
16586 (expand_arith_overflow): Likewise.
16587 * match.pd: Likewise.
16588 * stor-layout.c (layout_type): Likewise.
16589 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16590 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
16591 * tree-ssanames.c (get_range_info): Likewise.
16592 * tree-switch-conversion.c (array_value_type) Likewise.
16593 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
16594 (vect_recog_divmod_pattern): Likewise.
16595 (vect_recog_mixed_size_cond_pattern): Likewise.
16596 * tree-vrp.c (extract_range_basic): Likewise.
16597 (simplify_float_conversion_using_ranges): Likewise.
16598 * tree.c (int_fits_type_p): Likewise.
16599 * ubsan.c (instrument_bool_enum_load): Likewise.
16600 * varasm.c (mergeable_string_section): Likewise.
16601 (narrowing_initializer_constant_valid_p): Likewise.
16602 (output_constant): Likewise.
16604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16605 Alan Hayward <alan.hayward@arm.com>
16606 David Sherwood <david.sherwood@arm.com>
16608 * machmode.h (NARROWEST_INT_MODE): New macro.
16609 * expr.c (alignment_for_piecewise_move): Use it instead of
16610 GET_CLASS_NARROWEST_MODE (MODE_INT).
16611 (push_block): Likewise.
16612 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
16614 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16617 Alan Hayward <alan.hayward@arm.com>
16618 David Sherwood <david.sherwood@arm.com>
16620 * postreload.c (move2add_valid_value_p): Change the type of the
16621 mode parameter to scalar_int_mode.
16622 (move2add_use_add2_insn): Add a mode parameter and use it instead
16624 (move2add_use_add3_insn): Likewise.
16625 (reload_cse_move2add): Update accordingly.
16627 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16628 Alan Hayward <alan.hayward@arm.com>
16629 David Sherwood <david.sherwood@arm.com>
16631 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
16633 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
16634 * optabs.c (expand_unop): Likewise.
16636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16637 Alan Hayward <alan.hayward@arm.com>
16638 David Sherwood <david.sherwood@arm.com>
16640 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
16641 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
16642 (popcount_loc_descriptor): Likewise.
16643 (bswap_loc_descriptor): Likewise.
16644 (rotate_loc_descriptor): Likewise.
16645 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
16646 calling the functions above.
16648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16649 Alan Hayward <alan.hayward@arm.com>
16650 David Sherwood <david.sherwood@arm.com>
16652 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
16654 (try_combine): Likewise.
16655 (simplify_if_then_else): Likewise.
16656 * cse.c (cse_insn): Likewise.
16657 * dwarf2out.c (mem_loc_descriptor): Likewise.
16658 * emit-rtl.c (gen_lowpart_common): Likewise.
16659 * simplify-rtx.c (simplify_truncation): Likewise.
16660 (simplify_binary_operation_1): Likewise.
16661 (simplify_const_relational_operation): Likewise.
16662 (simplify_ternary_operation): Likewise.
16663 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
16665 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16666 Alan Hayward <alan.hayward@arm.com>
16667 David Sherwood <david.sherwood@arm.com>
16669 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
16670 * reload.c (push_reload): Likewise.
16671 (find_reloads): Likewise.
16673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16674 Alan Hayward <alan.hayward@arm.com>
16675 David Sherwood <david.sherwood@arm.com>
16677 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
16678 (make_compound_operation_int): Likewise.
16679 (change_zero_ext): Likewise.
16680 * expr.c (convert_move): Likewise.
16681 (convert_modes): Likewise.
16682 * fwprop.c (forward_propagate_subreg): Likewise.
16683 * loop-iv.c (get_biv_step_1): Likewise.
16684 * optabs.c (widen_operand): Likewise.
16685 * postreload.c (move2add_valid_value_p): Likewise.
16686 * recog.c (simplify_while_replacing): Likewise.
16687 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16688 (simplify_binary_operation_1): Likewise. Remove redundant
16689 mode equality check.
16691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16692 Alan Hayward <alan.hayward@arm.com>
16693 David Sherwood <david.sherwood@arm.com>
16695 * combine.c (combine_simplify_rtx): Add checks for
16696 is_a <scalar_int_mode>.
16697 (simplify_if_then_else): Likewise.
16698 (make_field_assignment): Likewise.
16699 (simplify_comparison): Likewise.
16700 * ifcvt.c (noce_try_bitop): Likewise.
16701 * loop-invariant.c (canonicalize_address_mult): Likewise.
16702 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16704 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16705 Alan Hayward <alan.hayward@arm.com>
16706 David Sherwood <david.sherwood@arm.com>
16708 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
16709 is_a <scalar_int_mode> instead of != BLKmode.
16711 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16712 Alan Hayward <alan.hayward@arm.com>
16713 David Sherwood <david.sherwood@arm.com>
16715 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
16716 instead of != VOIDmode.
16717 * combine.c (if_then_else_cond): Likewise.
16718 (change_zero_ext): Likewise.
16719 * dwarf2out.c (mem_loc_descriptor): Likewise.
16720 (loc_descriptor): Likewise.
16721 * rtlanal.c (canonicalize_condition): Likewise.
16722 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
16724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16725 Alan Hayward <alan.hayward@arm.com>
16726 David Sherwood <david.sherwood@arm.com>
16728 * simplify-rtx.c (simplify_binary_operation_1): Use
16729 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
16731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16732 Alan Hayward <alan.hayward@arm.com>
16733 David Sherwood <david.sherwood@arm.com>
16735 * wide-int.h (int_traits<unsigned char>) New class.
16736 (int_traits<unsigned short>) Likewise.
16737 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
16738 Use GET_MODE_UNIT_PRECISION and remove redundant test for
16740 * combine.c (set_nonzero_bits_and_sign_copies): Use
16741 is_a <scalar_int_mode>.
16742 (find_split_point): Likewise.
16743 (combine_simplify_rtx): Likewise.
16744 (simplify_logical): Likewise.
16745 (expand_compound_operation): Likewise.
16746 (expand_field_assignment): Likewise.
16747 (make_compound_operation): Likewise.
16748 (extended_count): Likewise.
16749 (change_zero_ext): Likewise.
16750 (simplify_comparison): Likewise.
16751 * dwarf2out.c (scompare_loc_descriptor): Likewise.
16752 (ucompare_loc_descriptor): Likewise.
16753 (minmax_loc_descriptor): Likewise.
16754 (mem_loc_descriptor): Likewise.
16755 (loc_descriptor): Likewise.
16756 * expmed.c (init_expmed_one_mode): Likewise.
16757 * lra-constraints.c (lra_constraint_offset): Likewise.
16758 * optabs.c (prepare_libcall_arg): Likewise.
16759 * postreload.c (move2add_note_store): Likewise.
16760 * reload.c (operands_match_p): Likewise.
16761 * rtl.h (load_extend_op): Likewise.
16762 * rtlhooks.c (gen_lowpart_general): Likewise.
16763 * simplify-rtx.c (simplify_truncation): Likewise.
16764 (simplify_unary_operation_1): Likewise.
16765 (simplify_binary_operation_1): Likewise.
16766 (simplify_const_binary_operation): Likewise.
16767 (simplify_const_relational_operation): Likewise.
16768 (simplify_subreg): Likewise.
16769 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16770 * var-tracking.c (adjust_mems): Likewise.
16771 (prepare_call_arguments): Likewise.
16773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16774 Alan Hayward <alan.hayward@arm.com>
16775 David Sherwood <david.sherwood@arm.com>
16777 * machmode.h (is_int_mode): New fuction.
16778 * combine.c (find_split_point): Use it.
16779 (combine_simplify_rtx): Likewise.
16780 (simplify_if_then_else): Likewise.
16781 (simplify_set): Likewise.
16782 (simplify_shift_const_1): Likewise.
16783 (simplify_comparison): Likewise.
16784 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
16785 * cse.c (notreg_cost): Likewise.
16786 (cse_insn): Likewise.
16787 * cselib.c (cselib_lookup_1): Likewise.
16788 * dojump.c (do_jump_1): Likewise.
16789 (do_compare_rtx_and_jump): Likewise.
16790 * dse.c (get_call_args): Likewise.
16791 * dwarf2out.c (rtl_for_decl_init): Likewise.
16792 (native_encode_initializer): Likewise.
16793 * expmed.c (emit_store_flag_1): Likewise.
16794 (emit_store_flag): Likewise.
16795 * expr.c (convert_modes): Likewise.
16796 (store_field): Likewise.
16797 (expand_expr_real_1): Likewise.
16798 * fold-const.c (fold_read_from_constant_string): Likewise.
16799 * gimple-ssa-sprintf.c (get_format_string): Likewise.
16800 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
16801 * optabs.c (expand_binop): Likewise.
16802 (expand_unop): Likewise.
16803 (expand_abs_nojump): Likewise.
16804 (expand_one_cmpl_abs_nojump): Likewise.
16805 * simplify-rtx.c (mode_signbit_p): Likewise.
16806 (val_signbit_p): Likewise.
16807 (val_signbit_known_set_p): Likewise.
16808 (val_signbit_known_clear_p): Likewise.
16809 (simplify_relational_operation_1): Likewise.
16810 * tree.c (vector_type_mode): Likewise.
16812 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16813 Alan Hayward <alan.hayward@arm.com>
16814 David Sherwood <david.sherwood@arm.com>
16816 * machmode.h (smallest_mode_for_size): Fix formatting.
16817 (smallest_int_mode_for_size): New function.
16818 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
16819 instead of smallest_mode_for_size.
16820 * combine.c (make_extraction): Likewise.
16821 * config/arc/arc.c (arc_expand_movmem): Likewise.
16822 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
16823 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16824 * config/s390/s390.c (s390_expand_insv): Likewise.
16825 * config/sparc/sparc.c (assign_int_registers): Likewise.
16826 * config/spu/spu.c (spu_function_value): Likewise.
16827 (spu_function_arg): Likewise.
16828 * coverage.c (get_gcov_type): Likewise.
16829 (get_gcov_unsigned_t): Likewise.
16830 * dse.c (find_shift_sequence): Likewise.
16831 * expmed.c (store_bit_field_1): Likewise.
16832 * expr.c (convert_move): Likewise.
16833 (store_field): Likewise.
16834 * internal-fn.c (expand_arith_overflow): Likewise.
16835 * optabs-query.c (get_best_extraction_insn): Likewise.
16836 * optabs.c (expand_twoval_binop_libfunc): Likewise.
16837 * stor-layout.c (layout_type): Likewise.
16838 (initialize_sizetypes): Likewise.
16839 * targhooks.c (default_get_mask_mode): Likewise.
16840 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
16842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16843 Alan Hayward <alan.hayward@arm.com>
16844 David Sherwood <david.sherwood@arm.com>
16846 * machmode.h (opt_mode::else_blk): New function.
16847 (int_mode_for_mode): Declare.
16848 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
16849 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
16851 * cfgexpand.c (expand_debug_expr): Likewise.
16852 * combine.c (gen_lowpart_or_truncate): Likewise.
16853 (gen_lowpart_for_combine): Likewise.
16854 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
16855 * config/avr/avr.c (avr_to_int_mode): Likewise.
16856 (avr_out_plus_1): Likewise.
16857 (avr_out_plus): Likewise.
16858 (avr_out_round): Likewise.
16859 * config/i386/i386.c (ix86_split_to_parts): Likewise.
16860 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16861 (s390_expand_vcond): Likewise.
16862 * config/spu/spu.c (spu_split_immediate): Likewise.
16863 (spu_expand_mov): Likewise.
16864 * dse.c (get_stored_val): Likewise.
16865 * expmed.c (store_bit_field_1): Likewise.
16866 (convert_extracted_bit_field): Use int_mode_for_mode instead of
16868 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
16869 (extract_low_bits): Likewise.
16870 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
16871 handling rather than repeating the check.
16872 (emit_group_store): Likewise.
16873 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
16874 * optabs.c (expand_absneg_bit): Likewise.
16875 (expand_copysign_absneg): Likewise.
16876 (expand_copysign_bit): Likewise.
16877 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
16878 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
16879 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
16880 * var-tracking.c (prepare_call_arguments): Likewise.
16881 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
16882 int_mode_for_mode instead of mode_for_size.
16883 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16886 Alan Hayward <alan.hayward@arm.com>
16887 David Sherwood <david.sherwood@arm.com>
16889 * machmode.h (int_mode_for_size): New function.
16890 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
16891 instead of mode_for_size.
16892 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
16894 * combine.c (expand_field_assignment): Use int_mode_for_size
16895 instead of mode_for_size.
16896 (make_extraction): Likewise.
16897 (simplify_shift_const_1): Likewise.
16898 (simplify_comparison): Likewise.
16899 * dojump.c (do_jump): Likewise.
16900 * dwarf2out.c (mem_loc_descriptor): Likewise.
16901 * emit-rtl.c (init_derived_machine_modes): Likewise.
16902 * expmed.c (flip_storage_order): Likewise.
16903 (convert_extracted_bit_field): Likewise.
16904 * expr.c (copy_blkmode_from_reg): Likewise.
16905 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
16906 * internal-fn.c (expand_mul_overflow): Likewise.
16907 * lower-subreg.c (simple_move): Likewise.
16908 * optabs-libfuncs.c (init_optabs): Likewise.
16909 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16910 * tree.c (vector_type_mode): Likewise.
16911 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
16912 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
16913 * tree-vect-generic.c (expand_vector_parallel): Likewise.
16914 * tree-vect-stmts.c (vectorizable_load): Likewise.
16915 (vectorizable_store): Likewise.
16917 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16918 Alan Hayward <alan.hayward@arm.com>
16919 David Sherwood <david.sherwood@arm.com>
16921 * coretypes.h (pod_mode): New type.
16922 (scalar_int_mode_pod): New typedef.
16923 * machmode.h (pod_mode): New class.
16924 (int_n_data_t::m): Change type to scalar_int_mode_pod.
16925 * genmodes.c (emit_mode_int_n): Update accordingly.
16926 * lower-subreg.h (target_lower_subreg): Change type to
16927 scalar_int_mode_pod.
16928 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16929 scalar_int_mode_pod.
16931 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16932 Alan Hayward <alan.hayward@arm.com>
16933 David Sherwood <david.sherwood@arm.com>
16935 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16936 machine_mode to scalar_int_mode.
16937 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16938 (rs6000_option_override_internal): Remove cast to int.
16939 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16940 machine_mode to scalar_int_mode.
16941 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16942 (rs6000_option_override_internal): Remove cast to int.
16943 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16944 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16946 * config/s390/s390.c (s390_expand_builtin): Likewise.
16947 * coretypes.h (scalar_int_mode): New type.
16948 (opt_scalar_int_mode): New typedef.
16949 * machmode.h (scalar_int_mode): New class.
16950 (scalar_int_mode::includes_p): New function.
16951 (byte_mode): Change type to scalar_int_mode.
16952 (word_mode): Likewise.
16953 (ptr_mode): Likewise.
16954 * emit-rtl.c (byte_mode): Likewise.
16955 (word_mode): Likewise.
16956 (ptr_mode): Likewise.
16957 (init_derived_machine_modes): Update accordingly.
16958 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16959 and MODE_PARTIAL_INT.
16960 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16961 opt_scalar_int_mode.
16963 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16964 Alan Hayward <alan.hayward@arm.com>
16965 David Sherwood <david.sherwood@arm.com>
16967 * target.def (libgcc_floating_mode_supported_p): Take a
16969 * doc/tm.texi: Regenerate.
16970 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16972 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16973 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16977 Alan Hayward <alan.hayward@arm.com>
16978 David Sherwood <david.sherwood@arm.com>
16980 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16981 * doc/tm.texi: Regenerate.
16982 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16983 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16984 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16985 * targhooks.h (default_floatn_mode): Likewise.
16986 * targhooks.c (default_floatn_mode): Likewise.
16987 * tree.c (build_common_tree_nodes): Update accordingly.
16989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16990 Alan Hayward <alan.hayward@arm.com>
16991 David Sherwood <david.sherwood@arm.com>
16993 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16994 (mode_iterator::iterate_p): Likewise.
16995 (mode_iterator::get_wider): Likewise.
16996 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16998 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16999 Alan Hayward <alan.hayward@arm.com>
17000 David Sherwood <david.sherwood@arm.com>
17002 * coretypes.h (opt_scalar_float_mode): New typedef.
17003 * machmode.h (float_mode_for_size): New function.
17004 * emit-rtl.c (double_mode): Delete.
17005 (init_emit_once): Use float_mode_for_size.
17006 * stor-layout.c (layout_type): Likewise.
17007 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
17009 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17010 Alan Hayward <alan.hayward@arm.com>
17011 David Sherwood <david.sherwood@arm.com>
17013 * output.h (assemble_real): Take a scalar_float_mode.
17014 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
17015 * config/arm/arm.md (consttable_4): Likewise.
17016 (consttable_8): Likewise.
17017 (consttable_16): Likewise.
17018 * config/mips/mips.md (consttable_float): Likewise.
17019 * config/s390/s390.c (s390_output_pool_entry): Likewise.
17020 * varasm.c (assemble_real): Take a scalar_float_mode.
17021 (output_constant_pool_2): Update accordingly.
17022 (output_constant): Likewise.
17024 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17025 Alan Hayward <alan.hayward@arm.com>
17026 David Sherwood <david.sherwood@arm.com>
17028 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
17029 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
17030 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
17031 (native_encode_real): Likewise.
17032 (native_interpret_real): Likewise.
17033 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
17034 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17037 Alan Hayward <alan.hayward@arm.com>
17038 David Sherwood <david.sherwood@arm.com>
17040 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
17041 <scalar_float_mode>. Simplify.
17042 (gen_extend_conv_libfunc): Likewise.
17044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17045 Alan Hayward <alan.hayward@arm.com>
17046 David Sherwood <david.sherwood@arm.com>
17048 * coretypes.h (scalar_float_mode): New type.
17049 * machmode.h (mode_traits::from_int): Use machine_mode if
17050 USE_ENUM_MODES is defined.
17051 (is_a): New function.
17053 (dyn_cast): Likewise.
17054 (scalar_float_mode): New class.
17055 (scalar_float_mode::includes_p): New function.
17056 (is_float_mode): Likewise.
17057 * gdbhooks.py (MachineModePrinter): New class.
17058 (build_pretty_printer): Use it for scalar_float_mode.
17059 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
17060 (format_helper::format_helper): Turn into a template.
17061 * genmodes.c (get_mode_class): New function.
17062 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
17063 or machine_mode if none.
17064 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
17065 as_a <scalar_float_mode>.
17066 * dwarf2out.c (mem_loc_descriptor): Likewise.
17067 (insert_float): Likewise.
17068 (add_const_value_attribute): Likewise.
17069 * simplify-rtx.c (simplify_immed_subreg): Likewise.
17070 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
17071 (expand_unop): Update accordingly.
17072 (expand_abs_nojump): Likewise.
17073 (expand_copysign_absneg): Take a scalar_float_mode.
17074 (expand_copysign_bit): Likewise.
17075 (expand_copysign): Update accordingly.
17077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17078 Alan Hayward <alan.hayward@arm.com>
17079 David Sherwood <david.sherwood@arm.com>
17081 * coretypes.h (opt_mode): New class.
17082 * machmode.h (opt_mode): Likewise.
17083 (opt_mode::else_void): New function.
17084 (opt_mode::require): Likewise.
17085 (opt_mode::exists): Likewise.
17086 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
17087 (GET_MODE_2XWIDER_MODE): Likewise.
17088 (mode_iterator::get_wider): Update accordingly.
17089 (mode_iterator::get_2xwider): Likewise.
17090 (mode_iterator::get_known_wider): Likewise, turning into a template.
17091 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
17092 forcing a wider mode to exist.
17093 * config/cr16/cr16.h (LONG_REG_P): Likewise.
17094 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17095 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
17096 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
17097 * lower-subreg.c (init_lower_subreg): Likewise.
17098 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
17099 on the final iteration.
17100 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
17101 a wider mode exists before asking for a move pattern.
17102 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
17103 forcing a wider mode to exist.
17104 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
17105 returning false if no such mode exists.
17106 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
17107 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
17108 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
17109 Avoid checking for a MODE_INT if we already know the mode is not a
17111 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
17112 forcing a wider mode to exist.
17113 (expmed_mult_highpart_optab): Likewise.
17114 (expmed_mult_highpart): Likewise.
17115 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
17117 * lto-streamer-in.c (lto_input_mode_table): Likewise.
17118 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
17119 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
17120 * internal-fn.c (expand_mul_overflow): Update use of
17121 GET_MODE_2XWIDER_MODE.
17122 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17123 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
17124 GET_MODE_WIDER_MODE.
17125 (convert_plusminus_to_widen): Likewise.
17126 * tree-switch-conversion.c (array_value_type): Likewise.
17127 * var-tracking.c (emit_note_insn_var_location): Likewise.
17128 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17129 Return false inside rather than outside the loop if no wider mode
17131 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
17132 and GET_MODE_2XWIDER_MODE
17133 (can_compare_p): Use else_void.
17134 * gdbhooks.py (OptMachineModePrinter): New class.
17135 (build_pretty_printer): Use it for opt_mode.
17137 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17138 Alan Hayward <alan.hayward@arm.com>
17139 David Sherwood <david.sherwood@arm.com>
17141 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
17142 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
17144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17145 Alan Hayward <alan.hayward@arm.com>
17146 David Sherwood <david.sherwood@arm.com>
17148 * machmode.h (mode_traits): New structure.
17149 (get_narrowest_mode): New function.
17150 (mode_iterator::start): Likewise.
17151 (mode_iterator::iterate_p): Likewise.
17152 (mode_iterator::get_wider): Likewise.
17153 (mode_iterator::get_known_wider): Likewise.
17154 (mode_iterator::get_2xwider): Likewise.
17155 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
17156 (FOR_EACH_MODE): Likewise.
17157 (FOR_EACH_MODE_FROM): Likewise.
17158 (FOR_EACH_MODE_UNTIL): Likewise.
17159 (FOR_EACH_WIDER_MODE): Likewise.
17160 (FOR_EACH_2XWIDER_MODE): Likewise.
17161 * builtins.c (expand_builtin_strlen): Use new mode iterators.
17162 * combine.c (simplify_comparison): Likewise
17163 * config/i386/i386.c (type_natural_mode): Likewise.
17164 * cse.c (cse_insn): Likewise.
17165 * dse.c (find_shift_sequence): Likewise.
17166 * emit-rtl.c (init_derived_machine_modes): Likewise.
17167 (init_emit_once): Likewise.
17168 * explow.c (hard_function_value): Likewise.
17169 * expmed.c (extract_fixed_bit_field_1): Likewise.
17170 (extract_bit_field_1): Likewise.
17171 (expand_divmod): Likewise.
17172 (emit_store_flag_1): Likewise.
17173 * expr.c (init_expr_target): Likewise.
17174 (convert_move): Likewise.
17175 (alignment_for_piecewise_move): Likewise.
17176 (widest_int_mode_for_size): Likewise.
17177 (emit_block_move_via_movmem): Likewise.
17178 (copy_blkmode_to_reg): Likewise.
17179 (set_storage_via_setmem): Likewise.
17180 (compress_float_constant): Likewise.
17181 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17182 * optabs-query.c (get_best_extraction_insn): Likewise.
17183 * optabs.c (expand_binop): Likewise.
17184 (expand_twoval_unop): Likewise.
17185 (expand_twoval_binop): Likewise.
17186 (widen_leading): Likewise.
17187 (widen_bswap): Likewise.
17188 (expand_parity): Likewise.
17189 (expand_unop): Likewise.
17190 (prepare_cmp_insn): Likewise.
17191 (prepare_float_lib_cmp): Likewise.
17192 (expand_float): Likewise.
17193 (expand_fix): Likewise.
17194 (expand_sfix_optab): Likewise.
17195 * postreload.c (move2add_use_add2_insn): Likewise.
17196 * reg-stack.c (reg_to_stack): Likewise.
17197 * reginfo.c (choose_hard_reg_mode): Likewise.
17198 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17199 * stor-layout.c (mode_for_size): Likewise.
17200 (smallest_mode_for_size): Likewise.
17201 (mode_for_vector): Likewise.
17202 (finish_bitfield_representative): Likewise.
17203 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
17204 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
17205 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17206 * var-tracking.c (prepare_call_arguments): Likewise.
17208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17209 Alan Hayward <alan.hayward@arm.com>
17210 David Sherwood <david.sherwood@arm.com>
17212 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
17213 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
17214 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
17215 * machmode.h (mode_size): Move earlier in file.
17216 (mode_precision): Likewise.
17217 (mode_inner): Likewise.
17218 (mode_nunits): Likewise.
17219 (mode_unit_size): Likewise.
17220 (unit_unit_precision): Likewise.
17221 (mode_wider): Likewise.
17222 (mode_2xwider): Likewise.
17223 (machine_mode): New class.
17224 (mode_to_bytes): New function.
17225 (mode_to_bits): Likewise.
17226 (mode_to_precision): Likewise.
17227 (mode_to_inner): Likewise.
17228 (mode_to_unit_size): Likewise.
17229 (mode_to_unit_precision): Likewise.
17230 (mode_to_nunits): Likewise.
17231 (GET_MODE_SIZE): Use mode_to_bytes.
17232 (GET_MODE_BITSIZE): Use mode_to_bits.
17233 (GET_MODE_PRECISION): Use mode_to_precision.
17234 (GET_MODE_INNER): Use mode_to_inner.
17235 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
17236 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
17237 (GET_MODE_NUNITS): Use mode_to_nunits.
17238 * system.h (ALWAYS_INLINE): New macro.
17239 * config/powerpcspe/powerpcspe-c.c
17240 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
17241 int for arg1_mode and arg2_mode.
17243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17244 Alan Hayward <alan.hayward@arm.com>
17245 David Sherwood <david.sherwood@arm.com>
17247 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17248 Prefix mode names with E_ in case statements.
17249 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17250 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
17251 (aarch64_split_simd_move): Likewise.
17252 (aarch64_gen_storewb_pair): Likewise.
17253 (aarch64_gen_loadwb_pair): Likewise.
17254 (aarch64_gen_store_pair): Likewise.
17255 (aarch64_gen_load_pair): Likewise.
17256 (aarch64_get_condition_code_1): Likewise.
17257 (aarch64_constant_pool_reload_icode): Likewise.
17258 (get_rsqrte_type): Likewise.
17259 (get_rsqrts_type): Likewise.
17260 (get_recpe_type): Likewise.
17261 (get_recps_type): Likewise.
17262 (aarch64_gimplify_va_arg_expr): Likewise.
17263 (aarch64_simd_container_mode): Likewise.
17264 (aarch64_emit_load_exclusive): Likewise.
17265 (aarch64_emit_store_exclusive): Likewise.
17266 (aarch64_expand_compare_and_swap): Likewise.
17267 (aarch64_gen_atomic_cas): Likewise.
17268 (aarch64_emit_bic): Likewise.
17269 (aarch64_emit_atomic_swap): Likewise.
17270 (aarch64_emit_atomic_load_op): Likewise.
17271 (aarch64_evpc_trn): Likewise.
17272 (aarch64_evpc_uzp): Likewise.
17273 (aarch64_evpc_zip): Likewise.
17274 (aarch64_evpc_ext): Likewise.
17275 (aarch64_evpc_rev): Likewise.
17276 (aarch64_evpc_dup): Likewise.
17277 (aarch64_gen_ccmp_first): Likewise.
17278 (aarch64_gen_ccmp_next): Likewise.
17279 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17280 (alpha_emit_xfloating_libcall): Likewise.
17281 (emit_insxl): Likewise.
17282 (alpha_arg_type): Likewise.
17283 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
17284 (arc_preferred_simd_mode): Likewise.
17285 (arc_secondary_reload): Likewise.
17286 (get_arc_condition_code): Likewise.
17287 (arc_print_operand): Likewise.
17288 (arc_legitimate_constant_p): Likewise.
17289 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17290 * config/arc/arc.md (casesi_load): Likewise.
17291 (casesi_compact_jump): Likewise.
17292 * config/arc/predicates.md (proper_comparison_operator): Likewise.
17293 (cc_use_register): Likewise.
17294 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17295 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17296 (arm_init_iwmmxt_builtins): Likewise.
17297 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
17298 (neon_expand_vector_init): Likewise.
17299 (arm_attr_length_move_neon): Likewise.
17300 (maybe_get_arm_condition_code): Likewise.
17301 (arm_emit_vector_const): Likewise.
17302 (arm_preferred_simd_mode): Likewise.
17303 (arm_output_iwmmxt_tinsr): Likewise.
17304 (thumb1_output_casesi): Likewise.
17305 (thumb2_output_casesi): Likewise.
17306 (arm_emit_load_exclusive): Likewise.
17307 (arm_emit_store_exclusive): Likewise.
17308 (arm_expand_compare_and_swap): Likewise.
17309 (arm_evpc_neon_vuzp): Likewise.
17310 (arm_evpc_neon_vzip): Likewise.
17311 (arm_evpc_neon_vrev): Likewise.
17312 (arm_evpc_neon_vtrn): Likewise.
17313 (arm_evpc_neon_vext): Likewise.
17314 (arm_validize_comparison): Likewise.
17315 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
17316 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
17317 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
17318 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
17319 (c6x_preferred_simd_mode): Likewise.
17320 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
17321 (epiphany_rtx_costs): Likewise.
17322 * config/epiphany/predicates.md (proper_comparison_operator):
17324 * config/frv/frv.c (condexec_memory_operand): Likewise.
17325 (frv_emit_move): Likewise.
17326 (output_move_single): Likewise.
17327 (output_condmove_single): Likewise.
17328 (frv_hard_regno_mode_ok): Likewise.
17329 (frv_matching_accg_mode): Likewise.
17330 * config/h8300/h8300.c (split_adds_subs): Likewise.
17331 (h8300_rtx_costs): Likewise.
17332 (h8300_print_operand): Likewise.
17333 (compute_mov_length): Likewise.
17334 (output_logical_op): Likewise.
17335 (compute_logical_op_length): Likewise.
17336 (compute_logical_op_cc): Likewise.
17337 (h8300_shift_needs_scratch_p): Likewise.
17338 (output_a_shift): Likewise.
17339 (compute_a_shift_length): Likewise.
17340 (compute_a_shift_cc): Likewise.
17341 (expand_a_rotate): Likewise.
17342 (output_a_rotate): Likewise.
17343 * config/i386/i386.c (classify_argument): Likewise.
17344 (function_arg_advance_32): Likewise.
17345 (function_arg_32): Likewise.
17346 (function_arg_64): Likewise.
17347 (function_value_64): Likewise.
17348 (ix86_gimplify_va_arg): Likewise.
17349 (ix86_legitimate_constant_p): Likewise.
17350 (put_condition_code): Likewise.
17351 (split_double_mode): Likewise.
17352 (ix86_avx256_split_vector_move_misalign): Likewise.
17353 (ix86_expand_vector_logical_operator): Likewise.
17354 (ix86_split_idivmod): Likewise.
17355 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
17356 (ix86_build_const_vector): Likewise.
17357 (ix86_build_signbit_mask): Likewise.
17358 (ix86_match_ccmode): Likewise.
17359 (ix86_cc_modes_compatible): Likewise.
17360 (ix86_expand_branch): Likewise.
17361 (ix86_expand_sse_cmp): Likewise.
17362 (ix86_expand_sse_movcc): Likewise.
17363 (ix86_expand_int_sse_cmp): Likewise.
17364 (ix86_expand_vec_perm_vpermi2): Likewise.
17365 (ix86_expand_vec_perm): Likewise.
17366 (ix86_expand_sse_unpack): Likewise.
17367 (ix86_expand_int_addcc): Likewise.
17368 (ix86_split_to_parts): Likewise.
17369 (ix86_vectorize_builtin_gather): Likewise.
17370 (ix86_vectorize_builtin_scatter): Likewise.
17371 (avx_vpermilp_parallel): Likewise.
17372 (inline_memory_move_cost): Likewise.
17373 (ix86_tieable_integer_mode_p): Likewise.
17374 (x86_maybe_negate_const_int): Likewise.
17375 (ix86_expand_vector_init_duplicate): Likewise.
17376 (ix86_expand_vector_init_one_nonzero): Likewise.
17377 (ix86_expand_vector_init_one_var): Likewise.
17378 (ix86_expand_vector_init_concat): Likewise.
17379 (ix86_expand_vector_init_interleave): Likewise.
17380 (ix86_expand_vector_init_general): Likewise.
17381 (ix86_expand_vector_set): Likewise.
17382 (ix86_expand_vector_extract): Likewise.
17383 (emit_reduc_half): Likewise.
17384 (ix86_emit_i387_round): Likewise.
17385 (ix86_mangle_type): Likewise.
17386 (ix86_expand_round_sse4): Likewise.
17387 (expand_vec_perm_blend): Likewise.
17388 (canonicalize_vector_int_perm): Likewise.
17389 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
17390 (expand_vec_perm_1): Likewise.
17391 (expand_vec_perm_interleave3): Likewise.
17392 (expand_vec_perm_even_odd_pack): Likewise.
17393 (expand_vec_perm_even_odd_1): Likewise.
17394 (expand_vec_perm_broadcast_1): Likewise.
17395 (ix86_vectorize_vec_perm_const_ok): Likewise.
17396 (ix86_expand_vecop_qihi): Likewise.
17397 (ix86_expand_mul_widen_hilo): Likewise.
17398 (ix86_expand_sse2_abs): Likewise.
17399 (ix86_expand_pextr): Likewise.
17400 (ix86_expand_pinsr): Likewise.
17401 (ix86_preferred_simd_mode): Likewise.
17402 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
17403 * config/i386/sse.md (*andnot<mode>3): Likewise.
17404 (<mask_codefor><code><mode>3<mask_name>): Likewise.
17405 (*<code><mode>3): Likewise.
17406 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
17407 (ia64_expand_atomic_op): Likewise.
17408 (ia64_arg_type): Likewise.
17409 (ia64_mode_to_int): Likewise.
17410 (ia64_scalar_mode_supported_p): Likewise.
17411 (ia64_vector_mode_supported_p): Likewise.
17412 (expand_vec_perm_broadcast): Likewise.
17413 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
17414 (iq2000_function_arg_advance): Likewise.
17415 (iq2000_function_arg): Likewise.
17416 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
17417 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
17418 (m68k_libcall_value): Likewise.
17419 (m68k_function_value): Likewise.
17420 (sched_attr_op_type): Likewise.
17421 * config/mcore/mcore.c (mcore_output_move): Likewise.
17422 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
17424 (microblaze_function_arg): Likewise.
17425 * config/mips/mips.c (mips16_build_call_stub): Likewise.
17426 (mips_print_operand): Likewise.
17427 (mips_mode_ok_for_mov_fmt_p): Likewise.
17428 (mips_vector_mode_supported_p): Likewise.
17429 (mips_preferred_simd_mode): Likewise.
17430 (mips_expand_vpc_loongson_even_odd): Likewise.
17431 (mips_expand_vec_unpack): Likewise.
17432 (mips_expand_vi_broadcast): Likewise.
17433 (mips_expand_vector_init): Likewise.
17434 (mips_expand_vec_reduc): Likewise.
17435 (mips_expand_msa_cmp): Likewise.
17436 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
17437 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
17438 (cc_flags_for_mode): Likewise.
17439 * config/msp430/msp430.c (msp430_print_operand): Likewise.
17440 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
17441 (nds32_output_casesi_pc_relative): Likewise.
17442 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17443 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
17444 (nvptx_gen_unpack): Likewise.
17445 (nvptx_gen_pack): Likewise.
17446 (nvptx_gen_shuffle): Likewise.
17447 (nvptx_gen_wcast): Likewise.
17448 (nvptx_preferred_simd_mode): Likewise.
17449 * config/pa/pa.c (pa_secondary_reload): Likewise.
17450 * config/pa/predicates.md (base14_operand): Likewise.
17451 * config/powerpcspe/powerpcspe-c.c
17452 (altivec_resolve_overloaded_builtin): Likewise.
17453 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
17455 (rs6000_preferred_simd_mode): Likewise.
17456 (output_vec_const_move): Likewise.
17457 (rs6000_expand_vector_extract): Likewise.
17458 (rs6000_split_vec_extract_var): Likewise.
17459 (reg_offset_addressing_ok_p): Likewise.
17460 (rs6000_legitimate_offset_address_p): Likewise.
17461 (rs6000_legitimize_address): Likewise.
17462 (rs6000_emit_set_const): Likewise.
17463 (rs6000_const_vec): Likewise.
17464 (rs6000_emit_move): Likewise.
17465 (spe_build_register_parallel): Likewise.
17466 (rs6000_darwin64_record_arg_recurse): Likewise.
17467 (swap_selector_for_mode): Likewise.
17468 (spe_init_builtins): Likewise.
17469 (paired_init_builtins): Likewise.
17470 (altivec_init_builtins): Likewise.
17471 (do_load_for_compare): Likewise.
17472 (rs6000_generate_compare): Likewise.
17473 (rs6000_expand_float128_convert): Likewise.
17474 (emit_load_locked): Likewise.
17475 (emit_store_conditional): Likewise.
17476 (rs6000_output_function_epilogue): Likewise.
17477 (rs6000_handle_altivec_attribute): Likewise.
17478 (rs6000_function_value): Likewise.
17479 (emit_fusion_gpr_load): Likewise.
17480 (emit_fusion_p9_load): Likewise.
17481 (emit_fusion_p9_store): Likewise.
17482 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
17483 (fusion_gpr_mem_load): Likewise.
17484 (fusion_addis_mem_combo_load): Likewise.
17485 (fusion_addis_mem_combo_store): Likewise.
17486 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
17487 (fusion_gpr_mem_load): Likewise.
17488 (fusion_addis_mem_combo_load): Likewise.
17489 (fusion_addis_mem_combo_store): Likewise.
17490 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17492 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
17493 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17494 (rs6000_preferred_simd_mode): Likewise.
17495 (output_vec_const_move): Likewise.
17496 (rs6000_expand_vector_extract): Likewise.
17497 (rs6000_split_vec_extract_var): Likewise.
17498 (reg_offset_addressing_ok_p): Likewise.
17499 (rs6000_legitimate_offset_address_p): Likewise.
17500 (rs6000_legitimize_address): Likewise.
17501 (rs6000_emit_set_const): Likewise.
17502 (rs6000_const_vec): Likewise.
17503 (rs6000_emit_move): Likewise.
17504 (rs6000_darwin64_record_arg_recurse): Likewise.
17505 (swap_selector_for_mode): Likewise.
17506 (paired_init_builtins): Likewise.
17507 (altivec_init_builtins): Likewise.
17508 (rs6000_expand_float128_convert): Likewise.
17509 (emit_load_locked): Likewise.
17510 (emit_store_conditional): Likewise.
17511 (rs6000_output_function_epilogue): Likewise.
17512 (rs6000_handle_altivec_attribute): Likewise.
17513 (rs6000_function_value): Likewise.
17514 (emit_fusion_gpr_load): Likewise.
17515 (emit_fusion_p9_load): Likewise.
17516 (emit_fusion_p9_store): Likewise.
17517 * config/rx/rx.c (rx_gen_move_template): Likewise.
17518 (flags_from_mode): Likewise.
17519 * config/s390/predicates.md (s390_alc_comparison): Likewise.
17520 (s390_slb_comparison): Likewise.
17521 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
17522 (s390_vector_mode_supported_p): Likewise.
17523 (s390_cc_modes_compatible): Likewise.
17524 (s390_match_ccmode_set): Likewise.
17525 (s390_canonicalize_comparison): Likewise.
17526 (s390_emit_compare_and_swap): Likewise.
17527 (s390_branch_condition_mask): Likewise.
17528 (s390_rtx_costs): Likewise.
17529 (s390_secondary_reload): Likewise.
17530 (__SECONDARY_RELOAD_CASE): Likewise.
17531 (s390_expand_cs): Likewise.
17532 (s390_preferred_simd_mode): Likewise.
17533 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
17534 * config/sh/sh.c (sh_print_operand): Likewise.
17535 (dump_table): Likewise.
17536 (sh_secondary_reload): Likewise.
17537 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17538 * config/sh/sh.md (casesi_worker_1): Likewise.
17539 (casesi_worker_2): Likewise.
17540 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
17541 (fcc_comparison_operator): Likewise.
17542 * config/sparc/sparc.c (sparc_expand_move): Likewise.
17543 (emit_soft_tfmode_cvt): Likewise.
17544 (sparc_preferred_simd_mode): Likewise.
17545 (output_cbranch): Likewise.
17546 (sparc_print_operand): Likewise.
17547 (sparc_expand_vec_perm_bmask): Likewise.
17548 (vector_init_bshuffle): Likewise.
17549 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17550 (spu_vector_mode_supported_p): Likewise.
17551 (spu_expand_insv): Likewise.
17552 (spu_emit_branch_or_set): Likewise.
17553 (spu_handle_vector_attribute): Likewise.
17554 (spu_builtin_splats): Likewise.
17555 (spu_builtin_extract): Likewise.
17556 (spu_builtin_promote): Likewise.
17557 (spu_expand_sign_extend): Likewise.
17558 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17559 (tilegx_simd_int): Likewise.
17560 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
17561 (tilepro_simd_int): Likewise.
17562 * config/v850/v850.c (const_double_split): Likewise.
17563 (v850_print_operand): Likewise.
17564 (ep_memory_offset): Likewise.
17565 * config/vax/vax.c (vax_rtx_costs): Likewise.
17566 (vax_output_int_move): Likewise.
17567 (vax_output_int_add): Likewise.
17568 (vax_output_int_subtract): Likewise.
17569 * config/visium/predicates.md (visium_branch_operator): Likewise.
17570 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
17571 (visium_print_operand_address): Likewise.
17572 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17573 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
17574 (xtensa_expand_conditional_branch): Likewise.
17575 (xtensa_copy_incoming_a7): Likewise.
17576 (xtensa_output_literal): Likewise.
17577 * dfp.c (decimal_real_maxval): Likewise.
17578 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17581 Alan Hayward <alan.hayward@arm.com>
17582 David Sherwood <david.sherwood@arm.com>
17584 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
17585 (mode_nunits_inline): Likewise.
17586 (mode_inner_inline): Likewise.
17587 (mode_unit_size_inline): Likewise.
17588 (mode_unit_precision_inline): Likewise.
17589 (emit_insn_modes_h): Likewise. Also emit a #define of the
17591 (emit_mode_wider): Add an E_ prefix to mode names.
17592 (emit_mode_complex): Likewise.
17593 (emit_mode_inner): Likewise.
17594 (emit_mode_adjustments): Likewise.
17595 (emit_mode_int_n): Likewise.
17596 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
17597 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
17598 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
17599 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
17600 (CRC32_BUILTIN, ENTRY): Likewise.
17601 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
17602 (aarch64_pop_regs): Likewise.
17603 (aarch64_process_components): Likewise.
17604 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
17605 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
17606 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
17607 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
17608 * config/arm/arm.c (arm_init_libfuncs): Likewise.
17609 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
17611 * config/i386/i386-builtin.def (pcmpestr): Likewise.
17612 (pcmpistr): Likewise.
17613 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
17614 * config/mmix/mmix.c (mmix_output_condition): Likewise.
17615 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17617 * config/rl78/rl78.c (mduc_regs): Likewise.
17618 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17619 (htm_expand_builtin): Likewise.
17620 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
17621 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
17622 * config/xtensa/xtensa.c (print_operand): Likewise.
17623 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
17624 (NUM_MODE_VECTOR_INT): Likewise.
17625 * genoutput.c (null_operand): Likewise.
17626 (output_operand_data): Likewise.
17627 * genrecog.c (print_parameter_value): Likewise.
17628 * lra.c (debug_operand_data): Likewise.
17630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17632 * dwarf2out.c (scompare_loc_descriptor_wide)
17633 (scompare_loc_descriptor_narrow): New functions, split out from...
17634 (scompare_loc_descriptor): ...here.
17635 * expmed.c (emit_store_flag_int): New function, split out from...
17636 (emit_store_flag): ...here.
17638 2017-08-30 Richard Biener <rguenther@suse.de>
17640 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
17641 (dwarf2out_early_finish): Move setting of AT_pubnames from
17642 early debug output to early finish.
17644 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17646 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
17647 and -mdata-region to the assembler.
17649 2017-08-30 Richard Biener <rguenther@suse.de>
17651 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
17653 (gen_subprogram_die): Add DW_AT_object_pointer only early.
17654 (dwarf2out_early_global_decl): Only generate a DIE for the
17655 abstract origin if it doesn't already exist or is a declaration DIE.
17656 (resolve_addr): Do not add the linkage name twice when
17657 generating a stub DIE for the DW_TAG_GNU_call_site target.
17659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17661 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17662 Use machine_mode rather than int for arg1_mode.
17664 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
17667 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
17668 that the second argument of the built-in functions to unpack
17669 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
17670 switch statement instead a lot of if statements.
17671 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
17672 Allow 64-bit values to be in Altivec registers as well as
17673 traditional floating point registers.
17674 (pack<mode>, FMOVE128_VSX iterator): Likewise.
17676 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
17678 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
17679 MAX_REGS_PER_ADDRESS == 1.
17681 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
17683 * config/i386/i386.opt (flag_fentry): Do not init to -1.
17684 * config/i386/i386.c (ix86_option_override_internal): Simplify
17685 setting of opts->x_flag_entry.
17687 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17688 Jakub Jelinek <jakub@redhat.com>
17689 Richard Biener <rguenther@suse.de>
17691 PR tree-optimization/81503
17692 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
17693 folded constant fits in the target type; reorder tests for clarity.
17695 2017-08-29 Martin Liska <mliska@suse.cz>
17697 * passes.def: Include pass_lower_switch.
17698 * stmt.c (dump_case_nodes): Remove and move to
17699 tree-switch-conversion.
17700 (case_values_threshold): Likewise.
17701 (expand_switch_as_decision_tree_p): Likewise.
17702 (emit_case_decision_tree): Likewise.
17703 (expand_case): Likewise.
17704 (balance_case_nodes): Likewise.
17705 (node_has_low_bound): Likewise.
17706 (node_has_high_bound): Likewise.
17707 (node_is_bounded): Likewise.
17708 (emit_case_nodes): Likewise.
17709 (struct simple_case_node): New struct.
17710 (add_case_node): Remove.
17711 (emit_case_dispatch_table): Use vector instead of case_list.
17712 (reset_out_edges_aux): Remove.
17713 (compute_cases_per_edge): Likewise.
17714 (expand_case): Build list of simple_case_node.
17715 (expand_sjlj_dispatch_table): Use it.
17716 * tree-switch-conversion.c (struct case_node): Moved from
17717 stmt.c and adjusted.
17718 (emit_case_nodes): Likewise.
17719 (node_has_low_bound): Likewise.
17720 (node_has_high_bound): Likewise.
17721 (node_is_bounded): Likewise.
17722 (case_values_threshold): Likewise.
17723 (reset_out_edges_aux): Likewise.
17724 (compute_cases_per_edge): Likewise.
17725 (add_case_node): Likewise.
17726 (dump_case_nodes): Likewise.
17727 (balance_case_nodes): Likewise.
17728 (expand_switch_as_decision_tree_p): Likewise.
17729 (emit_jump): Likewise.
17730 (emit_case_decision_tree): Likewise.
17731 (try_switch_expansion): Likewise.
17732 (do_jump_if_equal): Likewise.
17733 (emit_cmp_and_jump_insns): Likewise.
17734 (fix_phi_operands_for_edge): New function.
17735 (record_phi_operand_mapping): Likewise.
17736 (class pass_lower_switch): New pass.
17737 (pass_lower_switch::execute): New function.
17738 (make_pass_lower_switch): Likewise.
17739 (conditional_probability):
17740 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
17741 * tree-pass.h: Add make_pass_lower_switch.
17743 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17746 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
17749 2017-08-29 Richard Biener <rguenther@suse.de>
17751 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
17752 we do not add a DW_AT_inline attribute twice.
17753 (gen_subprogram_die): Remove code setting DW_AT_inline on
17754 DECL_ABSTRACT_P nodes.
17756 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
17758 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
17759 calls to internal functions.
17760 (gimplify_modify_expr): Likewise.
17761 * tree-call-cdce.c (use_internal_fn): Likewise.
17762 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
17763 (convert_to_divmod): Set the nothrow flag.
17764 * tree-if-conv.c (predicate_mem_writes): Likewise.
17765 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17766 (vectorizable_call): Likewise.
17767 (vectorizable_store): Likewise.
17768 (vectorizable_load): Likewise.
17769 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
17770 (vect_recog_mask_conversion_pattern): Likewise.
17772 2017-08-29 Martin Liska <mliska@suse.cz>
17775 * gcc.c (driver_handle_option): Add new argument.
17776 * opts-common.c (handle_option): Pass
17777 target_option_override_hook.
17778 * opts-global.c (lang_handle_option): Add new option.
17779 (set_default_handlers): Add new argument.
17780 (decode_options): Likewise.
17781 * opts.c (target_handle_option): Likewise.
17782 (common_handle_option): Call target_option_override_hook.
17783 * opts.h (struct cl_option_handler_func): Add hook for
17784 target option override.
17785 (struct cl_option_handlers): Likewise.
17786 (set_default_handlers): Add new argument.
17787 (decode_options): Likewise.
17788 (common_handle_option): Likewise.
17789 (target_handle_option): Likewise.
17790 * toplev.c (toplev::main): Pass targetm.target_option.override
17793 2017-08-29 Richard Biener <rguenther@suse.de>
17794 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
17796 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
17797 life to the active subtree.
17799 2017-08-28 Jeff Law <law@redhat.com>
17801 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
17802 derive_equivalences.
17803 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
17804 Code moved into....
17805 (edge_info::derive_equivalences): New private member function
17807 * tree-ssa-dom.c (class edge_info): Changed from a struct
17808 to a class. Add ctor/dtor, methods and data members.
17809 (edge_info::edge_info): Renamed from allocate_edge_info.
17810 Initialize additional members.
17811 (edge_info::~edge_info): New.
17812 (free_dom_edge_info): Delete the edge info.
17813 (record_edge_info): Use new class & associated member functions.
17814 Tighten forms for testing for edge equivalences.
17815 (record_temporary_equivalences): Iterate over the simple
17816 equivalences rather than assuming there's only one per edge.
17817 (cprop_into_successor_phis): Iterate over the simple
17818 equivalences rather than assuming there's only one per edge.
17819 (optimize_stmt): Use operand_equal_p rather than pointer
17820 equality for mini-DSE code.
17822 2017-08-28 Nathan Sidwell <nathan@acm.org>
17824 * gcc.c (execute): Fold SIGPIPE handling into switch
17825 statement. Adjust internal error message.
17827 2017-08-28 Richard Biener <rguenther@suse.de>
17830 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
17831 Do nothing for removed DIEs.
17833 2017-08-28 Richard Biener <rguenther@suse.de>
17835 PR tree-optimization/81977
17836 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
17839 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
17842 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
17843 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
17844 using targetm.gen_mem_thread_fence.
17846 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
17849 * config/i386/i386.md (*<btsc><mode>): Change operand 2
17850 predicate to register_operand. Reorder operands.
17851 (*btr<mode>): Ditto.
17852 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
17853 (*btr<mode>_mask): Ditto.
17855 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
17857 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
17858 * config/rs6000/xmmintrin.h: New file.
17859 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
17861 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17864 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
17865 parameter and_insn and return it.
17866 (recombine_lvx_pattern): Insert a copy to ensure availability of
17867 the base register of the copied masking operation at the point of
17868 the instruction replacement.
17869 (recombine_stvx_pattern): Likewise.
17871 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17873 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
17874 undocumented switches.
17875 (-mpower9-dform-vector): Likewise.
17876 (-mpower9-dform): Likewise.
17877 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17878 comments to delete references to -mpower9-dform* switches.
17879 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
17880 Delete reference to -mpower9-dform* switches, test for
17881 -mpower9-vector instead.
17882 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
17883 (OTHER_P9_VECTOR_MASKS): Likewise.
17884 (POWERPC_MASKS): Likewise.
17885 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
17886 tests against -mpower9-dform* to -mpower9-vector. Delete code
17887 that checked for -mpower9-dform* consistancy with other options.
17888 Add test for -mpower9-misc to enable other power9 switches.
17889 (rs6000_init_hard_regno_mode_ok): Likewise.
17890 (rs6000_option_override_internal): Likewise.
17891 (rs6000_emit_prologue): Likewise.
17892 (rs6000_emit_epilogue): Likewise.
17893 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
17894 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
17895 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
17897 (emit_fusion_p9_store): Likewise.
17898 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
17899 resetting these macros if the assembler does not support ISA 3.0
17901 (TARGET_P9_DFORM_VECTOR): Likewise.
17902 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
17903 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
17905 2017-08-25 Alan Modra <amodra@gmail.com>
17907 PR rtl-optimization/81747
17908 * cse.c (cse_extended_basic_block): Don't attempt to record
17909 equivalences for degenerate conditional branches that branch
17910 to their fall-through.
17912 2017-08-24 Martin Sebor <msebor@redhat.com>
17914 PR middle-end/81908
17915 * gimple-fold.c (size_must_be_zero_p): New function.
17916 (gimple_fold_builtin_memory_op): Call it.
17918 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
17920 * config/rs6000/mm_malloc.h: New file.
17922 2017-08-24 Bin Cheng <bin.cheng@arm.com>
17924 PR tree-optimization/81913
17925 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
17926 analysis when either IVs in condition can wrap.
17928 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17930 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17931 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17933 2017-08-24 Richard Biener <rguenther@suse.de>
17936 * targhooks.c (default_target_can_inline_p): Properly
17937 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17938 is present and always compare.
17939 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17940 infer -mfpmath=sse from TARGET_SSE_P.
17941 (ix86_can_inline_p): Properly use target_option_default_node when
17942 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17944 2017-08-24 Richard Biener <rguenther@suse.de>
17947 * dwarf2out.c (output_die): Handle flag_generate_offload like
17949 (output_comp_unit): Likewise.
17950 (gen_array_type_die): Likewise.
17951 (dwarf2out_early_finish): Likewise.
17952 (note_variable_value_in_expr): Likewise.
17953 (dwarf2out_finish): Likewise. Adjust assert.
17954 * cgraphunit.c (symbol_table::compile): Move setting of
17955 flag_generate_offload earlier ...
17956 (symbol_table::finalize_compilation_unit): ... here, before
17957 early debug finalization.
17959 2017-08-24 Richard Biener <rguenther@suse.de>
17961 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17962 and ipa-fnsummary.h.
17963 (ix86_can_inline_p): When ix86_fpmath flags do not match
17964 check whether the callee uses FP math at all.
17966 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17968 PR middle-end/81931
17969 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17970 instead of TYPE_PRECISION.
17972 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17973 Alan Hayward <alan.hayward@arm.com>
17974 David Sherwood <david.sherwood@arm.com>
17976 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17978 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17980 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17982 (ix86_frame::stack_realign_allocate): New field.
17983 (struct machine_frame_state): Modify comments.
17984 (machine_frame_state::sp_realigned_fp_end): New field.
17985 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17986 layout calculation.
17987 (sp_valid_at): Add assertion to assure no attempt to access invalid
17988 offset of a realigned stack.
17989 (fp_valid_at): Likewise.
17990 (choose_baseaddr): Modify comments.
17991 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17992 ix86_expand_prologue.
17993 (ix86_expand_prologue): Modify stack realignment and allocation.
17994 (ix86_expand_epilogue): Modify comments.
17995 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17996 avx2_runtime, avx512f, and avx512f_runtime.
17998 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
18000 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
18001 (mstackrealign): Do not init to -1.
18002 * config/i386/i386.c (ix86_option_override_internal):
18003 Check opts_set, not opts when setting default value of
18004 opts->x_ix86_force_align_arg_pointer.
18006 2017-08-23 Richard Biener <rguenther@suse.de>
18008 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
18009 lang_hooks.decl_printable_name.
18010 * print-rtl-function.c (print_rtx_function): Likewise.
18011 * tree-pretty-print.c (dump_function_header): Likewise.
18013 2017-08-23 Richard Biener <rguenther@suse.de>
18016 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
18017 -g0 at compile-time.
18019 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18021 PR middle-end/19706
18022 * doc/sourcebuild.texi (Other hardware attributes):
18025 2017-08-23 Tamar Christina <tamar.christina@arm.com>
18027 PR middle-end/19706
18028 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
18029 Fix single-use check.
18031 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
18033 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
18035 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18037 * doc/install.texi: Modify to add more details on running selected
18040 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
18042 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
18043 is combined with -mabi=ms.
18044 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
18047 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18049 PR tree-optimization/81488
18050 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
18051 and cached_basis fields.
18052 (MAX_SPREAD): New constant.
18053 (alloc_cand_and_find_basis): Initialize new fields.
18054 (clear_visited): New function.
18055 (create_phi_basis_1): Rename from create_phi_basis, set visited
18056 and cached_basis fields.
18057 (create_phi_basis): New wrapper function.
18058 (phi_add_costs_1): Rename from phi_add_costs, add spread
18059 parameter, set visited field, short-circuit when limits reached.
18060 (phi_add_costs): New wrapper function.
18061 (record_phi_increments_1): Rename from record_phi_increments, set
18063 (record_phi_increments): New wrapper function.
18064 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
18065 (phi_incr_cost): New wrapper function.
18066 (all_phi_incrs_profitable_1): Rename from
18067 all_phi_incrs_profitable, set visited field.
18068 (all_phi_incrs_profitable): New wrapper function.
18070 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18071 Alan Hayward <alan.hayward@arm.com>
18072 David Sherwood <david.sherwood@arm.com>
18074 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
18075 that takes the outer and inner modes.
18076 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
18077 comparison as the canonical test for a paradoxical subreg.
18078 * combine.c (simplify_set): Use paradoxical_subreg_p.
18079 (make_extraction): Likewise.
18080 (force_to_mode): Likewise.
18081 (rtx_equal_for_field_assignment_p): Likewise.
18082 (gen_lowpart_for_combine): Likewise.
18083 (simplify_comparison): Likewise.
18084 * cse.c (equiv_constant): Likewise.
18085 * expmed.c (store_bit_field_1): Likewise.
18086 * final.c (alter_subreg): Likewise.
18087 * fwprop.c (propagate_rtx): Likewise.
18088 (forward_propagate_subreg): Likewise.
18089 * ira-conflicts.c (ira_build_conflicts): Likewise.
18090 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
18091 * lra-constraints.c (curr_insn_transform): Likewise.
18092 (split_reg): Likewise.
18093 * lra-eliminations.c (move_plus_up): Likewise.
18094 (lra_eliminate_regs_1): Likewise.
18095 * recog.c (general_operand): Likewise.
18096 * ree.c (combine_reaching_defs): Likewise.
18097 * reload.c (push_reload): Likewise.
18098 (find_reloads): Likewise.
18099 * reload1.c (elimination_effects): Likewise.
18100 (compute_reload_subreg_offset): Likewise.
18101 (choose_reload_regs): Likewise.
18102 * rtlanal.c (subreg_lsb_1): Likewise.
18103 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18104 (simplify_subreg): Likewise.
18105 * var-tracking.c (track_loc_p): Likewise.
18106 * emit-rtl.c (byte_lowpart_offset): Likewise.
18107 (paradoxical_subreg_p): Delete out-of-line definition.
18109 2017-08-22 Jeff Law <law@redhat.com>
18111 PR tree-optimization/81741
18112 PR tree-optimization/71947
18113 * tree-ssa-dom.c: Include tree-inline.h.
18114 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
18115 equivalences if one is more expensive to compute than the other.
18116 * tree-ssa-scopedtables.h (class const_or_copies): Make
18117 record_const_or_copy_raw method private.
18118 (class avail_exprs_stack): New method simplify_binary_operation.
18119 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
18120 avail_exprs_stack::simplify_binary_operation as needed.
18121 (avail_exprs_stack::simplify_binary_operation): New function.
18123 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
18125 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
18126 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
18127 (DOT_SYMBOLS): Likewise.
18128 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
18129 (RELOCATABLE_NEEDS_FIXUP): Likewise.
18130 (RS6000_ABI_NAME): Likewise.
18131 (TARGET_CMODEL): Likewise.
18132 (TOC_SECTION_ASM_OP): Likewise.
18133 (SET_CMODEL): New macro.
18134 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
18136 2017-08-22 Richard Biener <rguenther@suse.de>
18138 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
18139 to allow for free-lang-data replacements similar to verify_type_variant.
18141 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18142 Alan Hayward <alan.hayward@arm.com>
18143 David Sherwood <david.sherwood@arm.com>
18145 * config/aarch64/aarch64.md (casesi): Use DImode rather than
18146 VOIDmode for the LABEL_REF.
18148 2017-08-22 Richard Biener <rguenther@suse.de>
18150 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
18152 2017-08-22 Richard Biener <rguenther@suse.de>
18154 * common.opt (feliminate-dwarf2-dups): Ignore.
18155 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
18156 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
18157 same_die_p_wrap, compute_section_prefix,
18158 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
18159 (comdat_symbol_id, comdat_symbol_number): Likewise.
18160 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
18162 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
18163 (output_die): Mark unreachable path unreachable.
18164 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
18165 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
18166 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
18167 (dwarf2out_early_finish): Likewise.
18169 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
18171 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
18173 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
18176 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
18177 not VAR_P. Filter attribute warnings with OPT_Wattributes.
18178 (avr_attribute_table) <io, io_low, address>: Initialize
18179 .decl_required with true.
18181 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
18183 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
18184 undocumented debugging options.
18185 (-mvsx-scalar-double): Likewise.
18186 (-mallow-df-permute): Likewise.
18187 (-mvectorize-builtins): Likewise.
18188 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18189 (rs6000_builtin_vectorized_function): Likewise.
18190 (rs6000_builtin_md_vectorized_function): Likewise.
18191 (rs6000_opt_vars): Likewise.
18193 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
18196 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
18197 (*btrq_imm): Rename from *btrq.
18198 (*btcq_imm): Rename from *btcq.
18199 (btsc): New code attribute.
18200 (*<btsc><mode>): New insn pattern.
18201 (*btr<mode>): Ditto.
18202 (*<btsc><mode>_mask): New insn_and_split pattern.
18203 (*btr<mode>_mask): Ditto.
18205 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18206 Alan Hayward <alan.hayward@arm.com>
18207 David Sherwood <david.sherwood@arm.com>
18209 * function.c (pad_below): Simplify padding calculation.
18211 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18212 Alan Hayward <alan.hayward@arm.com>
18213 David Sherwood <david.sherwood@arm.com>
18215 * target.def (function_prologue): Remove frame size argument.
18216 (function_epilogue): Likewise.
18217 * doc/tm.texi: Regenerate.
18218 * final.c (final_start_function): Update call to function_prologue.
18219 (final_end_function): Update call to function_epilogue.
18220 (default_function_pro_epilogue): Remove frame size argument.
18221 * output.h (default_function_pro_epilogue): Likewise.
18222 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
18223 (arm_output_function_prologue): Likewise.
18224 * config/frv/frv.c (frv_function_prologue): Likewise.
18225 (frv_function_epilogue): Likewise.
18226 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
18227 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
18228 (ia64_output_function_epilogue): Likewise.
18229 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
18230 (m32r_output_function_epilogue): Likewise.
18231 * config/microblaze/microblaze.c (microblaze_function_prologue)
18232 (microblaze_function_epilogue): Likewise.
18233 * config/mips/mips.c (mips_output_function_prologue): Likewise.
18234 (mips_output_function_epilogue): Likewise.
18235 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
18236 (mmix_target_asm_function_epilogue): Likewise.
18237 * config/msp430/msp430.c (msp430_start_function): Likewise.
18238 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
18239 (nds32_asm_function_epilogue): Likewise.
18240 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
18241 * config/pa/pa.c (pa_output_function_prologue): Likewise.
18242 (pa_output_function_epilogue): Likewise.
18243 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
18244 (rs6000_output_function_epilogue): Likewise.
18245 * config/rl78/rl78.c (rl78_start_function): Likewise.
18246 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
18247 (rs6000_output_function_epilogue): Likewise.
18248 * config/rx/rx.c (rx_output_function_prologue): Likewise.
18249 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
18250 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
18251 (sparc_asm_function_epilogue): Likewise.
18253 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18255 * tree.h (type_has_mode_precision_p): New function.
18256 * convert.c (convert_to_integer_1): Use it.
18257 * expr.c (expand_expr_real_2): Likewise.
18258 (expand_expr_real_1): Likewise.
18259 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
18260 * match.pd: Likewise.
18261 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
18262 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
18263 * tree-tailcall.c (process_assignment): Likewise.
18264 * tree-vect-loop.c (vectorizable_reduction): Likewise.
18265 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
18266 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
18267 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18268 (vectorizable_assignment): Likewise.
18269 (vectorizable_shift): Likewise.
18270 (vectorizable_operation): Likewise.
18271 * tree-vrp.c (register_edge_assert_for_2): Likewise.
18273 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
18275 * match.pd: Add pow (C, x) simplification.
18277 2017-08-21 Richard Biener <rguenther@suse.de>
18279 PR tree-optimization/81900
18280 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
18281 for blocks with abnormal predecessors.
18282 (compute_antic): Do not set visited flag prematurely.
18284 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
18287 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
18289 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18291 * stor-layout.h (vector_type_mode): Move to...
18292 * tree.h (vector_type_mode): ...here.
18293 * stor-layout.c (vector_type_mode): Move to...
18294 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
18296 2017-08-21 Richard Biener <rguenther@suse.de>
18298 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
18299 register_external_die hooks.
18300 (debug_false_tree_charstarstar_uhwistar): Declare.
18301 (debug_nothing_tree_charstar_uhwi): Likewise.
18302 * debug.c (do_nothing_debug_hooks): Adjust.
18303 (debug_false_tree_charstarstar_uhwistar): New do nothing.
18304 (debug_nothing_tree_charstar_uhwi): Likewise.
18305 * dbxout.c (dbx_debug_hooks): Adjust.
18306 (xcoff_debug_hooks): Likewise.
18307 * sdbout.c (sdb_debug_hooks): Likewise.
18308 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18309 * dwarf2out.c (macinfo_label_base): New global.
18310 (dwarf2out_register_external_die): New function for the
18311 register_external_die hook.
18312 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
18313 (dwarf2_debug_hooks): Use them.
18314 (dwarf2_lineno_debug_hooks): Adjust.
18315 (struct die_struct): Add with_offset flag.
18316 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
18317 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
18318 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
18319 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
18320 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
18321 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
18322 defining section names for the early LTO debug variants.
18323 (reset_indirect_string): New helper.
18324 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
18325 (print_dw_val): Add support for offsetted symbol references.
18326 (get_ultimate_context): Split out from is_cxx.
18327 (is_cxx): Use get_ultimate_context.
18328 (is_fortran): Add decl overload.
18329 (compute_comp_unit_symbol): Split out worker from
18330 compute_section_prefix.
18331 (compute_section_prefix): Call compute_comp_unit_symbol and
18332 set comdat_type_p here.
18333 (output_die): Skip DIE symbol output for the LTO added one.
18334 Handle DIE symbol references with offset.
18335 (output_comp_unit): Guard section name mangling properly.
18336 For LTO debug sections emit a symbol at the section beginning
18337 which we use to refer to its DIEs.
18338 (add_abstract_origin_attribute): For DIEs registered via
18339 dwarf2out_register_external_die directly refer to the early
18340 DIE rather than indirectly through the shadow one we created.
18341 Remove obsolete call to dwarf2out_abstract_function for
18342 non-function/block origins.
18343 (gen_array_type_die): When generating early LTO debug do
18344 not emit DW_AT_string_length.
18345 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
18346 late when in LTO. As suggested place a gcc_unreachable for
18347 the DECL_ABSTRACT_P case.
18348 (gen_subprogram_die): Avoid another specification DIE
18349 for early built declarations/definitions for the late LTO case.
18350 (gen_variable_die): Add type references for late duplicated VLA dies
18352 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
18353 we have the abstract instance already.
18354 (process_scope_var): Adjust decl DIE contexts in LTO which
18355 first puts them in limbo.
18356 (gen_decl_die): Do not generate type DIEs late apart from
18357 types for VLAs or for decls we do not yet have a DIE. Do not
18358 call dwarf2out_abstract_function late.
18359 (dwarf2out_early_global_decl): Make sure to create DIEs
18360 for abstract instances of a decl first.
18361 (dwarf2out_late_global_decl): Adjust comment.
18362 (output_macinfo_op): With multiple macro sections use
18363 macinfo_label_base to distinguish labels.
18364 (output_macinfo): Likewise. Update macinfo_label_base.
18365 Pass in the line info label.
18366 (note_variable_value_in_expr): When generating LTO resolve
18367 all variable values here by generating DIEs as needed.
18368 (init_sections_and_labels): Add early LTO debug flag parameter
18369 and generate different sections and names if set. Add generation
18370 counter for the labels so we can have multiple of them.
18371 (reset_dies): Helper to allow DIEs to be output multiple times.
18372 (dwarf2out_finish): When outputting DIEs to the fat part of an
18373 LTO object first reset DIEs.
18374 (dwarf2out_early_finish): Output early DIEs when generating LTO.
18375 (modified_type_die): Check for decl_ultimate_origin being self
18377 (gen_type_die_with_usage): Likewise.
18378 (gen_typedef_die): Allow decl_ultimate_origin being self.
18379 (set_decl_abstract_flags): Remove.
18380 (set_block_abstract_flags): Likewise.
18381 (dwarf2out_abstract_function): Treat the early generated DIEs
18382 as the abstract copy and only add DW_AT_inline and
18383 DW_AT_artificial here and call set_decl_origin_self.
18384 If the DIE has an abstract origin don't do anything.
18385 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
18386 if we have none yet (Go fails to build one, PR78628).
18387 (variably_modified_type_p): Prevent endless recursion for Ada
18388 cyclic pointer types.
18389 * lto-streamer-in.c: Include debug.h.
18390 (dref_queue): New global.
18391 (lto_read_tree_1): Stream in DIE references.
18392 (lto_input_tree): Register DIE references.
18393 (input_function): Stream DECL_DEBUG_ARGS.
18394 * lto-streamer-out.c: Include debug.h.
18395 (lto_write_tree_1): Output DIE references.
18396 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
18397 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
18398 (output_function): Stream DECL_DEBUG_ARGS.
18399 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
18400 Stream DECL_ABSTRACT_ORIGIN.
18401 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
18402 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
18403 DECL_CONTEXT for file-scope decls.
18404 * lto-streamer.h (struct dref_entry): Declare.
18405 (dref_queue): Likewise.
18406 * cfgexpand.c (pass_expand::execute): Do not call the
18407 outlining_inline_function hook here.
18408 * lto-wrapper.c (debug_obj): New global.
18409 (tool_cleanup): Unlink it if required.
18410 (debug_objcopy): New function.
18411 (run_gcc): Handle early debug sections in the IL files by
18412 extracting them to separate files, partially linkin them and
18413 feeding the result back as result to the linker.
18414 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
18415 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
18416 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
18417 sections into a separate segment.
18418 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
18420 (darwin_asm_dwarf_section): Likewise.
18421 (darwin_asm_output_dwarf_offset): Likewise.
18422 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
18424 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18425 Alan Hayward <alan.hayward@arm.com>
18426 David Sherwood <david.sherwood@arm.com>
18428 * read-md.h (md_reader::record_potential_iterator_use): Replace
18429 pointer argument with an rtx and an index.
18430 * read-rtl.c (iterator_group::apply_iterator): Likewise.
18431 (apply_mode_iterator): Likewise.
18432 (apply_code_iterator): Likewise.
18433 (apply_int_iterator): Likewise.
18434 (apply_subst_iterator): Likewise.
18435 (record_iterator_use): Likewise.
18436 (record_attribute_use): Likewise.
18437 (md_reader::record_potential_iterator_use): Likewise. Update calls
18438 to record_iterator_use and apply_iterator.
18439 (iterator_use): Replace ptr with x and index.
18440 (attribute_use): Likewise.
18441 (apply_attribute_uses): Update calls to apply_iterator.
18442 (apply_iterators): Likewise. Update initialization of iterator_use.
18443 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
18444 and record_potential_iterator_use.
18445 (rtx_reader::read_rtx_operand): Likewise.
18447 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18448 Alan Hayward <alan.hayward@arm.com>
18449 David Sherwood <david.sherwood@arm.com>
18451 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
18454 2017-08-21 Richard Biener <rguenther@suse.de>
18456 PR middle-end/81884
18457 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
18458 at struct end conservatively when comparing common bases.
18460 2017-08-21 Richard Biener <rguenther@suse.de>
18462 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
18463 (mem_ref_in_stmt): Remove.
18464 (determine_max_movement): Use ref index to get at the reference.
18465 (invariantness_dom_walker::before_dom_children): Deal with
18466 lim data already initialized.
18467 (gather_mem_refs_stmt): Initialize lim data and record ref index.
18469 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18471 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
18472 (TARGET_ISA_ROUND): Ditto.
18473 (TARGET_ROUND): Ditto.
18474 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
18475 * config/i386/i386.md: Ditto.
18476 * config/i386/sse.md: Ditto.
18477 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
18478 with OPTION_MASK_ISA_SSE4_1.
18480 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18483 * doc/extend.texi (x86 Built-in Functions): Correct the name of
18484 __builtin_ia32_lzcnt_u16.
18486 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
18489 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
18490 (rs6000_set_current_function): Rewrite function to use it.
18492 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
18495 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
18496 and TYPE_WARN_IF_NOT_ALIGN.
18497 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
18498 (handle_warn_if_not_align): New.
18499 (place_union_field): Call handle_warn_if_not_align.
18500 (place_field): Call handle_warn_if_not_align.
18501 Copy TYPE_WARN_IF_NOT_ALIGN.
18502 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
18503 (layout_type): Likewise.
18504 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
18506 (tree_decl_common): Add warn_if_not_align.
18507 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
18508 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
18509 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
18510 (DECL_WARN_IF_NOT_ALIGN): Likewise.
18511 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
18512 * doc/extend.texi: Document warn_if_not_aligned attribute.
18513 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
18515 2017-08-17 Martin Liska <mliska@suse.cz>
18518 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
18519 (get_data_dependence): Use it as pointer type.
18520 (distribute_loop): Likewise.
18522 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18524 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
18525 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
18526 (p8_vmrgow_<mode>_direct): New define_insn.
18527 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
18528 handle endianness for vmrgew and vmrgow permute patterns.
18530 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18532 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
18533 * config/rs6000/rs6000-cpus.def: Remove comment.
18534 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
18535 (POWERPC_MASKS): Likewise.
18536 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
18537 use of TARGET_VSX_TIMODE.
18538 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
18539 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
18540 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
18541 (rs6000_option_override_internal): Remove dead code.
18542 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
18543 (rs6000_legitimize_reload_address): Likewise.
18544 (rs6000_legitimate_address_p): Likewise.
18545 (rs6000_opt_masks): Delete "vsx-timode".
18546 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
18547 from function comment.
18548 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
18549 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
18550 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
18552 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
18553 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
18554 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
18555 (VSX_TI): Likewise.
18557 (define_peephole2): Likewise.
18559 2017-08-17 Martin Sebor <msebor@redhat.com>
18562 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
18563 past the end of an array.
18564 (test_pp_format): Add test cases.
18566 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
18568 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
18569 missing ECF_NOTHROW flags.
18571 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18574 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
18575 operands residing in integer registers.
18576 (*vsx_le_perm_load_<mode>): Likewise.
18577 (*vsx_le_perm_store_<mode>): Likewise.
18578 (define_peephole2): Add peepholes to optimize the above.
18580 2017-08-17 Marek Polacek <polacek@redhat.com>
18582 PR middle-end/81814
18583 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
18584 to mimic what shorten_compare did. Change the return type to bool.
18585 (fold_cond_expr_with_comparison): Update call to
18586 operand_equal_for_comparison_p.
18587 (fold_ternary_loc): Likewise.
18589 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
18591 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
18593 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
18595 2017-08-17 Richard Biener <rguenther@suse.de>
18597 * tree-ssa-structalias.c (solve_graph): When propagating
18598 to successors update the graphs succ edges and avoid duplicate work.
18600 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
18603 * config/i386/i386.c (ix86_option_override_internal): Save target
18604 specific options after ix86_stack_protector_guard_reg was changed.
18606 2017-08-17 Richard Biener <rguenther@suse.de>
18608 PR tree-optimization/81827
18609 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
18610 (new_var_info): Initialize it conservatively.
18611 (get_call_vi): Mark register vars.
18612 (new_scalar_tmp_constraint_exp): Likewise.
18613 (handle_rhs_call): Likewise.
18614 (handle_const_call): Likewise.
18615 (create_function_info_for): Likewise.
18616 (solve_constraints): Sort varinfos to separate register from
18617 non-register vars to pack points-to solution bitmaps during
18620 2017-08-17 Marek Polacek <polacek@redhat.com>
18622 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
18624 2017-08-17 Richard Biener <rguenther@suse.de>
18626 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
18627 to true when overflow is undefined and we saturated the result.
18629 2017-08-17 Alan Modra <amodra@gmail.com>
18632 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
18633 Don't use store multiple if only one reg needs saving.
18634 (interesting_frame_related_regno): New function.
18635 (rs6000_frame_related): Don't emit frame info for regs that
18637 (rs6000_emit_epilogue): Likewise.
18639 2017-08-16 Nathan Sidwell <nathan@acm.org>
18641 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
18642 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
18643 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
18644 * tree.c (free_lang_data_in_type): Use else-if chain. Always
18645 clear TYPE_LANG_1. Remove obsolete member-function stripping.
18646 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
18647 (verify_type): Adjust for TYPE_BINFO move.
18648 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
18649 process TYPE_BINFO directly.
18650 (hash_tree): Likewise.
18651 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18653 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18656 2017-08-16 David Malcolm <dmalcolm@redhat.com>
18658 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
18660 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
18663 * config/i386/i386.md (*anddi_1_btr): Change predicates of
18664 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
18665 Add ix86_binary_operator_ok to insn constraint.
18666 (*iordi_1_bts): Ditto.
18667 (*xordi_1_btc): Ditto.
18668 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
18669 Update corresponding peephole2 pattern.
18673 2017-08-16 Bin Cheng <bin.cheng@arm.com>
18675 PR tree-optimization/81832
18676 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
18677 copy loop header which has IFN_LOOP_DIST_ALIAS call.
18679 2017-08-16 Marek Polacek <polacek@redhat.com>
18682 * fold-const.c (fold_indirect_ref_1): Restore original behavior
18683 regarding size_zero_node.
18685 2017-08-16 Martin Liska <mliska@suse.cz>
18688 * config.gcc: Respect previously set extra_objs in case
18691 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
18693 PR tree-optimization/81835
18694 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
18695 the phi SSA_NAME. Check that the condition in a COND_EXPR does
18696 not depend on the phi.
18698 2017-08-16 Alan Modra <amodra@gmail.com>
18700 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
18703 2017-08-16 Alan Modra <amodra@gmail.com>
18705 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
18706 (save_reg_p): ..into this. Update all callers.
18707 (first_reg_to_save): Simplify.
18709 2017-08-16 Alan Modra <amodra@gmail.com>
18711 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
18714 2017-08-15 Joseph Myers <joseph@codesourcery.com>
18718 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
18719 constant count if that count is less than 32.
18721 2017-08-15 Nathan Sidwell <nathan@acm.org>
18723 * gcc.c (execute): Emit friendlier message if inferior is killed
18724 by an external cause.
18726 2017-08-15 Richard Biener <rguenther@suse.de>
18728 PR tree-optimization/81790
18729 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
18730 CONSTRUCTORs from simplifying and VN.
18732 2017-08-14 Martin Sebor <msebor@redhat.com>
18734 * builtin-attrs.def: Add comments.
18736 2017-08-14 Martin Sebor <msebor@redhat.com>
18739 * doc/extend.texi (attribute nonstring): Document new attribute.
18741 2017-08-14 Martin Sebor <msebor@redhat.com>
18744 * tree-diagnostic.c (default_tree_printer): Handle %G.
18745 * gimple-pretty-print.h (percent_G_format): Declare new function.
18746 * gimple-pretty-print.c (percent_G_format): Define.
18747 * tree-pretty-print.c (percent_K_format): Add argument.
18749 2017-08-14 Martin Sebor <msebor@redhat.com>
18751 PR translation/79998
18752 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
18753 Remove a stray space.
18755 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
18758 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
18759 (*iordi_1_bts): Ditto.
18760 (*xordi_1_btc): Ditto.
18762 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18765 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
18766 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18768 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
18769 quoted strings, and make more translator-friendly.
18770 (darwin_rs6000_override_options): Likewise.
18771 (rs6000_option_override_internal): Likewise.
18772 (rs6000_return_in_memory): Fix overlong line.
18773 (init_cmulative_args): Use quoted strings, and make more
18774 translator-friendly.
18775 (rs6000_pass_by_reference): Fix overlong line.
18776 (def_builtin): Use quoted strings.
18777 (altivec_expand_predicate_builtin): Use quoted strings, and make
18778 more translator-friendly.
18779 (htm_expand_builtin): Use quoted strings.
18780 (cpu_expand_builtin): Use quoted strings, and make more
18781 translator-friendly.
18782 (altivec_expand_builtin): Likewise.
18783 (paired_expand_predicate_builtin): Likewise.
18784 (rs6000_invalid_builtin): Likewise.
18785 (builtin_function_type): Use quoted strings.
18786 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
18787 more translator-friendly.
18788 (rs6000_trampoline_init): Likewise.
18789 (rs6000_handle_altivec_attribute): Likewise.
18790 (rs6000_inner_target_options): Use quoted strings.
18791 (rs6000_disable_incompatible_switches): Likewise.
18792 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
18793 strings, and make more translator-friendly.
18794 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
18796 2017-08-14 Bin Cheng <bin.cheng@arm.com>
18798 PR tree-optimization/81799
18799 * tree-loop-distribution.c (version_loop_by_alias_check): Force
18800 cond_expr to simple gimple operand.
18802 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
18804 PR middle-end/46932
18805 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
18807 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
18811 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
18812 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
18813 TARGET_GASISR_PROLOGUES.
18814 * config/avr/avr.c (avr_option_override): Same.
18815 (avr_pass_pre_proep::execute): Same.
18817 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
18820 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
18821 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
18823 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
18825 * config/i386/i386.md (*load_tp_<mode>): Redefine as
18826 define_insn_and_split. Split to a memory load from 0 in
18827 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
18828 using PTR mode iterator.
18829 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
18830 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
18831 (*add_tp_<mode>): Redefine as define_insn_and_split.
18832 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
18833 address space. Merge with *add_tp_x32 using PTR mode iterator.
18834 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
18835 Split to an add with a memory load from 0 in
18836 DEFAULT_TLS_SEG_REG address space.
18838 2017-08-12 Andrew Pinski <apinski@cavium.com>
18840 * config/aarch64/aarch64-option-extensions.def (rdma):
18841 Fix feature string to what Linux prints out in /proc/cpuinfo.
18843 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
18846 * dwarf2out.c (modified_type_die): For C typedef types that have
18847 an ultimate origin, process the ultimate origin instead of the
18849 (gen_typedef_die): Assert that input DECLs have no ultimate
18851 (gen_type_die_with_usage): For typedef variants that have an
18852 ultimate origin, just call gen_decl_die on the original DECL.
18853 (process_scope_var): Avoid creating DIEs for local typedefs and
18854 concrete static variables.
18856 2017-08-12 Alan Modra <amodra@gmail.com>
18860 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
18861 match gnu-user.h startfile.
18862 (ENDFILE_LINUX_SPEC): Similarly.
18864 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
18867 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
18869 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
18871 2017-08-11 Tamar Christina <tamar.christina@arm.com>
18872 * config/aarch64/aarch64.md (mov<mode>): Change.
18873 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18874 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
18875 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
18877 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
18879 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
18880 for storage order barriers.
18882 2017-08-11 Martin Liska <mliska@suse.cz>
18885 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
18886 variables of void type.
18888 2017-08-11 Martin Liska <mliska@suse.cz>
18890 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
18891 TARGET_SUPPORTS_ALIASES.
18892 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
18893 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
18894 (optimize_weakref): Likewise.
18895 * symtab.c (symtab_node::noninterposable_alias): Likewise.
18896 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
18897 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
18899 2017-08-11 Martin Liska <mliska@suse.cz>
18902 * config/i386/i386.c (make_resolver_func): Do complete
18903 refactoring of the function.
18905 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18908 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
18909 * config/i386/i386.c (ix86_stack_protect_guard): Use
18910 ix86_stack_protect_guard_symbol_str to generate varible declaration.
18911 * doc/invoke.texi (x86 Options): Document
18912 -mstack-protector-guard-symbol= option.
18914 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18916 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
18917 * config/i386/i386.c (ix86_split_stack_guard): New function.
18918 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
18919 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
18920 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
18921 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
18922 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
18923 (split_stack_space_check): Call ix86_split_stack_guard.
18925 2017-08-10 Martin Sebor <msebor@redhat.com>
18927 * print-tree.c (print_node): Print location using the established
18929 Replace spaces with colons.
18930 (debug_raw, debug): Ditto.
18932 2017-08-10 Martin Sebor <msebor@redhat.com>
18935 * pretty-print.c (pp_format): Correct the handling of %s precision.
18937 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18940 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18942 (ix86_finalize_stack_frame_flags): This. Also clear
18943 frame_pointer_needed if -fno-omit-frame-pointer is used without
18945 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18946 with ix86_finalize_stack_frame_flags.
18947 (ix86_expand_epilogue): Likewise.
18948 (ix86_expand_split_stack_prologue): Likewise.
18949 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18951 2017-08-10 Martin Liska <mliska@suse.cz>
18954 * c-attribs.c (handle_target_attribute):
18955 Report warning for an empty string argument of target attribute.
18957 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18960 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18962 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18963 or DECL_NONLOCAL labels.
18964 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18965 or DECL_NONLOCAL labels here.
18967 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18969 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18970 to indicate when early gimple folding has been disabled.
18971 (rs6000_gimple_fold_builtin): Add debug content.
18972 (rs6000_invalid_builtin): Fix whitespace.
18973 (rs6000_expand_builtin): Fix whitespace.
18974 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18976 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18979 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18980 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18982 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18984 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18985 (qdf24xx): Likewise.
18986 * config/aarch64/aarch64.md: Include falkor.md.
18987 * config/aarch64/falkor.md: New.
18989 2017-08-09 Marek Polacek <polacek@redhat.com>
18992 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18993 * diagnostic.c (emit_diagnostic): Add a comment.
18994 (emit_diagnostic_valist): New function.
18996 2017-08-09 Marek Polacek <polacek@redhat.com>
18999 * input.c (make_location): New overload.
19000 * input.h (make_location): Declare.
19002 2017-08-08 Alan Modra <amodra@gmail.com>
19003 H.J. Lu <hongjiu.lu@intel.com>
19006 * gcc.c (NO_PIE_SPEC): Delete.
19007 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
19009 (LINK_PIE_SPEC): ..to here.
19010 (LINK_COMMAND_SPEC): Support -no-pie.
19011 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
19012 chain of crtbegin*.o selection, update for PIE_SPEC changes and
19014 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
19015 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
19016 (ENDFILE_CRTEND_SPEC): Similarly.
19018 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
19021 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
19022 (mstack-protector-guard-offset=): Ditto.
19023 * config/i386/i386.c (ix86_option_override): Handle
19024 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
19026 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
19027 ix86_stack_protect_guard_offset variables.
19028 (TARGET_STACK_PROTECT_GUARD): Always define.
19029 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
19030 and -mstack-protector-guard-offset= options.
19032 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19034 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
19035 boundary case for the last candidate.
19037 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19039 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
19040 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
19042 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19044 PR middle-end/19706
19045 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
19046 * config/aarch64/aarch64-builtins.c
19047 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
19048 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
19049 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
19051 2017-08-08 Tamar Christina <tamar.christina@arm.com>
19052 Andrew Pinski <pinskia@gmail.com>
19054 PR middle-end/19706
19055 * internal-fn.def (XORSIGN): New.
19056 * optabs.def (xorsign_optab): New.
19057 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
19058 (convert_expand_mult_copysign): New.
19059 (pass_optimize_widening_mul::execute): Call
19060 convert_expand_mult_copysign.
19062 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19064 PR tree-optimization/81354
19065 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
19066 Insert on edges rather than explicitly creating landing pads.
19067 (analyze_candidates_and_replace): Commit edge inserts.
19069 2017-08-08 Richard Biener <rguenther@suse.de>
19071 PR middle-end/81719
19072 * tree-ssa-loop-niter.c: Include tree-dfa.h.
19073 (expand_simple_operations): Also look through ADDR_EXPRs with
19074 MEM_REF bases treating them as POINTER_PLUS_EXPR.
19076 2017-08-08 Richard Biener <rguenther@suse.de>
19078 PR tree-optimization/81723
19079 * tree-vect-slp.c (struct bst_traits): New hash traits.
19080 (bst_fail): New global.
19081 (vect_build_slp_tree_2): New worker, split out from ...
19082 (vect_build_slp_tree): ... this now wrapping it with using
19083 bst_fail set to cache SLP tree build fails. Properly handle
19085 (vect_analyze_slp_instance): Allocate and free bst_fail.
19087 2017-08-08 Martin Liska <mliska@suse.cz>
19090 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
19091 LABEL_DECLs that can be from a different function.
19093 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19095 PR tree-optimization/81744
19096 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
19097 loop's number of iterations.
19099 2017-08-08 Martin Liska <mliska@suse.cz>
19101 * asan.c: Include header files.
19102 * attribs.c (build_decl_attribute_variant): New function moved
19104 (build_type_attribute_qual_variant): Likewise.
19105 (cmp_attrib_identifiers): Likewise.
19106 (simple_cst_list_equal): Likewise.
19107 (omp_declare_simd_clauses_equal): Likewise.
19108 (attribute_value_equal): Likewise.
19109 (comp_type_attributes): Likewise.
19110 (build_type_attribute_variant): Likewise.
19111 (lookup_ident_attribute): Likewise.
19112 (remove_attribute): Likewise.
19113 (merge_attributes): Likewise.
19114 (merge_type_attributes): Likewise.
19115 (merge_decl_attributes): Likewise.
19116 (merge_dllimport_decl_attributes): Likewise.
19117 (handle_dll_attribute): Likewise.
19118 (attribute_list_equal): Likewise.
19119 (attribute_list_contained): Likewise.
19120 * attribs.h (lookup_attribute): New function moved from tree.[ch].
19121 (lookup_attribute_by_prefix): Likewise.
19122 * bb-reorder.c: Include header files.
19123 * builtins.c: Likewise.
19124 * calls.c: Likewise.
19125 * cfgexpand.c: Likewise.
19126 * cgraph.c: Likewise.
19127 * cgraphunit.c: Likewise.
19128 * convert.c: Likewise.
19129 * dwarf2out.c: Likewise.
19130 * final.c: Likewise.
19131 * fold-const.c: Likewise.
19132 * function.c: Likewise.
19133 * gimple-expr.c: Likewise.
19134 * gimple-fold.c: Likewise.
19135 * gimple-pretty-print.c: Likewise.
19136 * gimple.c: Likewise.
19137 * gimplify.c: Likewise.
19138 * hsa-common.c: Likewise.
19139 * hsa-gen.c: Likewise.
19140 * internal-fn.c: Likewise.
19141 * ipa-chkp.c: Likewise.
19142 * ipa-cp.c: Likewise.
19143 * ipa-devirt.c: Likewise.
19144 * ipa-fnsummary.c: Likewise.
19145 * ipa-inline.c: Likewise.
19146 * ipa-visibility.c: Likewise.
19148 * lto-cgraph.c: Likewise.
19149 * omp-expand.c: Likewise.
19150 * omp-general.c: Likewise.
19151 * omp-low.c: Likewise.
19152 * omp-offload.c: Likewise.
19153 * omp-simd-clone.c: Likewise.
19154 * opts-global.c: Likewise.
19155 * passes.c: Likewise.
19156 * predict.c: Likewise.
19157 * sancov.c: Likewise.
19158 * sanopt.c: Likewise.
19159 * symtab.c: Likewise.
19160 * toplev.c: Likewise.
19161 * trans-mem.c: Likewise.
19162 * tree-chkp.c: Likewise.
19163 * tree-eh.c: Likewise.
19164 * tree-into-ssa.c: Likewise.
19165 * tree-object-size.c: Likewise.
19166 * tree-parloops.c: Likewise.
19167 * tree-profile.c: Likewise.
19168 * tree-ssa-ccp.c: Likewise.
19169 * tree-ssa-live.c: Likewise.
19170 * tree-ssa-loop.c: Likewise.
19171 * tree-ssa-sccvn.c: Likewise.
19172 * tree-ssa-structalias.c: Likewise.
19173 * tree-ssa.c: Likewise.
19174 * tree-streamer-in.c: Likewise.
19175 * tree-vectorizer.c: Likewise.
19176 * tree-vrp.c: Likewise.
19177 * tsan.c: Likewise.
19178 * ubsan.c: Likewise.
19179 * varasm.c: Likewise.
19180 * varpool.c: Likewise.
19181 * tree.c: Remove functions moved to attribs.[ch].
19182 * tree.h: Likewise.
19183 * config/aarch64/aarch64.c: Add attrs.h header file.
19184 * config/alpha/alpha.c: Likewise.
19185 * config/arc/arc.c: Likewise.
19186 * config/arm/arm.c: Likewise.
19187 * config/avr/avr.c: Likewise.
19188 * config/bfin/bfin.c: Likewise.
19189 * config/c6x/c6x.c: Likewise.
19190 * config/cr16/cr16.c: Likewise.
19191 * config/cris/cris.c: Likewise.
19192 * config/darwin.c: Likewise.
19193 * config/epiphany/epiphany.c: Likewise.
19194 * config/fr30/fr30.c: Likewise.
19195 * config/frv/frv.c: Likewise.
19196 * config/ft32/ft32.c: Likewise.
19197 * config/h8300/h8300.c: Likewise.
19198 * config/i386/winnt.c: Likewise.
19199 * config/ia64/ia64.c: Likewise.
19200 * config/iq2000/iq2000.c: Likewise.
19201 * config/lm32/lm32.c: Likewise.
19202 * config/m32c/m32c.c: Likewise.
19203 * config/m32r/m32r.c: Likewise.
19204 * config/m68k/m68k.c: Likewise.
19205 * config/mcore/mcore.c: Likewise.
19206 * config/microblaze/microblaze.c: Likewise.
19207 * config/mips/mips.c: Likewise.
19208 * config/mmix/mmix.c: Likewise.
19209 * config/mn10300/mn10300.c: Likewise.
19210 * config/moxie/moxie.c: Likewise.
19211 * config/msp430/msp430.c: Likewise.
19212 * config/nds32/nds32-isr.c: Likewise.
19213 * config/nds32/nds32.c: Likewise.
19214 * config/nios2/nios2.c: Likewise.
19215 * config/nvptx/nvptx.c: Likewise.
19216 * config/pa/pa.c: Likewise.
19217 * config/pdp11/pdp11.c: Likewise.
19218 * config/powerpcspe/powerpcspe.c: Likewise.
19219 * config/riscv/riscv.c: Likewise.
19220 * config/rl78/rl78.c: Likewise.
19221 * config/rx/rx.c: Likewise.
19222 * config/s390/s390.c: Likewise.
19223 * config/sh/sh.c: Likewise.
19224 * config/sol2.c: Likewise.
19225 * config/sparc/sparc.c: Likewise.
19226 * config/spu/spu.c: Likewise.
19227 * config/stormy16/stormy16.c: Likewise.
19228 * config/tilegx/tilegx.c: Likewise.
19229 * config/tilepro/tilepro.c: Likewise.
19230 * config/v850/v850.c: Likewise.
19231 * config/vax/vax.c: Likewise.
19232 * config/visium/visium.c: Likewise.
19233 * config/xtensa/xtensa.c: Likewise.
19235 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19238 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
19239 constraints since the -mupper-regs-* switches have been
19241 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
19242 into a vector from a double word element that was extracted from
19243 another vector, and eliminate extra XXPERMDI instructions.
19244 (vsx_concat_<mode>_2): Likewise.
19245 (vsx_concat_<mode>_3): Likewise.
19246 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
19247 concat to allow optimizing inserts from previous extracts.
19249 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
19251 * config/i386/i386.c (ix86_stack_protect_guard): Generate
19252 memory reference to a SSP offset in TLS address space.
19253 (ix86_print_operand) <case '@'>: Remove.
19254 (ix86_print_operand_punct_valid_p): Remove '@' code.
19255 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
19256 UNSPEC_SP_TLS_TEST.
19257 (stack_tls_protect_set_<mode>): Remove.
19258 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
19259 (stack_tls_protect_test_<mode>): Remove.
19260 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
19262 2017-08-07 Olivier Hainque <hainque@adacore.com>
19265 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
19267 2017-08-07 Douglas Rupp <rupp@adacore.com>
19269 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
19270 variable was referenced as multidir in command.
19272 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19275 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
19278 2017-08-07 Martin Liska <mliska@suse.cz>
19280 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
19281 * config/rl78/rl78.c: Add include of attribs.h.
19282 * config/sh/sh.c: Likewise.
19283 * config/v850/v850.c: Likewise.
19285 2017-08-07 Tom de Vries <tom@codesourcery.com>
19287 PR middle-end/78266
19288 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
19290 2017-08-07 Martin Liska <mliska@suse.cz>
19292 * config/mips/mips.c: Include attribs.h.
19294 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19297 * doc/invoke.texi: Document change in behvaior for -Ofast for
19300 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
19302 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
19304 (aarch64_pop_regs): Likewise.
19305 (aarch64_gen_load_pair): Likewise.
19306 (aarch64_save_callee_saves): Likewise.
19307 (aarch64_restore_callee_saves): Likewise.
19309 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19311 * config/i386/i386.c: Revert the last change.
19313 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19316 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19318 (ix86_finalize_stack_frame_flags): This. Also clear
19319 frame_pointer_needed if -fno-omit-frame-pointer is used without
19321 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19322 with ix86_finalize_stack_frame_flags.
19323 (ix86_expand_epilogue): Likewise.
19324 (ix86_expand_split_stack_prologue): Likewise.
19326 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19329 * config/i386/i386.c (get_builtin_code_for_version): Set priority
19330 to P_AES for Westmere.
19332 2017-08-07 Jonathan Yong <10walls@gmail.com>
19334 * config/i386/mingw.opt (fset-stack-executable): Removed.
19335 * config/i386/cygming.opt (fset-stack-executable): Moved
19337 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
19339 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
19341 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
19343 2017-08-07 Marek Polacek <polacek@redhat.com>
19345 PR middle-end/81737
19346 * fold-const.c (fold_indirect_ref_1): Check type_domain.
19348 2017-08-07 Martin Liska <mliska@suse.cz>
19350 * attribs.h (canonicalize_attr_name): New function.
19351 (cmp_attribs): Move from c-format.c and adjusted.
19352 (is_attribute_p): Moved from tree.h.
19353 * tree-inline.c: Add new includes.
19354 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
19355 (private_is_attribute_p): Remove.
19356 (private_lookup_attribute): Likewise.
19357 (private_lookup_attribute_by_prefix): Simplify.
19358 (remove_attribute): Use is_attribute_p.
19359 * tree.h: Remove removed declarations.
19361 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19363 PR middle-end/81698
19364 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
19365 instead of computing it in the function. Formatting fix.
19366 (expand_case): Don't rely on default_edge being the first edge,
19367 clear it if removing it, pass default_edge to
19368 emit_case_dispatch_table.
19369 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
19372 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
19374 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
19375 insn in the function, emit NOP after the insn.
19377 2017-08-06 Tom de Vries <tom@codesourcery.com>
19379 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
19382 2017-08-06 Tom de Vries <tom@codesourcery.com>
19384 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
19387 2017-08-04 Yury Gribov <tetra2005@gmail.com>
19389 PR tree-optimization/57371
19390 * match.pd: New pattern.
19392 2017-08-04 Marek Polacek <polacek@redhat.com>
19394 PR middle-end/81695
19395 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
19396 perform the computation in offset_int.
19398 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19400 PR tree-optimization/81136
19401 * tree-vectorizer.h: Include tree-hash-traits.h.
19402 (vec_base_alignments): New typedef.
19403 (vec_info): Add a base_alignments field.
19404 (vect_record_base_alignments): Declare.
19405 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
19407 (DR_IS_CONDITIONAL_IN_STMT): New macro.
19408 (create_data_ref): Add an is_conditional_in_stmt argument.
19409 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
19410 the is_conditional_in_stmt field.
19411 (data_ref_loc): Add an is_conditional_in_stmt field.
19412 (get_references_in_stmt): Set the is_conditional_in_stmt field.
19413 (find_data_references_in_stmt): Update call to create_data_ref.
19414 (graphite_find_data_references_in_stmt): Likewise.
19415 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
19416 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
19417 (vect_record_base_alignment): New function.
19418 (vect_record_base_alignments): Likewise.
19419 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
19420 for nested statements even if we fail to compute a misalignment.
19421 Use pooled base alignments for unconditional references.
19422 (vect_find_same_alignment_drs): Compare base addresses instead
19424 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
19425 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
19427 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19429 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
19430 Add an explicit name for the enum. Use auto_vec for slp_instances
19431 and grouped_stores.
19432 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
19434 (_bb_vec_info): Add a constructor and destructor.
19435 (vinfo_for_stmt): Return NULL for uids of -1 as well.
19436 (destroy_loop_vec_info): Delete.
19437 (vect_destroy_datarefs): Likewise.
19438 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
19439 (vec_info::vec_info): New function.
19440 (vec_info::~vec_info): Likewise.
19441 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
19442 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
19443 destroy_loop_vec_info.
19444 * tree-vect-loop.c (new_loop_vec_info): Replace with...
19445 (_loop_vec_info::_loop_vec_info): ...this.
19446 (destroy_loop_vec_info): Replace with...
19447 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
19448 the stmt_vec_infos. Leave handling of vec_info information to its
19449 destructor. Remove explicit vector releases.
19450 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
19451 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
19452 * tree-vect-slp.c (new_bb_vec_info): Replace with...
19453 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
19454 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
19455 (destroy_bb_vec_info): Replace with...
19456 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
19457 information to its destructor.
19458 (vect_slp_analyze_bb_1): Use new and delete instead of
19459 new_bb_vec_info and destroy_bb_vec_info.
19460 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
19463 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19465 * tree-data-ref.h (subscript): Add access_fn field.
19466 (data_dependence_relation): Add could_be_independent_p.
19467 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
19468 (same_access_functions): Move to tree-data-ref.c.
19469 * tree-data-ref.c (ref_contains_union_access_p): New function.
19470 (access_fn_component_p): Likewise.
19471 (access_fn_components_comparable_p): Likewise.
19472 (dr_analyze_indices): Add a reference to access_fn_component_p.
19473 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
19475 (constant_access_functions): Likewise.
19476 (add_other_self_distances): Likewise.
19477 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
19478 (initialize_data_dependence_relation): Use XCNEW and remove
19479 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
19480 of access functions that have the same type. Allow the
19481 subsequence to end with different bases in some circumstances.
19482 Record the chosen access functions in SUB_ACCESS_FN.
19483 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
19484 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
19485 (subscript_dependence_tester_1): Likewise dra and drb.
19486 (build_classic_dist_vector): Update calls accordingly.
19487 (subscript_dependence_tester): Likewise.
19488 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
19489 DDR_COULD_BE_INDEPENDENT_P.
19490 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
19491 comp_alias_ddrs instead of may_alias_ddrs.
19492 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
19494 (vect_analyze_data_ref_dependence): Use it if
19495 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
19496 distance vectors if that fails.
19497 (dependence_distance_ge_vf): New function.
19498 (vect_prune_runtime_alias_test_list): Use it. Don't clear
19499 LOOP_VINFO_MAY_ALIAS_DDRS.
19501 2017-08-04 Richard Biener <rguenther@suse.de>
19503 PR middle-end/81705
19504 * fold-const.c (fold_binary_loc): Properly restrict
19505 minus_var0 && minus_var1 case when associating undefined overflow
19508 2017-08-04 Tom de Vries <tom@codesourcery.com>
19510 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
19512 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19514 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19515 Don't start diagnostic messages with a capital letter.
19516 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19518 (rs6000_invalid_builtin): Likewise.
19519 (rs6000_trampoline_init): Likewise.
19521 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19524 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
19525 after setting changeable df flags.
19527 2017-08-03 Richard Biener <rguenther@suse.de>
19529 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
19530 up if the use is in USE - X.
19532 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
19534 * toplev.c (dumpfile.h): New include.
19535 (internal_error_reentered): New static function. Use it...
19536 (internal_error_function): ...here to handle reentered internal_error.
19538 2017-08-03 Richard Biener <rguenther@suse.de>
19540 PR middle-end/81148
19541 * fold-const.c (split_tree): Add minus_var and minus_con
19542 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
19543 here but always use minus_*.
19544 (associate_trees): Assert we never associate with MINUS_EXPR
19545 and NULL first operand. Do not recurse for PLUS_EXPR operands
19546 when associating as MINUS_EXPR either.
19547 (fold_binary_loc): Track minus_var and minus_con.
19549 2017-08-03 Tom de Vries <tom@codesourcery.com>
19552 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
19553 ACCEL_COMPILER, apply finish_options on
19554 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
19556 2017-08-03 Tom de Vries <tom@codesourcery.com>
19559 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
19560 function_entry_patch_area_size > 0.
19562 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19565 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
19566 instead of 10??LU, perform unit multiplication in wide_int,
19567 don't change alloc_object_size_limit if the limit is larger
19570 PR tree-optimization/81655
19571 PR tree-optimization/81588
19572 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
19573 the case when ranges[i].low and high are 1 for unsigned type with
19576 PR middle-end/81052
19577 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
19578 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
19580 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19582 * tree-vrp.h: Add include guard.
19584 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
19587 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
19588 (ud2): New insn pattern.
19589 * config/i386/i386.c (ix86_expand_epilogue):
19590 For naked functions, generate ud2 instead of trap insn.
19592 2017-08-02 Marek Polacek <polacek@redhat.com>
19595 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
19597 2017-08-02 Tom de Vries <tom@codesourcery.com>
19598 Cesar Philippidis <cesar@codesourcery.com>
19600 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
19601 Add missing edge probabilities.
19603 2017-08-02 Tamar Christina <tamar.christina@arm.com>
19605 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
19606 Correct endianness.
19608 2017-08-02 Jakub Jelinek <jakub@redhat.com>
19610 PR middle-end/79499
19611 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
19612 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
19613 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
19614 prologue_seq sequences - if any.
19616 2017-08-02 Richard Biener <rguenther@suse.de>
19618 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
19619 via vectors if supported, integer extracts via punning if supported
19620 or otherwise vector extracts.
19622 2017-08-02 Richard Biener <rguenther@suse.de>
19624 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
19626 (bitmap_insert_into_set): ... this.
19628 2017-08-02 Richard Biener <rguenther@suse.de>
19630 PR tree-optimization/81633
19632 2015-08-17 Alan Hayward <alan.hayward@arm.com>
19634 PR tree-optimization/71752
19635 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
19637 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
19639 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
19640 (machine_function::call_ms2sysv_pad_out): Remove field.
19641 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
19642 (ix86_compute_frame_layout): Likewise.
19644 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
19647 * config/i386/i386.c (ix86_set_func_type): Disallow naked
19648 attribute with interrupt attribute.
19650 2017-08-01 Andrew Pinski <apinski@cavium.com>
19652 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
19653 BIT_INSERT_EXPR's operand 1
19654 to see if the types precision matches.
19656 2017-08-01 Martin Liska <mliska@suse.cz>
19658 PR middle-end/70140
19659 * builtins.c (expand_builtin_memcpy_args): Remove.
19660 (expand_builtin_memcpy): Call newly added function
19661 expand_builtin_memory_copy_args.
19662 (expand_builtin_memcpy_with_bounds): Likewise.
19663 (expand_builtin_mempcpy): Remove last argument.
19664 (expand_builtin_mempcpy_with_bounds): Likewise.
19665 (expand_builtin_memory_copy_args): New function created from
19666 expand_builtin_mempcpy_args with small modifications.
19667 (expand_builtin_mempcpy_args): Remove.
19668 (expand_builtin_stpcpy): Remove unused argument.
19669 (expand_builtin): Likewise.
19670 (expand_builtin_with_bounds): Likewise.
19672 2017-08-01 Martin Liska <mliska@suse.cz>
19675 Make mempcpy more optimal (PR middle-end/70140).
19677 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19680 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
19681 __builtin_vec_cmpne verify both arguments are compatible vectors
19682 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
19683 verify arg1_type is a pointer or array type. For __builtin_vec_st,
19684 move computation of aligned to after checking the argument types.
19688 * config/rs6000/vsx.md (vextract_fp_from_shorth,
19689 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
19692 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
19693 Doug Rupp <rupp@adacore.com>
19694 Olivier Hainque <hainque@adacore.com>
19696 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
19697 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
19699 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
19700 for TARGET_OS_CPP_BUILTIN.
19701 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
19702 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
19704 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
19705 passing required abi options to the assembler for EABI configurations.
19706 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
19707 of .text.hot and .text.unlikely sections for kernel modules when
19708 using ARM style exceptions.
19709 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
19710 options. Add EXTRA_CC1_SPEC.
19711 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
19713 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
19715 (ARM_TARGET2_DWARF_FORMAT): Define.
19716 * config/arm/t-vxworks: Adjust multilib control to removal of the
19717 Diab command line options.
19719 2017-08-01 Martin Liska <mliska@suse.cz>
19721 PR gcov-profile/81561
19722 * gcov.c (unblock): Make unblocking safe as we need to preserve
19723 index correspondence of blocks and block_lists.
19725 2017-08-01 Richard Biener <rguenther@suse.de>
19727 PR tree-optimization/81181
19728 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
19729 (compute_antic): ... end of iteration here.
19731 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
19733 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
19734 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
19735 (ftree-slp-vectorize): Likewise.
19736 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
19737 can no longer be set independent of flag_tree_loop_vectorize.
19738 * omp-general.c (emp_max_vf): Likewise.
19739 * opts.c (enable_fdo_optimizations): Remove references to
19740 flag_tree_vectorize, these are now implicit.
19741 (common_handle_option): Remove handling for OPT_ftree_vectorize,
19742 and leave it for the options machinery.
19744 2017-08-01 Martin Liska <mliska@suse.cz>
19746 PR middle-end/70140
19747 * builtins.c (expand_builtin_memcpy_args): Remove.
19748 (expand_builtin_memcpy): Call newly added function
19749 expand_builtin_memory_copy_args.
19750 (expand_builtin_memcpy_with_bounds): Likewise.
19751 (expand_builtin_mempcpy): Remove last argument.
19752 (expand_builtin_mempcpy_with_bounds): Likewise.
19753 (expand_builtin_memory_copy_args): New function created from
19754 expand_builtin_mempcpy_args with small modifications.
19755 (expand_builtin_mempcpy_args): Remove.
19756 (expand_builtin_stpcpy): Remove unused argument.
19757 (expand_builtin): Likewise.
19758 (expand_builtin_with_bounds): Likewise.
19760 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19763 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
19764 print "ds:" only for immediates in generic address space.
19766 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19769 * config/i386/i386.c (ix86_funciton_naked): New prototype.
19770 (ix86_function_ok_for_sibcall): Return false for naked functions.
19772 2017-08-01 Richard Biener <rguenther@suse.de>
19774 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
19775 (compute_antic): Seed worklist with exit block predecessors.
19776 * cfganal.c (dfs_find_deadend): For a cycle return the source
19777 of the edge closing it.
19779 2017-08-01 Tamar Christina <tamar.christina@arm.com>
19781 * config/aarch64/aarch64.c
19782 (aarch64_can_const_movi_rtx_p): Move 0 check.
19784 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19786 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
19787 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
19789 * match.pd: Ditto in address comparison pattern.
19791 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19793 PR tree-optimization/81627
19794 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
19795 closed ssa form for store-store chain.
19797 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19799 PR tree-optimization/81620
19800 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
19801 for store-store chain.
19803 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19805 PR tree-optimization/81588
19806 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
19807 ranges[i].in_p, invert comparison code ccode. For >/>=,
19808 swap rhs1 and rhs2 and comparison code unconditionally,
19809 for </<= don't do that. Don't swap rhs1/rhs2 again if
19810 ranges[i].in_p, instead invert comparison code ccode if
19811 opcode or oe->rank is BIT_IOR_EXPR.
19814 * optabs.def (vec_extract_optab, vec_init_optab): Change from
19815 a direct optab to conversion optab.
19816 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
19817 with GET_MODE_INNER as last argument instead of optab_handler.
19818 * expmed.c (extract_bit_field_1): Likewise. Use vector from
19819 vector extraction if possible and optab is available.
19820 * expr.c (store_constructor): Use convert_optab_handler instead
19821 of optab_handler. Use vector initialization from smaller
19822 vectors if possible and optab is available.
19823 * tree-vect-stmts.c (vectorizable_load): Likewise.
19824 * doc/md.texi (vec_extract, vec_init): Document that the optabs
19825 now have two modes.
19826 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
19827 of vec_init from half-sized vectors with the same element mode.
19828 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
19829 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
19830 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
19831 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
19832 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
19833 after mode in gen_vec_extract* calls.
19834 (vec_extract<mode>): Renamed to ...
19835 (vec_extract<mode><ssescalarmodelower>): ... this.
19836 (vec_extract<mode><ssehalfvecmodelower>): New expander.
19837 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
19838 element mode after mode in gen_vec_init* calls.
19839 (VEC_INIT_HALF_MODE): New mode iterator.
19840 (vec_init<mode>): Renamed to ...
19841 (vec_init<mode><ssescalarmodelower>): ... this.
19842 (vec_init<mode><ssehalfvecmodelower>): New expander.
19843 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
19844 (vec_extractv2sfsf): ... this.
19845 (vec_initv2sf): Renamed to ...
19846 (vec_initv2sfsf): ... this.
19847 (vec_extractv2si): Renamed to ...
19848 (vec_extractv2sisi): ... this.
19849 (vec_initv2si): Renamed to ...
19850 (vec_initv2sisi): ... this.
19851 (vec_extractv4hi): Renamed to ...
19852 (vec_extractv4hihi): ... this.
19853 (vec_initv4hi): Renamed to ...
19854 (vec_initv4hihi): ... this.
19855 (vec_extractv8qi): Renamed to ...
19856 (vec_extractv8qiqi): ... this.
19857 (vec_initv8qi): Renamed to ...
19858 (vec_initv8qiqi): ... this.
19859 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
19860 (vec_init<mode>): Renamed to ...
19861 (vec_init<mode><VEC_base_l>): ... this.
19862 (vec_extract<mode>): Renamed to ...
19863 (vec_extract<mode><VEC_base_l>): ... this.
19864 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
19865 (vec_initv2sfsf): ... this.
19866 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
19867 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19868 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
19869 element mode after mode in gen_vec_init* calls.
19870 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
19871 (vec_init<mode><Vel>): ... this.
19872 (vec_extract<mode>): Renamed to ...
19873 (vec_extract<mode><Vel>): ... this.
19874 * config/aarch64/iterators.md (Vel): New mode attribute.
19875 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
19876 Add element mode after mode in gen_vec_extract* calls.
19877 * config/s390/vector.md (non_vec_l): New mode attribute.
19878 (vec_extract<mode>): Renamed to ...
19879 (vec_extract<mode><non_vec_l>): ... this.
19880 (vec_init<mode>): Renamed to ...
19881 (vec_init<mode><non_vec_l>): ... this.
19882 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
19883 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
19885 * config/arm/iterators.md (V_elem_l): New mode attribute.
19886 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
19887 (vec_extract<mode><V_elem_l>): ... this.
19888 (vec_extractv2di): Renamed to ...
19889 (vec_extractv2didi): ... this.
19890 (vec_init<mode>): Renamed to ...
19891 (vec_init<mode><V_elem_l>): ... this.
19892 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
19893 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
19894 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
19895 Add element mode after gen_vec_extract* calls.
19896 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
19897 (vec_init<mode><unitmode>): ... this.
19898 (vec_extract<mode>): Renamed to ...
19899 (vec_extract<mode><unitmode>): ... this.
19900 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
19901 (vec_init<mode><unitmode>): ... this.
19902 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
19903 (vec_initv2sfsf): ... this.
19904 (vec_extractv2sf): Renamed to ...
19905 (vec_extractv2sfsf): ... this.
19906 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
19907 Add element mode after gen_vec_extract* calls.
19908 * config/mips/mips.md (unitmode): New mode iterator.
19909 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
19910 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
19911 * config/spu/spu.md (inner_l): New mode attribute.
19912 (vec_init<mode>): Renamed to ...
19913 (vec_init<mode><inner_l>): ... this.
19914 (vec_extract<mode>): Renamed to ...
19915 (vec_extract<mode><inner_l>): ... this.
19916 * config/sparc/sparc.md (veltmode): New mode iterator.
19917 (vec_init<VMALL:mode>): Renamed to ...
19918 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
19919 * config/ia64/vect.md (vec_initv2si): Renamed to ...
19920 (vec_initv2sisi): ... this.
19921 (vec_initv2sf): Renamed to ...
19922 (vec_initv2sfsf): ... this.
19923 (vec_extractv2sf): Renamed to ...
19924 (vec_extractv2sfsf): ... this.
19925 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
19926 (vec_init<mode>): Renamed to ...
19927 (vec_init<mode><VEC_base_l>): ... this.
19928 (vec_extract<mode>): Renamed to ...
19929 (vec_extract<mode><VEC_base_l>): ... this.
19930 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19931 (vec_initv2sfsf): ... this.
19932 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19933 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19934 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19935 gen_vec_init* calls.
19937 2017-08-01 Richard Biener <rguenther@suse.de>
19939 PR tree-optimization/81297
19940 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19941 TREE_OVERFLOW from INTEGER_CSTs.
19943 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19945 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19947 2017-07-31 Carl Love <cel@us.ibm.com>
19949 * config/rs6000/rs6000-c: Add support for built-in functions
19950 vector signed char vec_xl_be (signed long long, signed char *);
19951 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19952 vector signed int vec_xl_be (signed long long, signed int *);
19953 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19954 vector signed long long vec_xl_be (signed long long, signed long long *);
19955 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19956 vector signed short vec_xl_be (signed long long, signed short *);
19957 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19958 vector double vec_xl_be (signed long long, double *);
19959 vector float vec_xl_be (signed long long, float *);
19960 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19961 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19962 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19964 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19965 (altivec_expand_builtin): Add switch statement to call
19966 altivec_expand_xl_be for each builtin.
19967 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19968 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19969 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19970 __builtin_vsx_le_be_v16qi.
19971 * doc/extend.texi: Update the built-in documentation file for the
19972 new built-in functions.
19974 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19977 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19979 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19981 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19983 * config.gcc: Add z14.
19984 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19985 CPU model numbers for z13s and z14.
19986 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19988 * config/s390/s390-opts.h (enum processor_type): Rename
19989 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19990 * config/s390/s390.c (processor_table): Add field for CPU name to
19991 be passed to Binutils.
19992 (s390_asm_output_machine_for_arch): Use the new field in
19993 processor_table for Binutils.
19994 (s390_expand_builtin): Replace arch12 with z14.
19995 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19996 (s390_get_sched_attrmask): Likewise.
19997 (s390_get_unit_mask): Likewise.
19998 * config/s390/s390.opt: Add z14 to processor_type enum.
20000 2017-07-31 Martin Jambor <mjambor@suse.cz>
20003 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
20004 regardless of optimization level.
20006 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
20007 Martin Liska <mliska@suse.cz>
20009 * predict.def: Remove old comment and adjust probability.
20010 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
20011 PREDICT statements.
20013 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
20016 * config/i386/i386.c (ix86_function_naked): New function.
20017 (ix86_can_use_return_insn_p): Return false for naked functions.
20018 (ix86_expand_prologue): Skip prologue for naked functions.
20019 (ix86_expand_epilogue): Skip epilogue for naked functions
20020 and emit trap instruction.
20021 (ix86_warn_func_return): New function.
20022 (ix86_attribute_table): Add "naked" attribute specification.
20023 (TARGET_WARN_FUNC_RETURN): Define.
20024 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
20026 2017-07-31 Martin Liska <mliska@suse.cz>
20028 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
20029 (dump_gimple_bb_header): Always dump BB info.
20030 (pp_cfg_jump): Do not append info about BB when dumping a jump.
20032 2017-07-31 Martin Liska <mliska@suse.cz>
20035 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
20036 also with current_function_decl non-null equality.
20038 2017-07-31 Jakub Jelinek <jakub@redhat.com>
20041 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
20042 change type to the element type, instead add eltype variable and
20043 use it where we are interested in the element type.
20045 PR tree-optimization/81603
20046 * ipa-polymorphic-call.c
20047 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
20048 offset arithmetic in offset_int, bail out if the resulting bit offset
20049 doesn't fit into shwi.
20051 2017-07-31 Martin Liska <mliska@suse.cz>
20053 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
20054 (gimplify_save_expr): Fix comment.
20056 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
20059 * config/i386/i386.c (ix86_function_arg): Update arguments for
20061 (ix86_compute_frame_layout): Set the initial stack offset to
20062 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
20063 INCOMING_FRAME_SP_OFFSET.
20064 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
20065 stack before exception handler returns.
20066 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
20067 the 'ERROR_CODE' for exception handler.
20069 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
20071 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
20072 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
20073 (ASM_OUTPUT_REG_POP): Ditto.
20074 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
20075 instead of asm_fprintf to output pure string.
20077 2017-07-29 Jakub Jelinek <jakub@redhat.com>
20079 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
20080 to imported_module_or_decl hook.
20081 (debug_nothing_tree_tree_tree_bool): Remove.
20082 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
20083 * debug.c (do_nothing_debug_hooks): Use
20084 debug_nothing_tree_tree_tree_bool_bool instead of
20085 debug_nothing_tree_tree_tree_bool.
20086 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20087 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
20088 * sdbout.c (sdb_debug_hooks): Likewise.
20089 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
20090 (gen_namespace_die): Add DW_AT_export_symbols attribute if
20092 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
20093 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
20094 attribute, don't add anything.
20096 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20098 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
20099 (fold_build2_stat_loc): Likewise.
20100 (fold_build3_stat_loc): Likewise.
20101 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
20102 (fold_build1_loc): Remove macro.
20103 (fold_build2_loc): Likewise.
20104 (fold_build3_loc): Likewise.
20106 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20108 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
20109 (gimple_build_debug_bind_source_stat): Likewise.
20110 * gimple.h (gimple_build_debug_bind): Remove macro.
20111 (gimple_build_debug_bind_source): Likewise.
20113 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20115 * bitmap.c (bitmap_alloc): Adjust.
20116 (bitmap_gc_alloc): Likewise.
20117 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
20119 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20121 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
20122 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
20123 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
20124 (bitmap_gc_alloc_stat): Likewise.
20125 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
20127 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20129 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
20130 * rtl.h (shallow_copy_rtx): Remove macro.
20132 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20134 * emit-rtl.c (gen_raw_REG): Adjust.
20135 * gengenrtl.c (gendef): Likewise.
20136 * rtl.c (rtx_alloc_stat): Remove _stat from name.
20137 * rtl.h (rtx_alloc): Remove macro.
20139 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20141 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
20142 (build_tree_list_stat): Likewise.
20143 * tree.h (build_tree_list): Remove macro.
20144 (build_tree_list_vec): Likewise.
20146 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20148 * tree.c (make_vector_stat): Remove _stat from name.
20149 (build_vector_stat): Likewise.
20150 * tree.h (make_vector_stat): Remove macro.
20151 (build_vector_stat): Likewise.
20153 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20155 * tree.h (build_var_debug_value): Remove prototype.
20157 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20159 * tree.c (tree_cons_stat): Remove _stat from name.
20160 * tree.h (tree_cons): Remove macro.
20162 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20164 * tree.c (build_vl_exp_stat): Remove _stat from name.
20165 * tree.h (build_vl_exp): Remove macro.
20167 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20169 * tree.c (build_decl_stat): Remove _stat from name.
20170 * tree.h (build_decl): Remove macro.
20172 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20174 * gimple.c (gimple_build_with_ops_stat): Adjust.
20175 (gimple_alloc_stat): Remove _stat from name.
20176 * gimple.h (gimple_alloc): Remove macro.
20178 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20180 * tree.c (make_tree_vec_stat): Remove _stat from name.
20181 (grow_tree_vec_stat): Likewise.
20182 * tree.h (make_tree_vec_stat): Adjust prototype.
20183 (grow_tree_vec_stat): Likewise.
20184 (make_tree_vec): Remove macro.
20185 (grow_tree_vec): Likewise.
20187 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20189 * fold-const.c (fold_build1_stat_loc): Adjust.
20190 (fold_build2_stat_loc): Likewise.
20191 (fold_build3_stat_loc): Likewise.
20192 * tree.c (build0_stat): Remove _stat from name.
20193 (build1_stat): Likewise.
20194 (build2_stat): Likewise.
20195 (build3_stat): Likewise.
20196 (build4_stat): Likewise.
20197 (build5_stat): Likewise.
20198 * tree.h (build1_loc): Remove macro, and rename _stat function
20200 (build2_loc): Likewise.
20201 (build3_loc): Likewise.
20202 (build4_loc): Likewise.
20203 (build5_loc): Likewise.
20205 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20207 * tree.c (make_int_cst_stat): Remove _stat from name.
20208 * tree.h (make_int_cst_stat): Adjust prototype.
20209 (make_int_cst): Remove macro.
20211 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20213 * tree.c (make_tre_binfo_stat): Remove _stat from name.
20214 * tree.h (make_tree_binfo_stat): Adjust prototype.
20215 (make_tree_binfo): Remove.
20217 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20219 * tree.c (copy_node_stat): Rename to copy_node.
20220 (build_distinct_type_copy): Adjust.
20221 * tree.h (copy_node_stat): Adjust prototype.
20222 (copy_node): Remove macro.
20224 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20226 * tree.c (make_node_stat): rename to make_node.
20227 (build_tree_list_stat): Adjust.
20228 (build0_stat): Likewise.
20229 (build2_stat): Likewise.
20230 (build3_stat): Likewise.
20231 (build4_stat): Likewise.
20232 (build5_stat): Likewise.
20233 (build_decl_stat): Likewise.
20234 * tree.h (make_node_stat): Adjust prototype.
20235 (make_node): remove macro.
20237 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
20239 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
20240 (PPC_FEATURE2_SCV): Likewise.
20241 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
20243 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20245 * config/aarch64/aarch64.c
20246 (aarch64_internal_mov_immediate): Add new special pattern.
20247 * config/aarch64/aarch64.md (*movdi_aarch64):
20248 Add reg/32bit const mov case.
20250 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20251 Richard Sandiford <richard.sandiford@linaro.org>
20253 * config/aarch64/aarch64.md (mov<mode>): Generalize.
20254 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20255 Add integer and movi cases.
20256 (movi-split-hf-df-sf split, fp16): New.
20257 (enabled): Added TARGET_FP_F16INST.
20258 * config/aarch64/iterators.md (GPF_HF): New.
20259 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
20261 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20263 * config/aarch64/aarch64.c
20264 (aarch64_simd_container_mode): Add prototype.
20265 (aarch64_expand_mov_immediate): Add HI support.
20266 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
20267 (aarch64_can_const_movi_rtx_p): New.
20268 (aarch64_preferred_reload_class):
20269 Remove restrictions of using FP registers for certain SIMD operations.
20270 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
20271 (aarch64_valid_floating_const): Add integer move validation.
20272 (aarch64_simd_imm_scalar_p): Remove.
20273 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
20274 (aarch64_legitimate_constant_p): Expand list of supported cases.
20275 * config/aarch64/aarch64-protos.h
20276 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
20277 (aarch64_reinterpret_float_as_int): New.
20278 (aarch64_simd_imm_scalar_p): Remove.
20279 * config/aarch64/constraints.md (Uvi): New.
20280 (Dd): Split into Ds and new Dd.
20281 * config/aarch64/aarch64.md (*movsi_aarch64):
20283 (*movdi_aarch64): Add SIMD mov case.
20285 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20287 * tree-predcom.c: (struct chain): Handle store-store chain in which
20288 stores for elimination only store loop invariant values.
20289 (execute_pred_commoning_chain): Ditto.
20290 (prepare_initializers_chain_store_elim): Ditto.
20291 (prepare_finalizers): Ditto.
20292 (is_inv_store_elimination_chain): New function.
20293 (initialize_root_vars_store_elim_1): New function.
20295 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20297 * tree-predcom.c: Revise general description of the pass.
20298 (enum chain_type): New enum type for store elimination.
20299 (struct chain): New field supporting store elimination.
20300 (struct component): Ditto.
20301 (dump_chain): Dump store-stores chain.
20302 (release_chain): Release resources.
20303 (split_data_refs_to_components): Compute and create component
20304 contains only stores for elimination.
20305 (get_chain_last_ref_at): New function.
20306 (make_invariant_chain): Initialization.
20307 (make_rooted_chain): Specify chain type in parameter and record it.
20308 (add_looparound_copies): Skip for store-stores chain.
20309 (determine_roots_comp): Compute type of chain and pass it to
20311 (initialize_root_vars_store_elim_2): New function.
20312 (finalize_eliminated_stores): New function.
20313 (remove_stmt): Handle store for elimination.
20314 (execute_pred_commoning_chain): Execute predictive commoning on
20315 store-store chains.
20316 (determine_unroll_factor): Skip unroll for store-stores chain.
20317 (prepare_initializers_chain_store_elim): New function.
20318 (prepare_initializers_chain): Hanlde store-store chain.
20319 (prepare_finalizers_chain, prepare_finalizers): New function.
20320 (tree_predictive_commoning_loop): Return integer value indicating
20321 if loop is unrolled or lcssa form is corrupted.
20322 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
20324 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20326 * tree-predcom.c (initialize_root): Delete.
20327 (execute_pred_commoning_chain): Initialize root vars and replace
20328 reference of non-combined chain directly, rather than call above
20331 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20333 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
20334 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
20336 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20338 * tree-predcom.c (struct chain): New field init_seq.
20339 (release_chain): Release init_seq.
20340 (prepare_initializers_chain): Record intialization stmts in above
20342 (insert_init_seqs): New function.
20343 (tree_predictive_commoning_loop): Call insert_init_seqs.
20345 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20347 * tree-predcom.c (determine_roots_comp): Skip trivial components.
20349 2017-07-28 Richard Biener <rguenther@suse.de>
20351 * match.pd: Remove superfluous :c.
20352 * genmatch.c (simplify::id): Add member.
20353 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
20355 (current_id): New global.
20356 (dt_node::parent): Move from ...
20357 (dt_operand::parent): ... here. Add for_id member.
20358 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
20359 (decision_tree::find_node): Relax order requirement when
20360 merging DT_TRUE nodes to ones inbetween the current simplify
20361 and the one we try to merge with. Add diagnostic whenever
20362 we need to enforce pattern order by not merging.
20363 (decision_tree::insert): Set current_id.
20364 (decision_tree::print_node): Dump parent node and for_id.
20365 (parser::last_id): Add member.
20366 (parser::push_simplify): Assign unique id.
20367 (parser::parser): Initialize last_id.
20369 2017-07-28 Martin Liska <mliska@suse.cz>
20372 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
20373 gimple_build_debug_bind.
20375 2017-07-28 Richard Biener <rguenther@suse.de>
20377 PR tree-optimization/81502
20378 * match.pd: Add pattern combining BIT_INSERT_EXPR with
20380 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
20382 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
20383 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
20384 for BIT_FIELD_REF args.
20385 * fold-const.c (make_bit_field_ref): Likewise.
20386 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
20388 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20391 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
20392 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
20393 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
20394 Or it into SANITIZER_UNDEFINED.
20395 * ubsan.c: Include gimple-fold.h and varasm.h.
20396 (ubsan_expand_ptr_ifn): New function.
20397 (instrument_pointer_overflow): New function.
20398 (maybe_instrument_pointer_overflow): New function.
20399 (instrument_object_size): Formatting fix.
20400 (pass_ubsan::execute): Call instrument_pointer_overflow
20401 and maybe_instrument_pointer_overflow.
20402 * internal-fn.c (expand_UBSAN_PTR): New function.
20403 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
20404 * sanitizer.def (__ubsan_handle_pointer_overflow,
20405 __ubsan_handle_pointer_overflow_abort): New builtins.
20406 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
20407 * internal-fn.def (UBSAN_PTR): New internal function.
20408 * opts.c (sanitizer_opts): Add pointer-overflow.
20409 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
20410 * fold-const.c (build_range_check): Compute pointer range check in
20411 integral type if pointer arithmetics would be needed. Formatting
20414 2017-07-28 Martin Liska <mliska@suse.cz>
20417 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
20418 parameters that are of a variable-length.
20420 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
20422 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
20424 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
20425 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
20426 (CRT_CALL_STATIC_FUNCTION): Likewise.
20427 (ASM_DEFAULT_SPEC): New define.
20428 (ASM_SPEC32): Likewise.
20429 (ASM_SPEC64): Likewise.
20430 (ASM_SPEC_COMMON): Likewise.
20431 (ASM_SPEC): Likewise.
20432 (INVALID_64BIT): Likewise.
20433 (LINK_OS_DEFAULT_SPEC): Likewise.
20434 (LINK_OS_SPEC32): Likewise.
20435 (LINK_OS_SPEC64): Likewise.
20436 (POWERPC_LINUX): Likewise.
20437 (PTRDIFF_TYPE): Likewise.
20438 (RESTORE_FP_PREFIX): Likewise.
20439 (RESTORE_FP_SUFFIX): Likewise.
20440 (SAVE_FP_PREFIX): Likewise.
20441 (SAVE_FP_SUFFIX): Likewise.
20442 (SIZE_TYPE): Likewise.
20443 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
20444 (TARGET_64BIT): Likewise.
20445 (TARGET_64BIT): Likewise.
20446 (TARGET_AIX): Likewise.
20447 (WCHAR_TYPE_SIZE): Likewise.
20448 (WCHAR_TYPE): Undefine.
20449 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
20450 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
20451 (CPP_OS_RTEMS_SPEC): Delete.
20452 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
20453 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
20455 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
20457 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20459 PR tree-optimization/81578
20460 * tree-parloops.c (build_new_reduction): Bail out if
20461 reduction_code isn't one of the standard OpenMP reductions.
20462 Move the details printing after that decision.
20464 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20466 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
20467 related to reload_in_progress.
20468 (splat_input_operand): Likewise.
20469 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
20471 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
20473 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
20474 (TARGET_INSTANTIATE_DECLS): Likewise.
20475 (legitimate_indexed_address_p): Delete reload_in_progress code.
20476 (rs6000_debug_legitimate_address_p): Likewise.
20477 (rs6000_eliminate_indexed_memrefs): Likewise.
20478 (rs6000_emit_le_vsx_store): Likewise.
20479 (rs6000_emit_move_si_sf_subreg): Likewise.
20480 (rs6000_emit_move): Likewise.
20481 (register_to_reg_type): Likewise.
20482 (rs6000_pre_atomic_barrier): Likewise.
20483 (rs6000_machopic_legitimize_pic_address): Likewise.
20484 (rs6000_allocate_stack_temp): Likewise.
20485 (rs6000_address_for_fpconvert): Likewise.
20486 (rs6000_address_for_altivec): Likewise.
20487 (rs6000_secondary_memory_needed_rtx): Delete function.
20488 (rs6000_check_sdmode): Likewise.
20489 (rs6000_alloc_sdmode_stack_slot): Likewise.
20490 (rs6000_instantiate_decls): Likewise.
20491 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
20492 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
20493 Delete reload_in_progress.
20494 (*vec_reload_and_plus_<mptrsize>): Likewise.
20495 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
20496 (vsx_div_v2di): Likewise.
20497 (vsx_udiv_v2di): Likewise.
20499 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20501 * config/rs6000/rs6000.opt (mlra): Replace with stub.
20502 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
20503 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
20504 (rs6000_debug_reg_global): Delete print of LRA status.
20505 (rs6000_option_override_internal): Delete dead LRA related code.
20506 (rs6000_lra_p): Delete function.
20507 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
20509 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
20511 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
20512 * config/riscv/rtems.h: New file.
20514 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20515 Sudakshina Das <sudi.das@arm.com>
20517 * config/aarch64/aarch64.md
20518 (define_split for and<mode>3nr_compare): Move
20519 non aarch64_logical_operand to a register.
20520 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
20521 register immediate operand to a register.
20522 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
20524 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20526 PR middle-end/81564
20527 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
20529 2017-07-27 Richard Biener <rguenther@suse.de>
20531 PR tree-optimization/81573
20532 PR tree-optimization/81494
20533 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
20534 multi defuse cycle case.
20536 2017-07-27 Richard Biener <rguenther@suse.de>
20538 PR tree-optimization/81571
20539 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
20542 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
20544 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
20545 earlier and only if MASK_FPU is set. Adjust formatting.
20547 2017-07-27 Martin Liska <mliska@suse.cz>
20549 * opt-functions.awk: Add validation of value of Init.
20550 * optc-gen.awk: Pass new argument.
20552 2017-07-27 Martin Liska <mliska@suse.cz>
20554 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
20555 Fix wrong condition.
20557 2017-07-27 Martin Liska <mliska@suse.cz>
20559 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
20560 BBs and edges seen by autoFDO.
20562 2017-07-27 Richard Biener <rguenther@suse.de>
20564 PR tree-optimization/81502
20565 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
20566 with incompatible but same sized type.
20567 (execute_update_addresses_taken): Likewise.
20569 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
20571 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
20572 flag_tree_loop_vectorize rather than flag_tree_vectorize.
20574 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20577 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
20578 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
20579 Change s_operand to memory_operand.
20581 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
20583 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
20584 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
20585 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
20586 Emit instructions rather than returning an expression. Handle TFmode
20587 and KFmode by casting to TImode.
20588 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
20589 (rs6000_emit_le_vsx_store): Likewise.
20590 * config/rs6000/vsx.md (VSX_TI): New iterator.
20591 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
20592 (*vsx_le_undo_permute_<mode>): Likewise.
20593 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
20594 emit the split sequence.
20595 (*vsx_le_perm_store_<mode>): Likewise.
20597 2017-07-27 Jakub Jelinek <jakub@redhat.com>
20599 PR tree-optimization/81555
20600 PR tree-optimization/81556
20601 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
20602 if true, force CHANGED for the recursive invocation.
20603 (reassociate_bb): Remember original length of ops array, pass
20604 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
20606 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
20607 attributes for noipa attribute. For naked attribute use
20608 lookup_attribute first before lookup_attribute_spec.
20609 * final.c (rest_of_handle_final): Disable IPA RA for functions with
20611 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
20612 for functions with noipa attribute.
20613 (cgraph_externally_visible_p): Return true for functions with noipa
20615 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
20616 for functions with noipa attribute.
20617 * doc/extend.texi: Document noipa function attribute.
20618 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
20619 also for functions with noipa attribute.
20620 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
20622 2017-07-26 Andrew Pinski <apinski@cavium.com>
20624 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
20625 vec_unalign_load_cost and vec_unalign_store_cost.
20627 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
20629 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
20630 -mvsx-small-integer option.
20631 (ISA_3_0_MASKS_IEEE): Likewise.
20632 (OTHER_VSX_VECTOR_MASKS): Likewise.
20633 (POWERPC_MASKS): Likewise.
20634 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
20635 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
20636 code, only testing for DImode being allowed in non-VSX floating
20638 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
20639 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
20641 (rs6000_option_override_internal): Delete -mvsx-small-integer.
20642 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
20643 TARGET_P8_VECTOR test.
20644 (rs6000_secondary_reload_simple_move): Likewise.
20645 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
20646 since TARGET_P9_VECTOR was already tested.
20647 (rs6000_opt_masks): Remove -mvsx-small-integer.
20648 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
20649 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20651 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
20652 test for TARGET_VEXTRACTUB was used, and that uses
20654 (p9 extract splitter): Likewise.
20655 (vsx_extract_<mode>_di_p9): Likewise.
20656 (vsx_extract_<mode>_store_p9): Likewise.
20657 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
20658 for TARGET_P9_VECTOR was used. Delete code that is now dead with
20659 the elimination of TARGET_VSX_SMALL_INTEGER.
20660 (vsx_extract_<mode>_p8): Likewise.
20661 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
20662 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
20663 (vsx_set_<mode>_p9): Likewise.
20664 (vsx_set_v4sf_p9): Likewise.
20665 (vsx_set_v4sf_p9_zero): Likewise.
20666 (vsx_insert_extract_v4sf_p9): Likewise.
20667 (vsx_insert_extract_v4sf_p9_2): Likewise.
20668 * config/rs6000/rs6000.md (sign extend splitter): Change
20669 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
20670 (floatsi<mode>2_lfiwax_mem): Likewise.
20671 (floatunssi<mode>2_lfiwzx_mem): Likewise.
20672 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
20673 since a test for TARGET_P9_VECTOR was used.
20674 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20675 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
20676 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20677 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20678 TARGET_P8_VECTOR test.
20679 (fix_trunc<mode>si2_stfiwx): Likewise.
20680 (fix_trunc<mode>si2_internal): Likewise.
20681 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
20682 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20684 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20685 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20686 TARGET_P8_VECTOR test.
20687 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20688 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
20689 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20691 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20692 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
20693 since a test for TARGET_P9_VECTOR was used.
20694 (splitter for loading small constants): Likewise.
20696 2017-07-26 Andrew Pinski <apinski@cavium.com>
20698 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
20701 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
20704 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
20705 (fp_valid_at): Likewise.
20707 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20709 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
20710 (qdf24xx_addrcost_table): Likewise.
20711 (cortexa57_tunings): Update to use generic_branch_cost.
20712 (cortexa72_tunings): Likewise.
20713 (cortexa73_tunings): Likewise.
20714 (qdf24xx_tunings): Likewise.
20716 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20718 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
20719 (thunderx2t99_branch_cost): Likewise.
20720 (cortexa35_tunings): Update to use generic_branch_cost.
20721 (cortexa53_tunings): Likewise.
20722 (cortexa57_tunings): Likewise.
20723 (cortexa72_tunings): Likewise.
20724 (cortexa73_tunings): Likewise.
20725 (thunderx2t99_tunings): Likewise.
20727 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20729 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
20730 (sparc_option_override): Honour MASK_FSMULD.
20731 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
20732 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
20733 * config/sparc/sparc.opt (mfsmuld): New option.
20734 * doc/invoke.texi (mfsmuld): Document option.
20736 2017-07-26 Marek Polacek <polacek@redhat.com>
20738 PR middle-end/70992
20739 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
20741 2017-07-26 Richard Biener <rguenther@suse.de>
20743 * gimple-match-head.c (do_valueize): Return OP if valueize
20745 (get_def): New helper to get at the def stmt of a SSA name
20746 if valueize allows.
20747 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
20748 do_valueize to get at the def stmt.
20749 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
20751 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
20753 PR middle-end/46932
20754 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
20756 2017-07-26 Martin Liska <mliska@suse.cz>
20759 * function.c (expand_function_start): Make expansion of
20760 nonlocal_goto_save_area after parm_birth_insn.
20762 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20764 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
20765 from all CPU target flags enable members.
20767 2017-07-26 Richard Biener <rguenther@suse.de>
20769 * genmatch.c (dt_simplify::gen): Make iterator vars const.
20770 (decision_tree::gen): Make 'type' const.
20771 (write_predicate): Likewise.
20773 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20775 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
20776 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
20777 (rs6000_option_override_internal): Likewise.
20778 (rs6000_expand_vector_set): Likewise.
20779 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
20780 (TARGET_UPPER_REGS_SF): Likewise.
20781 (TARGET_UPPER_REGS_DI): Likewise.
20782 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
20783 (TARGET_DIRECT_MOVE_64BIT): Likewise.
20784 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
20785 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20786 (Splitters for DI constants in Altivec registers): Likewise.
20787 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
20788 (vsx_set_v4sf_p9): Likewise.
20789 (vsx_set_v4sf_p9_zero): Likewise.
20790 (vsx_insert_extract_v4sf_p9): Likewise.
20791 (vsx_insert_extract_v4sf_p9_2): Likewise.
20793 2017-07-25 Carl Love <cel@us.ibm.com>
20795 * doc/extend.texi: Update the built-in documentation file for the
20796 existing built-in functions
20797 vector signed char vec_cnttz (vector signed char);
20798 vector unsigned char vec_cnttz (vector unsigned char);
20799 vector signed short vec_cnttz (vector signed short);
20800 vector unsigned short vec_cnttz (vector unsigned short);
20801 vector signed int vec_cnttz (vector signed int);
20802 vector unsigned int vec_cnttz (vector unsigned int);
20803 vector signed long long vec_cnttz (vector signed long long);
20804 vector unsigned long long vec_cnttz (vector unsigned long long);
20806 2017-07-25 Andrew Pinski <apinski@cavium.com>
20808 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
20809 accesses where the use is for the first operand of a BIT_INSERT.
20811 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20814 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20815 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20817 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20819 * config/i386/gstabs.h: Delete.
20820 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
20822 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
20824 * config/i386/i386.c (ix86_decompose_address): Do not check for
20825 register RTX when looking at index_reg or base_reg.
20826 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
20828 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
20830 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
20831 to update EH info here.
20833 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20835 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
20837 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20839 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
20841 2017-07-25 Torsten Duwe <duwe@suse.de>
20843 * common.opt: Introduce -fpatchable-function-entry
20844 command line option, and its variables function_entry_patch_area_size
20845 and function_entry_patch_area_start.
20846 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
20847 including a two-value parser.
20848 * target.def (print_patchable_function_entry): New target hook.
20849 * targhooks.h (default_print_patchable_function_entry): New function.
20850 * targhooks.c (default_print_patchable_function_entry): Likewise.
20851 * toplev.c (process_options): Switch off IPA-RA if
20852 patchable function entries are being generated.
20853 * varasm.c (assemble_start_function): Look at the
20854 patchable-function-entry command line switch and current
20855 function attributes and maybe generate NOP instructions by
20856 calling the print_patchable_function_entry hook.
20857 * doc/extend.texi: Document patchable_function_entry attribute.
20858 * doc/invoke.texi: Document -fpatchable_function_entry
20859 command line option.
20860 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
20862 * doc/tm.texi: Re-generate.
20864 2017-07-25 Jakub Jelinek <jakub@redhat.com>
20867 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
20868 TARGET_AVX512DQ rather than TARGET_AVX512BW.
20870 2017-07-25 Tamar Christina <tamar.christina@arm.com>
20872 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
20874 2017-07-25 Richard Biener <rguenther@suse.de>
20876 PR tree-optimization/81455
20877 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
20878 not walk in cycles when looking for guards.
20880 2017-07-25 Richard Biener <rguenther@suse.de>
20882 PR tree-optimization/81529
20883 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
20884 when optimizing backedge uses.
20886 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
20888 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
20890 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
20891 to dl_section_ref. On AIX, append an expression to subtract
20892 the size of the section length to dl_section_ref.
20894 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
20896 * configure.ac: If any of the config.* scripts fail, exit 1.
20897 * configure: Regenerate.
20899 2017-07-25 Richard Biener <rguenther@suse.de>
20901 PR middle-end/81546
20902 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
20903 of immediate uses, be more verbose on errors.
20905 2017-07-25 Richard Biener <rguenther@suse.de>
20907 PR tree-optimization/81510
20908 * tree-vect-loop.c (vect_is_simple_reduction): When the
20909 reduction stmt is not inside the loop bail out.
20911 2017-07-25 Richard Biener <rguenther@suse.de>
20913 PR tree-optimization/81303
20914 * tree-vect-loop-manip.c (vect_loop_versioning): Build
20915 profitability check against LOOP_VINFO_NITERSM1.
20917 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20919 * domwalk.c (cmp_bb_postorder): Simplify.
20920 (sort_bbs_postorder): New function. Use it...
20921 (dom_walker::walk): ...here to optimize common cases.
20923 2017-07-25 Martin Liska <mliska@suse.cz>
20926 * ipa-visibility.c (function_and_variable_visibility): Make the
20927 redirection just on target that supports aliasing.
20928 Fix GNU coding style.
20930 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20933 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20935 * config/arm/rtems.h: Likewise.
20936 * config/bfin/rtems.h: Likewise.
20937 * config/i386/rtemself.h: Likewise.
20938 * config/lm32/rtems.h: Likewise.
20939 * config/m32c/rtems.h: Likewise.
20940 * config/m68k/rtemself.h: Likewise.
20941 * config/microblaze/rtems.h: Likewise.
20942 * config/mips/rtems.h: Likewise.
20943 * config/moxie/rtems.h: Likewise.
20944 * config/nios2/rtems.h: Likewise.
20945 * config/powerpcspe/rtems.h: Likewise.
20946 * config/rs6000/rtems.h: Likewise.
20947 * config/rtems.h: Likewise.
20948 * config/sh/rtems.h: Likewise.
20949 * config/sh/rtemself.h: Likewise.
20950 * config/sparc/rtemself.h: Likewise.
20952 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20955 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20956 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20957 * tree-ssa-structalias.c (alias_get_name): Same.
20959 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20962 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20963 instructions if no du chain is found.
20965 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20967 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20969 2017-07-25 Richard Biener <rguenther@suse.de>
20971 PR middle-end/81505
20972 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20975 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20977 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20978 upper-regs options.
20979 (ISA_2_7_MASKS_SERVER): Likewise.
20980 (ISA_3_0_MASKS_IEEE): Likewise.
20981 (OTHER_P8_VECTOR_MASKS): Likewise.
20982 (OTHER_VSX_VECTOR_MASKS): Likewise.
20983 (POWERPC_MASKS): Likewise.
20984 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20985 duplicate list of options.
20986 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20987 explicit -mupper-regs options.
20988 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20989 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20990 alias for -mupper-regs-df.
20991 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20992 (rs6000_init_hard_regno_mode_ok): Likewise.
20993 (rs6000_option_override_internal): Likewise.
20994 (rs6000_opt_masks): Likewise.
20995 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20996 options in terms of whether -mvsx or -mpower8-vector was used.
20997 (TARGET_UPPER_REGS_DI): Likewise.
20998 (TARGET_UPPER_REGS_SF): Likewise.
20999 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
21000 -mupper-regs-* options.
21002 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21004 * passes.c (emergency_dump_function): Print some empty lines and a
21005 header before the RTL dump.
21007 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
21009 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
21011 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
21014 * config/aarch64/aarch64.c (aarch64_classify_symbol):
21015 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
21017 2017-07-24 Carl Love <cel@us.ibm.com>
21019 * config/rs6000/rs6000-c.c: Add support for built-in functions
21020 vector float vec_extract_fp32_from_shorth (vector unsigned short);
21021 vector float vec_extract_fp32_from_shortl (vector unsigned short);
21022 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
21023 vec_extract_fp_from_shortl): Add defines for the two builtins.
21024 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
21025 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
21027 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
21028 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
21029 * doc/extend.texi: Update the built-in documentation file for the
21030 new built-in function.
21032 2017-07-24 Jakub Jelinek <jakub@redhat.com>
21035 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
21037 * doc/generic.texi: Likewise.
21038 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
21039 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
21041 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
21043 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
21044 (aarch64_mls_elt_merge<mode>): Likewise.
21046 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
21048 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
21049 having __cxa_atexit.
21051 2017-07-23 Michael Collison <michael.collison@arm.com>
21053 * config/arm/arm.c (arm_option_override): Deprecate
21054 use of -mstructure-size-boundary.
21055 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
21056 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
21058 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21061 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
21062 Reduce cost estimate for direct moves.
21064 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
21067 * config/i386/i386.c (ix86_option_override_internal): Disable
21068 BMI, BMI2 and TBM instructions for -m16.
21070 2017-07-21 Carl Love <cel@us.ibm.com>
21072 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21073 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21074 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21075 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21076 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21077 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21078 VMULOSW): New enum "unspec" values.
21079 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21080 altivec_vmulosw): New patterns.
21081 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21082 VMULOSW): Add definitions.
21084 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
21086 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
21087 (qdf24xx): Likewise.
21088 * config/aarch64/aarch64-options-extensions.def (rdma); New.
21089 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
21090 (AARCH64_FL_V8_1): Renumber.
21091 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
21092 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
21093 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
21094 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
21095 rdma to feature modifiers list.
21097 2017-07-21 Yury Gribov <tetra2005@gmail.com>
21099 PR middle-end/56727
21100 * ipa-visibility (function_and_variable_visibility): Convert
21101 recursive PLT call to direct call if appropriate.
21103 2017-07-21 Andrew Pinski <apinski@cavium.com>
21105 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
21106 operand 1 to see if the types precision matches.
21107 * fold-const.c (operand_equal_p): Likewise.
21109 2017-07-21 Richard Biener <rguenther@suse.de>
21111 PR tree-optimization/81303
21112 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
21113 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
21114 (vect_peeling_hash_get_lowest_cost): Adjust.
21115 (vect_enhance_data_refs_alignment): Likewise. Use
21116 vect_get_peeling_costs_all_drs to compute the penalty for no
21117 peeling to match up costs.
21119 2017-07-21 Richard Biener <rguenther@suse.de>
21121 PR tree-optimization/81500
21122 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
21123 we didn't identify a reduction path.
21125 2017-07-21 Tom de Vries <tom@codesourcery.com>
21126 Cesar Philippidis <cesar@codesourcery.com>
21128 PR gcov-profile/81442
21129 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
21132 2017-07-21 Tom de Vries <tom@codesourcery.com>
21135 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
21137 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
21138 nvptx_override_options_after_change.
21140 2017-07-21 Ulrich Drepper <drepper@redhat.com>
21142 * dwarf2out.c (output_file_names): Avoid double testing for
21143 dwarf_version >= 5.
21145 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
21147 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
21149 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21151 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
21153 (try_crossjump_to_edge): Do not punt on partitioned functions.
21155 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21157 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
21158 Put all BBs reachable only via paths crossing cold region to cold
21160 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
21162 2016-07-21 Richard Biener <rguenther@suse.de>
21164 PR tree-optimization/81303
21165 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
21166 into account prologue and epilogue iterations when raising
21167 min_profitable_iters to sth at least covering one vector iteration.
21169 2017-07-21 Tamar Christina <tamar.christina@arm.com>
21171 * config/arm/arm.c (arm_test_cpu_arch_dat):
21174 2017-07-20 Nathan Sidwell <nathan@acm.org>
21176 Remove TYPE_METHODS.
21177 * tree.h (TYPE_METHODS): Delete.
21178 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
21179 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
21180 (dbxout_type_methods): Scan TYPE_FIELDS.
21181 (dbxout_type): Don't check TYPE_METHODS here.
21182 * function.c (use_register_for_decl): Always ignore register for
21183 class types when not optimizing.
21184 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
21185 * tree.c (free_lang_data_in_type): Stitch out member functions and
21186 templates from TYPE_FIELDS.
21187 (build_distinct_type_copy, verify_type_variant,
21188 verify_type): Member fns are on TYPE_FIELDS.
21189 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
21190 * tree-pretty-print.c (dump_generic_node): Likewise.
21192 2017-07-20 Jakub Jelinek <jakub@redhat.com>
21195 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
21196 V2TImode and V4TImode.
21197 (ix86_expand_vector_extract): Likewise.
21198 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
21199 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
21200 (ssescalarmode): Handle V4TImode and V2TImode.
21201 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
21202 (*vec_extractv2ti, *vec_extractv4ti): New insns.
21203 (VEXTRACTI128_MODE): New mode iterator.
21204 (splitter for *vec_extractv?ti first element): New.
21205 (VEC_INIT_MODE): New mode iterator.
21206 (vec_init<mode>): Consolidate 3 expanders into one using
21207 VEC_INIT_MODE mode iterator.
21209 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21211 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
21212 non_spilled_static_chain_regno_p.
21214 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21216 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
21218 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
21220 * bb-reorder.c (connect_traces): Allow copying of blocks within
21223 2017-07-20 Richard Biener <rguenther@suse.de>
21225 * gimple.h (gimple_phi_result): Add gphi * overload.
21226 (gimple_phi_result_ptr): Likewise.
21227 (gimple_phi_arg): Likewise. Adjust index assert to only
21228 allow actual argument accesses rather than all slots available
21230 (gimple_phi_arg_def): Add gphi * overload.
21231 * tree-phinodes.c (make_phi_node): Initialize only actual
21233 (resize_phi_node): Clear memory not covered by old node,
21234 do not initialize excess argument slots.
21235 (reserve_phi_args_for_new_edge): Initialize new argument slot
21238 2017-07-20 Bin Cheng <bin.cheng@arm.com>
21240 PR tree-optimization/81388
21242 2016-07-21 Bin Cheng <bin.cheng@arm.com>
21244 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
21245 by removing computation of may_be_zero.
21247 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21248 Tom de Vries <tom@codesourcery.com>
21250 PR middle-end/81030
21251 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
21252 when gimple level profile disagrees with what RTL expander did.
21254 2017-07-20 Richard Biener <rguenther@suse.de>
21256 PR tree-optimization/61171
21257 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
21258 (vect_analyze_stmt): Add slp instance parameter.
21259 (vectorizable_reduction): Likewise.
21260 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
21261 (vect_is_simple_reduction): Deal with chains not detected
21262 as SLP reduction chain, specifically not properly associated
21263 chains containing a mix of plus/minus.
21264 (get_reduction_op): Remove.
21265 (get_initial_defs_for_reduction): Simplify, pass in whether
21266 this is a reduction chain, pass in the SLP node for the PHIs.
21267 (vect_create_epilog_for_reduction): Get the SLP instance as
21269 (vectorizable_reduction): Get the SLP instance as arg.
21270 During analysis remember the SLP node with the PHIs in the
21271 instance. Simplify getting at the vectorized reduction PHIs.
21272 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
21273 through SLP instance.
21274 (vect_slp_analyze_operations): Likewise.
21275 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
21276 (vect_transform_stmt): Likewise.
21278 2017-07-20 Tom de Vries <tom@codesourcery.com>
21280 PR tree-optimization/81489
21281 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
21282 read of phi arg location to before loop that modifies phi.
21284 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21286 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
21289 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21291 PR middle-end/81331
21292 * except.c (execute): Fix ordering issue.
21294 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21296 PR rtl-optimization/81423
21297 * combine.c (make_compound_operation_int): Don't try to optimize
21298 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
21300 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21302 PR rtl-optimization/81423
21303 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
21304 with a constant that is -1 in the truncated to mode.
21306 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21308 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
21309 (determine_unlikely_bbs): ... here.
21310 * predict.h (propagate_unlikely_bbs_forward): Declare.
21311 * cfgexpand.c (pass_expand::execute): Use it.
21312 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
21314 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
21315 propagate_unlikely_bbs_forward.
21317 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21319 PR middle-end/81331
21320 * except.c (maybe_add_nop_after_section_switch): New function.
21323 2017-07-19 Tom de Vries <tom@codesourcery.com>
21325 * gimple.h (gimple_phi_set_arg): Make assert more strict.
21327 2017-07-19 Tom de Vries <tom@codesourcery.com>
21329 * gimple.h (gimple_phi_arg): Make assert more strict.
21331 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
21333 * config.gcc (powerpc*-*-*): Add mmintrin.h.
21334 * config/rs6000/mmintrin.h: New file.
21335 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
21337 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21339 PR tree-optimization/81346
21340 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
21342 2017-07-19 Tom de Vries <tom@codesourcery.com>
21344 * config/nvptx/nvptx.md (VECIM): Add V2DI.
21346 2017-07-19 Tom de Vries <tom@codesourcery.com>
21348 * config/nvptx/nvptx-modes.def: Add V2DImode.
21349 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
21350 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
21351 (nvptx_output_mov_insn): Handle lack of mov.b128.
21352 (nvptx_print_operand): Handle 'H' and 'L' codes.
21353 (nvptx_vector_mode_supported): Allow V2DImode.
21354 (nvptx_preferred_simd_mode): New function.
21355 (nvptx_data_alignment): New function.
21356 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
21357 nvptx_preferred_simd_mode.
21358 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
21360 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
21362 2017-07-19 Tom de Vries <tom@codesourcery.com>
21364 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
21365 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
21366 (nvptx_vector_mode_supported): New function. Allow V2SImode.
21367 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
21368 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
21369 (mov<VECIM>_insn): New define_insn.
21370 (define_expand "mov<VECIM>): New define_expand.
21372 2017-07-19 Tom de Vries <tom@codesourcery.com>
21374 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
21376 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21378 PR tree-optimization/81346
21379 * fold-const.h (fold_div_compare, range_check_type): Declare.
21380 * fold-const.c (range_check_type): New function.
21381 (build_range_check): Use range_check_type.
21382 (fold_div_compare): No longer static, rewritten into
21383 a match.pd helper function.
21384 (fold_comparison): Don't call fold_div_compare here.
21385 * match.pd (X / C1 op C2): New optimization using fold_div_compare
21386 as helper function.
21388 2017-07-19 Nathan Sidwell <nathan@acm.org>
21390 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
21391 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
21392 * tree.c (find_decls_types_r, verify_type): Use
21393 TYPE_{MIN,MAX}_VALUE_RAW.
21394 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
21395 (hash_tree): Likewise.
21396 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21398 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21401 2017-07-18 Tom de Vries <tom@codesourcery.com>
21403 PR middle-end/81464
21404 * omp-expand.c (expand_omp_for_static_chunk): Handle
21405 equal-argument loop exit phi.
21407 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
21410 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
21411 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
21412 operand 2 predicate.
21413 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
21414 operand 2 predicate.
21415 (ror,rol -> rorx splitters): Use const_int_operand as
21416 operand 2 predicate.
21418 2017-06-18 Richard Biener <rguenther@suse.de>
21420 PR tree-optimization/81410
21421 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
21422 the gap in the ! slp_perm SLP case after each group.
21424 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21426 PR middle-end/81463
21427 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
21430 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21432 PR middle-end/81462
21433 * predict.c (set_even_probabilities): Cleanup; do not affect
21434 probabilities that are already known.
21435 (combine_predictions_for_bb): Call even when count is set.
21437 2017-07-18 Nathan Sidwell <nathan@acm.org>
21439 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
21442 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21445 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
21446 optimization for loop niter analysis.
21448 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
21451 * config/avr/avr.c (avr_optimize_casesi): Don't use
21452 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
21454 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
21456 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
21457 body_cost_vec from _vect_peel_extended_info.
21458 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
21459 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
21462 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21464 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
21466 2017-07-18 Richard Biener <rguenther@suse.de>
21468 PR tree-optimization/80620
21469 PR tree-optimization/81403
21470 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
21471 info when re-using a VN table entry.
21473 2017-07-18 Richard Biener <rguenther@suse.de>
21475 PR tree-optimization/81418
21476 * tree-vect-loop.c (vectorizable_reduction): Properly compute
21477 vectype_in. Verify that with lane-reducing reduction operations
21478 we have a single def-use cycle.
21480 2017-07-17 Carl Love <cel@us.ibm.com>
21482 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
21484 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21485 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21486 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21487 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21488 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21489 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21490 VMULOSW): New enum "unspec" values.
21491 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21492 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21493 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21494 altivec_vmulosw): New patterns.
21495 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21496 VMULOSW): Add definitions.
21498 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
21500 * config/alpha/alpha.c: Include predict.h.
21502 2017-07-17 Yury Gribov <tetra2005@gmail.com>
21504 * tree-vrp.c (compare_assert_loc): Fix comparison function
21505 to return predictable results.
21507 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21509 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
21511 (subdi3): Likewise.
21512 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
21513 * doc/invoke.texi (mexpand-adddi): Update text.
21515 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21517 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
21518 that also clobbers the CC register. The old expand code is moved
21520 (*arc_clzsi2): ... here.
21521 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
21522 the CC register. The old expand code is moved to ...
21523 (arc_ctzsi2): ... here.
21525 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21527 * config/arc/arc.opt (mindexed-loads): Use initial value
21528 TARGET_INDEXED_LOADS_DEFAULT.
21529 (mauto-modify-reg): Use initial value
21530 TARGET_AUTO_MODIFY_REG_DEFAULT.
21531 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21532 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21533 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21534 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21536 2017-07-17 Martin Liska <mliska@suse.cz>
21539 * opts.c (finish_options): Do not allow -fgnu-tm
21540 w/ -fsanitize={kernel-,}address. Say sorry.
21542 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21545 * tree-loop-distribution.c (classify_partition): Only assert on
21546 numer of iterations.
21547 (merge_dep_scc_partitions): Delete prameter. Update function call.
21548 (distribute_loop): Remove code handling loop with unknown niters.
21549 (pass_loop_distribution::execute): Skip loop with unknown niters.
21551 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21554 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
21555 function sort_partitions_by_post_order.
21557 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21559 PR tree-optimization/81374
21560 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
21561 the max index of basic blocks, rather than number of basic blocks.
21563 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21565 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
21567 (arc_legitimate_pic_operand_p): Likewise.
21568 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
21570 (arc_needs_pcl_p): Likewise.
21571 (arc_legitimate_pc_offset_p): Likewise.
21572 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
21573 function is also used in constrains.md.
21574 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
21575 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
21576 PLUS. Only return true/false in known cases, otherwise assert.
21577 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
21578 is already called in arc_legitimate_constant_p.
21579 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
21581 (LEGITIMATE_PIC_OPERAND_P): Use
21582 arc_raw_symbolic_reference_mentioned_p function.
21583 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
21588 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21589 Andrew Burgess <andrew.burgess@embecosm.com>
21591 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
21592 (arc_return_address_register): New function.
21593 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
21594 (arc_handle_fndecl_attribute): Add naked attribute.
21595 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21596 (TARGET_WARN_FUNC_RETURN): Likewise.
21597 (arc_allocate_stack_slots_for_args): New function.
21598 (arc_warn_func_return): Likewise.
21599 (machine_function): Change type fn_type.
21600 (arc_compute_function_type): Consider new naked function type,
21601 change function return type.
21602 (arc_must_save_register): Adapt to handle new
21603 arc_compute_function_type's return type.
21604 (arc_expand_prologue): Likewise.
21605 (arc_expand_epilogue): Likewise.
21606 (arc_return_address_regs): Delete.
21607 (arc_return_address_register): New function.
21608 (arc_epilogue_uses): Use above function.
21609 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
21610 (arc_function_type): Change encoding, add naked type.
21611 (ARC_INTERRUPT_P): Change to handle the new encoding.
21612 (ARC_FAST_INTERRUPT_P): Likewise.
21613 (ARC_NORMAL_P): Define.
21614 (ARC_NAKED_P): Likewise.
21615 (arc_compute_function_type): Delete prototype.
21616 * config/arc/arc.md (in_ret_delay_slot): Use
21617 arc_return_address_register function.
21618 (simple_return): Likewise.
21619 (p_return_i): Likewise.
21621 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21623 PR tree-optimization/81428
21624 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
21625 can't be built for those types.
21627 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21629 Remove stuff dead since r239246.
21631 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
21632 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
21633 (avr_inform_devices): Remove dead stuff.
21635 2017-07-17 Tamar Christina <tamar.christina@arm.com>
21637 * config/arm/arm_neon.h: Fix softp typo.
21639 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21641 PR tree-optimization/81365
21642 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
21643 aggregate moves onto bb predecessor edges, make sure there are no
21644 loads that could alias the lhs in between the start of bb and the
21647 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21650 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
21651 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
21652 [LSHIFTRT, outer_code = TRUNCATE]: Same.
21654 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21656 PR tree-optimization/81396
21657 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
21658 (init_symbolic_number): Initialize it to 1.
21659 (perform_symbolic_merge): Add n_ops from both operands into the new
21661 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
21662 without base_addr as useless if they need more than one operation.
21663 (bswap_replace): Handle !bswap case for NULL base_addr.
21665 2017-07-17 Tom de Vries <tom@codesourcery.com>
21668 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
21671 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
21673 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
21674 conditional builtin define __FIX_LEON3FT_B2BST.
21676 2017-07-17 Daniel Cederman <cederman@gaisler.com>
21678 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
21679 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
21682 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21684 PR rtl-optimization/81424
21685 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
21686 to remove potential trapping from operands if -fnon-call-exceptions.
21688 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21690 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
21691 profile_proability for scalling.
21692 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
21694 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21696 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
21698 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21700 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
21701 fixpoint arithmetics.
21703 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21705 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
21706 fixpoint arithmetics.
21708 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21710 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
21711 fixpoint arithmetics.
21713 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21715 * profile-count.h (profile_probability::from_reg_br_prob_note,
21716 profile_probability::to_reg_br_prob_note): New functions.
21717 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
21718 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
21719 * predict.c (probability_reliable_p): Update.
21720 (edge_probability_reliable_p): Update.
21721 (br_prob_note_reliable_p): Update.
21722 (invert_br_probabilities): Update.
21723 (add_reg_br_prob_note): New function.
21724 (combine_predictions_for_insn): Update.
21725 * asan.c (asan_clear_shadow): Update.
21726 * cfgbuild.c (compute_outgoing_frequencies): Update.
21727 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21728 (update_br_prob_note): Update.
21729 (rtl_verify_edges): Update.
21730 (purge_dead_edges): Update.
21731 (fixup_reorder_chain): Update.
21732 * emit-rtl.c (try_split): Update.
21733 * ifcvt.c (cond_exec_process_insns): Update.
21734 (cond_exec_process_if_block): Update.
21735 (dead_or_predicable): Update.
21736 * internal-fn.c (expand_addsub_overflow): Update.
21737 (expand_neg_overflow): Update.
21738 (expand_mul_overflow): Update.
21739 * loop-doloop.c (doloop_modify): Update.
21740 * loop-unroll.c (compare_and_jump_seq): Update.
21741 * optabs.c (emit_cmp_and_jump_insn_1): Update.
21742 * predict.h: Update.
21743 * reorg.c (mostly_true_jump): Update.
21745 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
21746 * config/alpha/alpha.c (emit_unlikely_jump): Update.
21747 * config/arc/arc.c: (emit_unlikely_jump): Update.
21748 * config/arm/arm.c: (emit_unlikely_jump): Update.
21749 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
21750 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
21751 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
21752 (ix86_print_operand): Update.
21753 (ix86_split_fp_branch): Update.
21754 (predict_jump): Update.
21755 * config/ia64/ia64.c (ia64_print_operand): Update.
21756 * config/mmix/mmix.c (mmix_print_operand): Update.
21757 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
21758 (rs6000_expand_split_stack_prologue): Update.
21759 * config/rs6000/rs6000.c: Update.
21760 * config/s390/s390.c (s390_expand_vec_strlen): Update.
21761 (s390_expand_vec_movstr): Update.
21762 (s390_expand_cs_tdsi): Update.
21763 (s390_expand_split_stack_prologue): Update.
21764 * config/sh/sh.c (sh_print_operand): Update.
21765 (expand_cbranchsi4): Update.
21766 (expand_cbranchdi4): Update.
21767 * config/sparc/sparc.c (output_v9branch): Update.
21768 * config/spu/spu.c (get_branch_target): Update.
21769 (ea_load_store_inline): Update.
21770 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
21771 * config/tilepro/tilepro.c: Update.
21773 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21775 * gimplify.c (mostly_copy_tree_r): Revert latest change.
21776 (gimplify_save_expr): Likewise.
21778 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21780 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
21782 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21784 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
21786 * timevar.def (TV_IPA_FNSUMMARY): Define.
21788 2017-07-16 Daniel Cederman <cederman@gaisler.com>
21790 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
21791 to back store errata sensitive sequence from being generated.
21792 (sqrtdf2_fix): Likewise.
21794 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21796 * tree-ssa-threadupdate.c (compute_path_counts,
21797 update_joiner_offpath_counts): Use profile_probability.
21799 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
21802 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21804 * config/arm/arm-c.c (arm_cpu_builtins): Define
21805 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21807 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
21809 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21810 array entries to represent __ieee128 versions of the
21811 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
21812 scalar_extract_sig, and scalar_insert_exp built-in functions.
21813 (altivec_resolve_overloaded_builtin): Add special case handling
21814 for the __builtin_scalar_insert_exp function, as represented by
21815 the P9V_BUILTIN_VEC_VSIEDP constant.
21816 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
21817 exponent support for __ieee128 argument.
21818 (VSESQP): Add scalar extract signature support for __ieee128
21820 (VSTDCNQP): Add scalar test negative support for __ieee128
21822 (VSIEQP): Add scalar insert exponent support for __int128 argument
21823 with __ieee128 result.
21824 (VSIEQPF): Add scalar insert exponent support for __ieee128
21825 argument with __ieee128 result.
21826 (VSTDCQP): Add scalar test data class support for __ieee128
21828 (VSTDCNQP): Add overload support for scalar test negative with
21829 __ieee128 argument.
21830 (VSTDCQP): Add overload support for scalar test data class
21831 __ieee128 argument.
21832 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
21833 UNSPEC_VSX_SXSIGDP.
21834 (UNSPEC_VSX_SIEXPQP): New constant.
21835 (xsxexpqp): New insn for VSX scalar extract exponent quad
21837 (xsxsigqp): New insn for VSX scalar extract significand quad
21839 (xsiexpqpf): New insn for VSX scalar insert exponent quad
21840 precision with floating point argument.
21841 (xststdcqp): New expand for VSX scalar test data class quad
21843 (xststdcnegqp): New expand for VSX scalar test negative quad
21845 (xststdcqp): New insn to match expansions for VSX scalar test data
21846 class quad precision and VSX scalar test negative quad precision.
21847 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
21848 special case operand checking to enforce that second operand of
21849 VSX scalar test data class with quad precision argument is a 7-bit
21851 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21852 prototypes and descriptions of __ieee128 versions of
21853 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
21854 scalar_test_data_class, and scalar_test_neg built-in functions.
21856 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21858 PR tree-optimization/81162
21859 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
21860 replace a negate with an add.
21862 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
21864 * doc/invoke.texi (arm/-mcpu): Document +crypto.
21866 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21868 * config/arm/arm-c.c (arm_cpu_builtins): Define
21869 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21871 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21873 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
21874 (armv8-r): Set ARM Cortex-R52 as default CPU.
21875 * config/arm/arm-tables.opt: Regenerate.
21876 * config/arm/arm-tune.md: Regenerate.
21877 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21879 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
21880 extension for -mcpu=cortex-r52.
21882 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21884 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
21885 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
21886 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
21887 (fp-armv8): Define it as FP_ARMv8 only.
21888 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
21889 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
21891 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
21892 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
21893 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
21894 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
21895 than TARGET_FPU_ARMV8.
21896 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
21897 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
21898 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
21900 * config/arm/neon.md (neon_vrint): Likewise.
21901 (neon_vcvt): Likewise.
21902 (neon_<fmaxmin_op><mode>): Likewise.
21903 (<fmaxmin><mode>3): Likewise.
21904 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
21905 * config/arm/predicates.md (arm_cond_move_operator): Check against
21906 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
21908 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
21910 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
21911 to top of function.
21913 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21915 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
21916 loop in comment with memset.
21918 2017-07-14 Martin Liska <mliska@suse.cz>
21920 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
21921 * dwarf2out.c (is_java): Remove the function.
21922 (output_pubname): Remove usage of the function.
21923 (lower_bound_default): Remove usage of DW_LANG_Java.
21924 (gen_compile_unit_die): Likewise.
21925 * gcc.c: Remove compiler defaults for .java and .zip files.
21926 * gimple-expr.c (remove_suffix): Change as there's no longer
21927 extension than 4-letter one.
21928 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21929 (gimplify_save_expr): Likewise.
21930 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21931 as it's possible even for other languages than Java.
21932 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21933 * lto-opts.c (lto_write_options): Remove reference to Java.
21934 * opts.c (strip_off_ending): Update file extension handling.
21935 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21936 * tree-eh.c (lower_resx): Likewise.
21937 * tree.c (free_lang_data_in_type): Remove dead code.
21938 (find_decls_types_r): Likewise.
21939 (build_common_builtin_nodes): Remove Java from a comment.
21940 (verify_type): Remove dead code.
21941 * varasm.c (assemble_external): Remove Java from a comment.
21943 2017-07-14 Martin Liska <mliska@suse.cz>
21945 * opts.c (finish_options): Add quotes.
21946 (common_handle_option): Likewise.
21948 2017-07-14 Martin Liska <mliska@suse.cz>
21950 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21951 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21952 Remove N_SO_PASCAL.
21953 * dwarf2out.c (lower_bound_default): Do not handle
21955 (gen_compile_unit_die): Likewise.
21956 * gcc.c: Remove default extension binding for GNU Pascal.
21957 * stmt.c: Remove Pascal language from a comment.
21958 * xcoffout.c: Likewise.
21960 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21963 * diagnostic-show-locus.c (fixit_cmp): New function.
21964 (layout::layout): Sort m_fixit_hints.
21965 (column_range::column_range): Assert that the values are valid.
21966 (struct char_span): New struct.
21967 (correction::overwrite): New method.
21968 (struct source_line): New struct.
21969 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21970 calls in terms of classes source_line and char_span, and
21971 correction::overwrite.
21972 (selftest::test_overlapped_fixit_printing_2): New function.
21973 (selftest::diagnostic_show_locus_c_tests): Call it.
21975 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21977 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21978 early if there is no lhs.
21980 2017-07-13 Martin Liska <mliska@suse.cz>
21982 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21983 (gen_reference_type_die): Likewise.
21984 * stor-layout.c: Remove Pascal-related comment.
21986 2017-07-13 Martin Liska <mliska@suse.cz>
21988 * opts.c (finish_options): Add quotes to error messages.
21989 (parse_sanitizer_options): Likewise.
21991 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21993 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21995 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21997 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21999 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22001 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
22003 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
22005 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22008 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
22009 provides the hardware capability bits, define the macro
22010 __BUILTIN_CPU_SUPPORTS__.
22011 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
22012 if GLIBC does not provide the hardware capability bits. Add a
22013 gcc_unreachable call if the built-in cpu function is neither
22014 __builtin_cpu_is nor __builtin_cpu_supports.
22015 (rs6000_get_function_versions_dispatcher): Change the warning
22016 that an old GLIBC is used which does not export the capability
22017 bits to be an error.
22018 * doc/extend.texi (target_clones attribute): Document the
22019 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
22020 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
22021 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
22022 the macros defined by GCC if the newer GLIBC is available.
22024 2017-07-12 Jeff Law <law@redhat.com>
22026 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
22027 remaining includes slightly.
22028 * config/riscv/riscv-builtins.c: Include profile-count.h.
22030 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22033 * config/avr/avr.c (avr_set_current_function): In diagnostic
22034 messages: Quote keywords and (parts of) identifiers.
22035 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
22038 2017-07-12 Carl Love <cel@us.ibm.com>
22040 * config/rs6000/rs6000-c.c: Add support for built-in functions
22041 vector bool char vec_revb (vector bool char);
22042 vector bool short vec_revb (vector short char);
22043 vector bool int vec_revb (vector bool int);
22044 vector bool long long vec_revb (vector bool long long);
22045 * doc/extend.texi: Update the built-in documentation file for the
22046 new built-in functions.
22048 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22050 * config/s390/s390.md: Remove movcc splitter.
22052 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22054 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
22055 load/store on condition.
22057 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
22060 * config/avr/avr.c (avr_encode_section_info)
22061 [progmem && !TREE_READONLY]: Error if progmem object needs
22064 2017-07-11 Michael Collison <michael.collison@arm.com>
22066 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
22069 2017-07-11 Carl Love <cel@us.ibm.com>
22071 * config/rs6000/rs6000-c.c: Add support for builtins
22072 vector unsigned int vec_parity_lsbb (vector signed int);
22073 vector unsigned int vec_parity_lsbb (vector unsigned int);
22074 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
22075 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
22076 vector unsigned long long vec_parity_lsbb (vector signed long long);
22077 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
22078 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
22079 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
22080 * doc/extend.texi: Update the built-in documentation file for the
22081 new built-in functions.
22083 2017-07-11 David Malcolm <dmalcolm@redhat.com>
22085 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
22086 (layout::m_primary_loc): New field.
22087 (layout::layout): Initialize new field. Move location filtering
22088 logic from here to...
22089 (layout::maybe_add_location_range): ...this new method. Add
22090 support for filtering to just the lines already specified by other
22092 (layout::will_show_line_p): New method.
22093 (gcc_rich_location::add_location_if_nearby): New method.
22094 (selftest::test_add_location_if_nearby): New test function.
22095 (selftest::diagnostic_show_locus_c_tests): Call it.
22096 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
22099 2017-07-11 Tom de Vries <tom@codesourcery.com>
22101 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
22102 (bb_first_real_insn): New function.
22103 (nvptx_single): Add extra initialization of broadcasted condition
22106 2017-07-11 Nathan Sidwell <nathan@acm.org>
22108 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
22110 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
22112 * doc/extend.texi (AVR Function Attributes): Remove weblink to
22113 Binutils doc as TEXI will mess them up.
22114 * doc/invoke.texi (AVR Options): Same here.
22116 2017-07-11 Daniel Cederman <cederman@gaisler.com>
22118 * config/sparc/sparc.opt (mfix-ut700): New option.
22119 (mfix-gr712rc): Likewise.
22120 (sparc_fix_b2bst): New variable.
22121 * doc/invoke.texi (SPARC options): Document them.
22122 (ARM options): Fix warnings.
22123 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
22124 instructions to prevent sequences that can trigger the store-store
22125 errata for certain LEON3FT processors.
22126 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
22127 (sparc_option_override): Set sparc_fix_b2bst appropriately.
22128 * config/sparc/sparc.md (fix_b2bst): New attribute.
22129 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
22131 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
22134 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
22136 (*rsqrtsf2_sse): Ditto.
22138 (div<mode>3): Macroize insn from divdf3 and divsf3
22139 using MODEF mode iterator.
22141 2017-07-10 Martin Sebor <msebor@redhat.com>
22143 PR tree-optimization/80397
22144 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
22145 instead of testing for equality to INTEGER_TYPE.
22147 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
22149 * config.gcc: Remove uclibc from arc target spec.
22151 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
22153 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
22155 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22158 * lto-wrapper.c (remove_option): New function.
22159 (merge_and_complain): Merge PIC/PIE options more realistically.
22161 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22163 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
22167 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
22168 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
22169 * config.in: Regenerate.
22170 * configure: Regenerate.
22171 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
22172 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
22173 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
22174 (TARGET_GASISR_PROLOGUES): ...target mask.
22175 * common/config/avr/avr-common.c
22176 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
22177 Set -mgas-isr-prologues.
22178 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
22179 INSERT_PASS_BEFORE for it.
22180 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
22181 * config/avr/avr.c (avr_option_override)
22182 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
22183 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
22184 (avr_attribute_table) <no_gccisr>: Add new function attribute.
22185 (avr_set_current_function) <is_no_gccisr>: Init machine field.
22186 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
22188 (make_avr_pass_pre_proep): New function.
22189 (emit_push_sfr) <treg>: Add argument to function and use it
22190 instead of TMP_REG.
22191 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
22192 and set machine->gasisr.yes.
22193 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
22194 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
22195 __gcc_isr.n_pushed to .L__stack_usage.
22196 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
22197 (avr_asm_final_postscan_insn): ...this new static function.
22198 * config/avr/avr.h (machine_function)
22199 <is_no_gccisr, use_L__stack_usage>: New fields.
22200 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
22201 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
22202 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
22203 (gasisr, *gasisr): New expander and insn.
22204 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22205 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
22206 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
22208 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
22210 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
22213 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22215 Move jump-tables out of .text again.
22218 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
22219 (ASM_OUTPUT_ADDR_VEC): New function.
22220 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
22221 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
22222 INSN_ADDRESSes as asm comment.
22223 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
22224 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
22225 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
22226 * config/avr/avr.md (*tablejump): Adjust comment.
22227 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
22228 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
22230 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
22231 (avr_output_addr_vec): New proto.
22232 (avr_log_t) <insn_addresses>: New field.
22234 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
22237 * config/i386/i386.c (ix86_function_arg_advance): Set
22238 outgoing_args_on_stack to true if there are outgoing arguments
22240 (ix86_function_arg): Likewise.
22241 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
22242 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
22243 * config/i386/i386.h (machine_function): Add
22244 outgoing_args_on_stack.
22246 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
22248 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22249 supporting pthreds.
22250 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
22252 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
22254 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
22255 (REAL_H): Remove $(MACHMODE_H).
22256 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
22258 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
22259 $(MACHMODE_H) and double-int.h.
22260 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
22262 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
22265 2017-07-07 Andrew Pinski <apinski@cavium.com>
22267 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
22268 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
22270 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22272 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
22273 Add warning if GCC was not configured to link against a GLIBC that
22274 exports the hardware capability bits.
22275 (make_resolver_func): Make resolver function private and not a
22276 COMDAT function. Create the name with clone_function_name instead
22277 of make_unique_name.
22280 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
22281 correct operand in doing the split.
22283 2017-07-07 Carl Love <cel@us.ibm.com>
22285 * config/rs6000/rs6000-c: Add support for built-in function
22286 vector unsigned short vec_pack_to_short_fp32 (vector float,
22288 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
22289 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
22290 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
22291 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
22292 (convert_4f32_8i16): Add define_expand.
22293 * doc/extend.texi: Update the built-in documentation file for the
22294 new built-in function.
22296 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22298 * config/sparc/m8.md: New file.
22299 * config/sparc/sparc.md: Include m8.md.
22301 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22303 * config/sparc/sparc.opt: New option -mvis4b.
22304 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
22305 (sparc_option_override): Handle VIS4B.
22306 (enum sparc_builtins): Define
22307 SPARC_BUILTIN_DICTUNPACK{8,16,32},
22308 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
22309 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
22310 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
22311 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
22312 (check_constant_argument): New function.
22313 (sparc_vis_init_builtins): Define builtins
22314 __builtin_vis_dictunpack{8,16,32},
22315 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
22316 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
22317 __builtin_vis_fpcmpde{8,16,32}shl and
22318 __builtin_vis_fpcmpur{8,16,32}shl.
22319 (sparc_expand_builtin): Check that the constant operands to
22320 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
22321 constant and in range.
22322 * config/sparc/sparc-c.c (sparc_target_macros): Handle
22324 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
22325 (SPARC_IMM5_P): Likewise.
22326 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
22327 (enabled): Handle vis4b.
22328 (UNSPEC_DICTUNPACK): New unspec.
22329 (UNSPEC_FPCMPSHL): Likewise.
22330 (UNSPEC_FPUCMPSHL): Likewise.
22331 (UNSPEC_FPCMPDESHL): Likewise.
22332 (UNSPEC_FPCMPURSHL): Likewise.
22333 (cpu_feature): New CPU feature `vis4b'.
22334 (dictunpack{8,16,32}): New insns.
22335 (FPCSMODE): New mode iterator.
22336 (fpcscond): New code iterator.
22337 (fpcsucond): Likewise.
22338 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
22339 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
22340 (fpcmpde{8,16,32}{si,di}shl): Likewise.
22341 (fpcmpur{8,16,32}{si,di}shl): Likewise.
22342 * config/sparc/constraints.md: Define constraints `q' for unsigned
22343 2-bit integer constants and `t' for unsigned 5-bit integer
22345 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
22347 (imm5_operand_dictunpack16): Likewise.
22348 (imm5_operand_dictunpack32): Likewise.
22349 (imm2_operand): Likewise.
22350 * doc/invoke.texi (SPARC Options): Document -mvis4b.
22351 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
22352 ditunpack* and fpcmp*shl builtins.
22354 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22356 * config.gcc: Handle m8 in --with-{cpu,tune} options.
22357 * config.in: Add HAVE_AS_SPARC6 define.
22358 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
22360 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
22362 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
22363 (CPP_CPU_SPEC): Handle m8.
22364 (ASM_CPU_SPEC): Likewise.
22365 * config/sparc/sparc-opts.h (enum processor_type): Add
22367 * config/sparc/sparc.c (m8_costs): New struct.
22368 (sparc_option_override): Handle TARGET_CPU_m8.
22369 (sparc32_initialize_trampoline): Likewise.
22370 (sparc64_initialize_trampoline): Likewise.
22371 (sparc_issue_rate): Likewise.
22372 (sparc_register_move_cost): Likewise.
22373 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
22374 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
22375 (ASM_CPU64_DEFAULT_SPEC): Likewise.
22376 (CPP_CPU_SPEC): Handle M8.
22377 (ASM_CPU_SPEC): Likewise.
22378 (AS_M8_FLAG): Define.
22379 * config/sparc/sparc.md: Add m8 to the cpu attribute.
22380 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
22381 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
22383 * configure: Regenerate.
22384 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
22387 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22389 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
22391 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
22392 ("*movdi_insn_sp32"): Do not set v3pipe.
22393 ("*movsi_insn"): Likewise.
22394 ("*movdi_insn_sp64"): Likewise.
22395 ("*movsf_insn"): Likewise.
22396 ("*movdf_insn_sp32"): Likewise.
22397 ("*movdf_insn_sp64"): Likewise.
22398 ("*zero_extendsidi2_insn_sp64"): Likewise.
22399 ("*sign_extendsidi2_insn"): Likewise.
22400 ("*mov<VM32:mode>_insn"): Likewise.
22401 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22402 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22403 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22404 ("<vlop:code><VL:mode>3"): Likewise.
22405 ("*not_<vlop:code><VL:mode>3"): Likewise.
22406 ("*nand<VL:mode>_vis"): Likewise.
22407 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
22408 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
22409 ("one_cmpl<VL:mode>2"): Likewise.
22410 ("faligndata<VM64:mode>_vis"): Likewise.
22411 ("alignaddrsi_vis"): Likewise.
22412 ("alignaddrdi_vis"): Likweise.
22413 ("alignaddrlsi_vis"): Likewise.
22414 ("alignaddrldi_vis"): Likewise.
22415 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22416 ("bmaskdi_vis"): Likewise.
22417 ("bmasksi_vis"): Likewise.
22418 ("bshuffle<VM64:mode>_vis"): Likewise.
22419 ("cmask8<P:mode>_vis"): Likewise.
22420 ("cmask16<P:mode>_vis"): Likewise.
22421 ("cmask32<P:mode>_vis"): Likewise.
22422 ("pdistn<P:mode>_vis"): Likewise.
22423 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22425 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22427 * config/sparc/sparc.md ("subtype"): New insn attribute.
22428 ("*wrgsr_sp64"): Set insn subtype.
22429 ("*rdgsr_sp64"): Likewise.
22430 ("alignaddrsi_vis"): Likewise.
22431 ("alignaddrdi_vis"): Likewise.
22432 ("alignaddrlsi_vis"): Likewise.
22433 ("alignaddrldi_vis"): Likewise.
22434 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22435 ("fexpand_vis"): Likewise.
22436 ("fpmerge_vis"): Likewise.
22437 ("faligndata<VM64:mode>_vis"): Likewise.
22438 ("bshuffle<VM64:mode>_vis"): Likewise.
22439 ("cmask8<P:mode>_vis"): Likewise.
22440 ("cmask16<P:mode>_vis"): Likewise.
22441 ("cmask32<P:mode>_vis"): Likewise.
22442 ("fchksm16_vis"): Likewise.
22443 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
22444 ("fmean16_vis"): Likewise.
22445 ("fp<plusminus_insn>64_vis"): Likewise.
22446 ("<plusminus_insn>v8qi3"): Likewise.
22447 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22448 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
22449 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
22450 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
22451 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
22452 ("*movqi_insn"): Likewise.
22453 ("*movhi_insn"): Likewise.
22454 ("*movsi_insn"): Likewise.
22455 ("movsi_pic_gotdata_op"): Likewise.
22456 ("*movdi_insn_sp32"): Likewise.
22457 ("*movdi_insn_sp64"): Likewise.
22458 ("movdi_pic_gotdata_op"): Likewise.
22459 ("*movsf_insn"): Likewise.
22460 ("*movdf_insn_sp32"): Likewise.
22461 ("*movdf_insn_sp64"): Likewise.
22462 ("*zero_extendhisi2_insn"): Likewise.
22463 ("*zero_extendqihi2_insn"): Likewise.
22464 ("*zero_extendqisi2_insn"): Likewise.
22465 ("*zero_extendqidi2_insn"): Likewise.
22466 ("*zero_extendhidi2_insn"): Likewise.
22467 ("*zero_extendsidi2_insn_sp64"): Likewise.
22468 ("ldfsr"): Likewise.
22469 ("prefetch_64"): Likewise.
22470 ("prefetch_32"): Likewise.
22471 ("tie_ld32"): Likewise.
22472 ("tie_ld64"): Likewise.
22473 ("*tldo_ldub_sp32"): Likewise.
22474 ("*tldo_ldub1_sp32"): Likewise.
22475 ("*tldo_ldub2_sp32"): Likewise.
22476 ("*tldo_ldub_sp64"): Likewise.
22477 ("*tldo_ldub1_sp64"): Likewise.
22478 ("*tldo_ldub2_sp64"): Likewise.
22479 ("*tldo_ldub3_sp64"): Likewise.
22480 ("*tldo_lduh_sp32"): Likewise.
22481 ("*tldo_lduh1_sp32"): Likewise.
22482 ("*tldo_lduh_sp64"): Likewise.
22483 ("*tldo_lduh1_sp64"): Likewise.
22484 ("*tldo_lduh2_sp64"): Likewise.
22485 ("*tldo_lduw_sp32"): Likewise.
22486 ("*tldo_lduw_sp64"): Likewise.
22487 ("*tldo_lduw1_sp64"): Likewise.
22488 ("*tldo_ldx_sp64"): Likewise.
22489 ("*mov<VM32:mode>_insn"): Likewise.
22490 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22491 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22493 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22495 * config/sparc/sparc.md ("type"): New insn type viscmp.
22496 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
22498 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
22499 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
22500 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22501 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
22503 ("n7_vis_logical_11cycle"): Likewise.
22504 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
22505 * config/sparc/niagara2.md ("niag3_vis": Likewise.
22506 * config/sparc/niagara.md ("niag_vis"): Likewise.
22507 * config/sparc/ultra3.md ("us3_fga"): Likewise.
22508 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
22510 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22512 * config/sparc/sparc.md: New instruction type `bmask'.
22513 (bmaskdi_vis): Use the `bmask' type.
22514 (bmasksi_vis): Likewise.
22515 * config/sparc/ultra3.md (us3_array): Likewise.
22516 * config/sparc/niagara7.md (n7_array): Likewise.
22517 * config/sparc/niagara4.md (n4_array): Likewise.
22518 * config/sparc/niagara2.md (niag2_vis): Likewise.
22519 (niag3_vis): Likewise.
22520 * config/sparc/niagara.md (niag_vis): Likewise.
22522 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22524 * ipa-comdats.c: Remove optimize check from gate.
22525 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
22526 for functions not optimized.
22527 (ipa_fn_summary_read): Skip optimize check.
22528 (ipa_fn_summary_write): Likewise.
22529 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
22531 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
22533 (can_inline_edge_p): Check flag_pcc_struct_return for match.
22534 (check_callers): Give up on caller which is not optimized.
22535 (inline_small_functions): Likewise.
22536 (ipa_inline): Do not give up when not optimizing.
22537 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
22538 away unoptimizes cdtors.
22539 (whole_program_function_and_variable_visibility): Do
22540 ipa_discover_readonly_nonaddressable_vars in LTO mode.
22541 * ipa.c (process_references): Do not check optimize.
22542 (symbol_table::remove_unreachable_nodes): Update optimize check.
22543 (set_writeonly_bit): Update optimize check.
22544 (pass_ipa_cdtor_merge::gate): Do not check optimize.
22545 (pass_ipa_single_use::gate): Remove.
22547 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22549 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
22550 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
22551 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
22552 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
22553 permute_load, permute_store, adjust_extract, adjust_splat,
22554 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
22555 replace_swap_with_copy, dump_swap_insn_table,
22556 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
22557 recombine_lvx_pattern, recombine_stvx_pattern,
22558 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
22559 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
22560 to file rs6000-p8swap.c.
22561 * config/rs6000/rs6000-p8swap.c: New file.
22562 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
22563 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
22564 and rs6000*-*-* targets.
22566 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22568 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
22570 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22572 * lto-wrapper.c (merge_and_complain): Do not merge
22573 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
22574 fsigned_zeros, ftrapping_math, fwrapv.
22575 (append_compiler_options): Do not track these options.
22576 (append_linker_options): Likewie
22578 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22580 * cgraphunit.c (cgraph_node::finalize_function): When
22581 !flag_toplevel_reorde set no_reorder flag.
22582 (varpool_node::finalize_decl): Likewise.
22583 (symbol_table::compile): Drop no toplevel reorder path.
22585 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22587 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
22588 edges; zero probability is not better than uninitialized.
22590 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22592 * asan.h (asan_sanitize_allocas_p): Declare.
22593 * asan.c (asan_sanitize_allocas_p): New function.
22594 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
22595 (handle_builtin_alloca): Likewise.
22596 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
22597 if !asan_sanitize_allocas_p.
22598 * params.def (asan-instrument-allocas): Add new option.
22599 * params.h (ASAN_PROTECT_ALLOCAS): Define.
22600 * opts.c (common_handle_option): Disable allocas sanitization for
22603 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22605 * asan.c: Include gimple-fold.h.
22606 (get_last_alloca_addr): New function.
22607 (handle_builtin_stackrestore): Likewise.
22608 (handle_builtin_alloca): Likewise.
22609 (asan_emit_allocas_unpoison): Likewise.
22610 (get_mem_refs_of_builtin_call): Add new parameter, remove const
22611 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
22612 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
22613 (instrument_builtin_call): Pass gimple iterator to
22614 get_mem_refs_of_builtin_call.
22615 (last_alloca_addr): New global.
22616 * asan.h (asan_emit_allocas_unpoison): Declare.
22617 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
22618 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
22619 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
22620 if function calls alloca.
22621 * gimple-fold.c (replace_call_with_value): Remove static keyword.
22622 * gimple-fold.h (replace_call_with_value): Declare.
22623 * internal-fn.c: Include asan.h.
22624 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
22625 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
22627 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22629 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
22630 (C_SELFTEST_FLAGS): New.
22631 (CPP_SELFTEST_FLAGS): New.
22632 (SELFTEST_DEPS): New, from deps of s-selftest.
22633 (C_SELFTEST_DEPS): New, from deps of s-selftest.
22634 (CPP_SELFTEST_DEPS): New.
22635 (selftest): Add dependency on s-selftest-c++.
22636 (s-selftest): Rename to...
22637 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
22638 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
22639 than SELFTEST_FLAGS.
22640 (selftest-gdb): Rename to...
22641 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
22643 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
22644 (selftest-valgrind): Rename to...
22645 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
22647 (selftest-valgrind): Reintroduce as an alias for
22648 selftest-c-valgrind.
22649 (s-selftest-c++): New.
22650 (selftest-c++-gdb): New.
22651 (selftest-c++-valgrind): New.
22653 2017-07-06 Olivier Hainque <hainque@adacore.com>
22655 * gcc.c (process_command): When deciding if undefined variables
22656 should be ignored when processing specs, accept "gcc -v" as well.
22658 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22660 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
22661 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
22663 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22665 * config/arm/arm-cpus.in (armv8-r): Add new entry.
22666 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
22667 * config/arm/arm-tables.opt: Regenerate.
22668 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
22670 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
22672 2017-07-06 Carl Love <cel@us.ibm.com>
22674 * ChangeLog: Clean up from mid air collision
22676 2017-07-06 Carl Love <cel@us.ibm.com>
22678 * config/rs6000/rs6000-c.c: Add support for built-in functions
22679 vector signed int vec_subc (vector signed int, vector signed int);
22680 vector signed __int128 vec_subc (vector signed __int128,
22681 vector signed __int128);
22682 vector unsigned __int128 vec_subc (vector unsigned __int128,
22683 vector unsigned __int128);
22684 vector signed int vec_sube (vector signed int, vector signed int,
22685 vector signed int);
22686 vector unsigned int vec_sube (vector unsigned int,
22687 vector unsigned int,
22688 vector unsigned int);
22689 vector signed __int128 vec_sube (vector signed __int128,
22690 vector signed __int128,
22691 vector signed__int128);
22692 vector unsigned __int128 vec_sube (vector unsigned __int128,
22693 vector unsigned __int128,
22694 vector unsigned __int128);
22695 vector signed int vec_subec (vector signed int, vector signed int,
22696 vector signed int);
22697 vector unsigned int vec_subec (vector unsigned int,
22698 vector unsigned int,
22699 vector unsigned int);
22700 vector signed __int128 vec_subec (vector signed __int128,
22701 vector signed __int128,
22702 vector signed__int128);
22703 vector unsigned __int128 vec_subec (vector unsigned __int128,
22704 vector unsigned __int128,
22705 vector unsigned __int128);
22706 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
22707 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
22708 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
22709 BU_ALTIVEC_OVERLOAD_X definitions.
22710 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
22711 * doc/extend.texi: Update the built-in documentation file for the new
22712 built-in functions.
22714 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22717 * diagnostic-show-locus.c (layout::layout): Use start and finish
22718 spelling location for the start and finish of each range.
22719 * genmatch.c (linemap_client_expand_location_to_spelling_point):
22720 Add unused aspect param.
22721 * input.c (expand_location_1): Add "aspect" param, and use it
22722 to access the correct part of the location.
22723 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
22725 (expand_location_to_spelling_point): Likewise.
22726 (linemap_client_expand_location_to_spelling_point): Add "aspect"
22727 param, and pass it to expand_location_1.
22729 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
22731 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
22732 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
22733 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
22734 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
22735 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
22736 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
22737 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
22738 _mm_maskz_getmant_ss): New intrinsics.
22739 (__builtin_ia32_getexpss128_mask): Changed to ...
22740 __builtin_ia32_getexpss128_round ... this.
22741 (__builtin_ia32_getexpsd128_mask): Changed to ...
22742 __builtin_ia32_getexpsd128_round ... this.
22743 * config/i386/i386-builtin-types.def
22744 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
22745 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
22746 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
22747 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
22748 __builtin_ia32_getmantss_mask_round): New builtins.
22749 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
22750 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
22751 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
22752 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
22753 * config/i386/sse.md
22754 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
22755 avx512f_sgetexp<mode><mask_scalar_name>
22756 <round_saeonly_scalar_name> ... this.
22757 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
22758 %0, %1, %2<round_saeonly_op3>}): Changed to ...
22759 vgetexp<ssescalarmodesuffix>
22760 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22761 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
22762 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
22763 avx512f_vgetmant<mode><mask_scalar_name>
22764 <round_saeonly_scalar_name> ... this.
22765 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
22766 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
22767 vgetmant<ssescalarmodesuffix>
22768 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
22769 %0<mask_scalar_operand4>, %1, %2
22770 <round_saeonly_scalar_mask_op4>, %3} ... this.
22771 * config/i386/subst.md (mask_scalar_operand4,
22772 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
22773 round_saeonly_scalar_nimm_predicate): New subst attributes.
22775 2017-07-06 Julia Koval <julia.koval@intel.com>
22777 * config/i386/i386.c (ix86_erase_embedded_rounding):
22778 Remove code for old rounding pattern.
22780 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
22782 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
22784 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
22786 * doc/sourcebuild.texi (Test Directives, Variants of
22787 dg-require-support): Add documentation for dg-require-stack-check.
22789 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
22791 * config/i386/subst.md (mask_scalar, round_scalar,
22792 round_saeonly_scalar): New meta-templates.
22793 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
22794 round_scalar_mask_operand3, round_scalar_mask_op3,
22795 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
22796 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
22797 round_saeonly_scalar_constraint,
22798 round_saeonly_scalar_prefix): New subst attribute.
22799 * config/i386/sse.md
22800 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
22801 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
22802 <round_scalar_name> ... this.
22803 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
22804 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
22805 <round_scalar_name> ... this.
22806 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
22807 <sse>_vm<code><mode>3<mask_scalar_name>
22808 <round_saeonly_scalar_name> ... this.
22809 (v<plusminus_mnemonic><ssescalarmodesuffix>
22810 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22811 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22812 v<plusminus_mnemonic><ssescalarmodesuffix>
22813 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22814 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22815 (v<multdiv_mnemonic><ssescalarmodesuffix>
22816 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22817 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22818 v<multdiv_mnemonic><ssescalarmodesuffix>
22819 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22820 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22821 (v<maxmin_float><ssescalarmodesuffix>
22822 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
22823 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
22824 v<maxmin_float><ssescalarmodesuffix>
22825 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22826 %0<mask_scalar_operand3>, %1, %<iptr>2
22827 <round_saeonly_scalar_mask_op3>} ... this.
22829 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
22831 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
22832 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
22834 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
22835 Alan Hayward <alan.hayward@arm.com>
22836 David Sherwood <david.sherwood@arm.com>
22838 * combine.c (simplify_if_then_else): Remove "enum" before
22840 * compare-elim.c (can_eliminate_compare): Likewise.
22841 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
22843 (aarch64_lookup_simd_builtin_type): Likewise.
22844 (aarch64_simd_builtin_type): Likewise.
22845 (aarch64_init_simd_builtin_types): Likewise.
22846 (aarch64_simd_expand_args): Likewise.
22847 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
22849 (aarch64_reverse_mask): Likewise.
22850 (aarch64_simd_emit_reg_reg_move): Likewise.
22851 (aarch64_gen_adjusted_ldpstp): Likewise.
22852 (aarch64_ccmp_mode_to_code): Likewise.
22853 (aarch64_operands_ok_for_ldpstp): Likewise.
22854 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22855 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
22857 (aarch64_min_divisions_for_recip_mul): Likewise.
22858 (aarch64_reassociation_width): Likewise.
22859 (aarch64_get_condition_code_1): Likewise.
22860 (aarch64_simd_emit_reg_reg_move): Likewise.
22861 (aarch64_simd_attr_length_rglist): Likewise.
22862 (aarch64_reverse_mask): Likewise.
22863 (aarch64_operands_ok_for_ldpstp): Likewise.
22864 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22865 (aarch64_gen_adjusted_ldpstp): Likewise.
22866 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
22868 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
22869 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
22870 (arm_lookup_simd_builtin_type): Likewise.
22871 (arm_simd_builtin_type): Likewise.
22872 (arm_init_simd_builtin_types): Likewise.
22873 (arm_expand_builtin_args): Likewise.
22874 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
22875 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
22876 (ft32_setup_incoming_varargs): Likewise.
22877 (ft32_function_arg): Likewise.
22878 (ft32_function_arg_advance): Likewise.
22879 (ft32_pass_by_reference): Likewise.
22880 (ft32_arg_partial_bytes): Likewise.
22881 (ft32_valid_pointer_mode): Likewise.
22882 (ft32_addr_space_pointer_mode): Likewise.
22883 (ft32_addr_space_legitimate_address_p): Likewise.
22884 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
22886 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
22887 (ix86_emit_outlined_ms2sysv_restore): Likewise.
22888 (iamcu_alignment): Likewise.
22889 (canonicalize_vector_int_perm): Likewise.
22890 (ix86_noce_conversion_profitable_p): Likewise.
22891 (ix86_mpx_bound_mode): Likewise.
22892 (ix86_operands_ok_for_move_multiple): Likewise.
22893 * config/microblaze/microblaze-protos.h
22894 (microblaze_expand_conditional_branch_reg): Likewise.
22895 * config/microblaze/microblaze.c
22896 (microblaze_expand_conditional_branch_reg): Likewise.
22897 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
22899 (rs6000_reassociation_width): Likewise.
22900 (rs6000_invalid_binary_op): Likewise.
22901 (fusion_p9_p): Likewise.
22902 (emit_fusion_p9_load): Likewise.
22903 (emit_fusion_p9_store): Likewise.
22904 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
22906 (riscv_hard_regno_mode_ok_p): Likewise.
22907 (riscv_address_insns): Likewise.
22908 (riscv_split_symbol): Likewise.
22909 (riscv_legitimize_move): Likewise.
22910 (riscv_function_value): Likewise.
22911 (riscv_hard_regno_nregs): Likewise.
22912 (riscv_expand_builtin): Likewise.
22913 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
22914 (riscv_build_integer): Likewise.
22915 (riscv_split_integer): Likewise.
22916 (riscv_legitimate_constant_p): Likewise.
22917 (riscv_cannot_force_const_mem): Likewise.
22918 (riscv_regno_mode_ok_for_base_p): Likewise.
22919 (riscv_valid_base_register_p): Likewise.
22920 (riscv_valid_offset_p): Likewise.
22921 (riscv_valid_lo_sum_p): Likewise.
22922 (riscv_classify_address): Likewise.
22923 (riscv_legitimate_address_p): Likewise.
22924 (riscv_address_insns): Likewise.
22925 (riscv_load_store_insns): Likewise.
22926 (riscv_force_binary): Likewise.
22927 (riscv_split_symbol): Likewise.
22928 (riscv_force_address): Likewise.
22929 (riscv_legitimize_address): Likewise.
22930 (riscv_move_integer): Likewise.
22931 (riscv_legitimize_const_move): Likewise.
22932 (riscv_legitimize_move): Likewise.
22933 (riscv_address_cost): Likewise.
22934 (riscv_subword): Likewise.
22935 (riscv_output_move): Likewise.
22936 (riscv_canonicalize_int_order_test): Likewise.
22937 (riscv_emit_int_order_test): Likewise.
22938 (riscv_function_arg_boundary): Likewise.
22939 (riscv_pass_mode_in_fpr_p): Likewise.
22940 (riscv_pass_fpr_single): Likewise.
22941 (riscv_pass_fpr_pair): Likewise.
22942 (riscv_get_arg_info): Likewise.
22943 (riscv_function_arg): Likewise.
22944 (riscv_function_arg_advance): Likewise.
22945 (riscv_arg_partial_bytes): Likewise.
22946 (riscv_function_value): Likewise.
22947 (riscv_pass_by_reference): Likewise.
22948 (riscv_setup_incoming_varargs): Likewise.
22949 (riscv_print_operand): Likewise.
22950 (riscv_elf_select_rtx_section): Likewise.
22951 (riscv_save_restore_reg): Likewise.
22952 (riscv_for_each_saved_reg): Likewise.
22953 (riscv_register_move_cost): Likewise.
22954 (riscv_hard_regno_mode_ok_p): Likewise.
22955 (riscv_hard_regno_nregs): Likewise.
22956 (riscv_class_max_nregs): Likewise.
22957 (riscv_memory_move_cost): Likewise.
22958 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22959 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22960 (rl78_addr_space_address_mode): Likewise.
22961 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22963 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22964 (rs6000_reassociation_width): Likewise.
22965 (rs6000_invalid_binary_op): Likewise.
22966 (fusion_p9_p): Likewise.
22967 (emit_fusion_p9_load): Likewise.
22968 (emit_fusion_p9_store): Likewise.
22969 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22970 (ok_for_simple_move_operands): Likewise.
22971 (ok_for_simple_move_strict_operands): Likewise.
22972 (ok_for_simple_arith_logic_operands): Likewise.
22973 (visium_legitimize_reload_address): Likewise.
22974 (visium_select_cc_mode): Likewise.
22975 (output_cbranch): Likewise.
22976 (visium_split_double_move): Likewise.
22977 (visium_expand_copysign): Likewise.
22978 (visium_expand_int_cstore): Likewise.
22979 (visium_expand_fp_cstore): Likewise.
22980 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22981 (visium_function_arg): Likewise.
22982 (visium_function_arg_advance): Likewise.
22983 (visium_libcall_value): Likewise.
22984 (visium_setup_incoming_varargs): Likewise.
22985 (visium_legitimate_constant_p): Likewise.
22986 (visium_legitimate_address_p): Likewise.
22987 (visium_legitimize_address): Likewise.
22988 (visium_secondary_reload): Likewise.
22989 (visium_register_move_cost): Likewise.
22990 (visium_memory_move_cost): Likewise.
22991 (prepare_move_operands): Likewise.
22992 (ok_for_simple_move_operands): Likewise.
22993 (ok_for_simple_move_strict_operands): Likewise.
22994 (ok_for_simple_arith_logic_operands): Likewise.
22995 (visium_function_value_1): Likewise.
22996 (rtx_ok_for_offset_p): Likewise.
22997 (visium_legitimize_reload_address): Likewise.
22998 (visium_split_double_move): Likewise.
22999 (visium_expand_copysign): Likewise.
23000 (visium_expand_int_cstore): Likewise.
23001 (visium_expand_fp_cstore): Likewise.
23002 (visium_split_cstore): Likewise.
23003 (visium_select_cc_mode): Likewise.
23004 (visium_split_cbranch): Likewise.
23005 (output_cbranch): Likewise.
23006 (visium_print_operand_address): Likewise.
23007 * expmed.c (flip_storage_order): Likewise.
23008 * expmed.h (emit_cstore): Likewise.
23009 (flip_storage_order): Likewise.
23010 * genrecog.c (validate_pattern): Likewise.
23011 * hsa-gen.c (gen_hsa_addr): Likewise.
23012 * internal-fn.c (expand_arith_overflow): Likewise.
23013 * ira-color.c (allocno_copy_cost_saving): Likewise.
23014 * lra-assigns.c (find_hard_regno_for_1): Likewise.
23015 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
23016 (process_invariant_for_inheritance): Likewise.
23017 * lra-eliminations.c (move_plus_up): Likewise.
23018 * omp-low.c (lower_oacc_reductions): Likewise.
23019 * simplify-rtx.c (simplify_subreg): Likewise.
23020 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
23021 (TARGET_CHKP_BOUND_MODE): Likewise..
23022 * targhooks.c (default_chkp_bound_mode): Likewise.
23023 (default_setup_incoming_vararg_bounds): Likewise.
23024 * targhooks.h (default_chkp_bound_mode): Likewise.
23025 (default_setup_incoming_vararg_bounds): Likewise.
23026 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
23027 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
23028 (have_whole_vector_shift): Likewise.
23029 * tree-vect-stmts.c (vectorizable_load): Likewise.
23030 * doc/tm.texi: Regenerate.
23032 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23034 Graceful degrade if Binutils PR21472 is not available.
23037 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
23038 .rodata in flash test fails.
23039 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
23040 * confgure: Regenerate.
23041 * config.in: Regenerate.
23042 * config/avr/avr.c (avr_asm_named_section)
23043 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
23044 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
23045 (avr_asm_init_sections): Same.
23047 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23049 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
23050 (fma<VH:mode>4_intrinsic): Likewise.
23051 (*fmsub<VCVTF:mode>4): Likewise.
23052 (*fmsub<VH:mode>4_intrinsic): Likewise.
23054 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
23057 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
23058 Don't depend on "optimize > 0".
23059 (out_movhi_r_mr, out_movqi_mr_r): Same.
23060 (out_movhi_mr_r, out_movqi_r_mr): Same.
23061 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
23062 io_address_operand on "optimize > 0".
23064 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23066 * tree-loop-distribution.c: Add general explanantion on the pass.
23067 (generate_loops_for_partition): Mark distributed loop.
23068 (pg_add_dependence_edges): New parameter. Handle alias data
23069 dependence specially and record it in the parameter if asked.
23070 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
23071 (init_partition_graph_vertices, add_partition_graph_edge): New.
23072 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
23073 (free_partition_graph_vdata, build_partition_graph): New.
23074 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
23075 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
23076 (data_ref_segment_size, latch_dominated_by_data_ref): New.
23077 (compute_alias_check_pairs, version_loop_by_alias_check): New.
23078 (version_for_distribution_p, finalize_partitions): New.
23079 (distribute_loop): Handle alias data dependence specially. Factor
23080 out loop fusion code as functions and call these functions.
23082 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23084 * tree-loop-distribution.c (classify_partition): New parameter and
23085 better handle reduction statement.
23086 (rdg_build_partitions): Revise comment.
23087 (distribute_loop): Compute statements in all partitions and pass it
23088 to classify_partition.
23090 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23092 * tree-loop-distribution.c (enum partition_type): New.
23093 (struct partition): New field type.
23094 (partition_merge_into): Add parameter. Update partition type.
23095 (data_dep_in_cycle_p, update_type_for_merge): New functions.
23096 (build_rdg_partition_for_vertex): Compute partition type.
23097 (rdg_build_partitions): Dump partition type.
23098 (distribute_loop): Update calls to partition_merge_into.
23100 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23102 * tree-loop-distribution.c (struct ddr_hasher): New.
23103 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
23105 (classify_partition): Call get_data_dependence.
23106 (pg_add_dependence_edges): Ditto.
23107 (distribute_loop): Release data dependence hash table.
23109 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23111 * tree-loop-distribution.c (ref_base_address): Delete.
23112 (similar_memory_accesses): Rename ...
23113 (share_memory_accesses): ... to this. Check if partitions access
23114 the same memory reference.
23115 (distribute_loop): Call share_memory_accesses.
23117 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23119 * tree-loop-distribution.c (struct partition): New field recording
23120 its data reference.
23121 (partition_alloc, partition_free): Init and release data refs.
23122 (partition_merge_into): Merge data refs.
23123 (build_rdg_partition_for_vertex): Collect data refs for partition.
23124 (pg_add_dependence_edges): Change parameters from vector to bitmap.
23126 (distribute_loop): Remve data refs from vertice data of partition
23129 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23131 * tree-loop-distribution.c (params.h): Include header file.
23132 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
23133 (datarefs_vec): New global var.
23134 (create_rdg_vertices): Use datarefs_vec directly.
23135 (free_rdg): Don't free data references.
23136 (build_rdg): Update use. Don't free data references.
23137 (distribute_loop): Compute global variable for data references.
23138 Bail out if there are too many data references.
23140 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23142 * tree-loop-distribution.c (loop_nest): New global var.
23143 (build_rdg): Use loop directly, rather than loop nest.
23144 (pg_add_dependence_edges): Remove loop nest parameter. Use global
23146 (distribute_loop): Compute global variable loop nest. Update use.
23148 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23150 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
23151 (partition_merge_into): New parameter. Dump reason for fusion.
23152 (distribute_loop): Update use of partition_merge_into.
23154 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23156 * tree-loop-distribution.c (bb_top_order_index): New.
23157 (bb_top_order_index_size, bb_top_order_cmp): New.
23158 (stmts_from_loop): Use topological order.
23159 (pass_loop_distribution::execute): Compute and release topological
23160 order for basic blocks.
23162 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23164 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
23167 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23169 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
23170 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
23171 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
23172 * internal-fn.def (LOOP_DIST_ALIAS): New.
23173 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
23174 (fold_loop_internal_call): ... this.
23175 (vect_loop_dist_alias_call): New function.
23176 (set_uid_loop_bbs): Call fold_loop_internal_call.
23177 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
23180 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23183 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
23184 Require dead FLAGS_REG at the beginning of a peephole.
23186 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23189 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
23190 arguments in the call to __builtin_ia32_sbb_u32.
23191 (_subborrow_u64): Swap _X and _Y arguments in the call to
23192 __builtin_ia32_sbb_u64.
23194 2017-07-04 Jakub Jelinek <jakub@redhat.com>
23197 * tree-vrp.c (compare_assert_loc): Turn into a function template
23198 with stable template parameter. Only test if a->e is NULL,
23199 !a->e == !b->e has been verified already. Use e == NULL or
23200 e != NULL instead of e or ! e tests. If stable is true, don't use
23201 iterative_hash_expr, on the other side allow a or b or both NULL
23202 and sort the NULLs last.
23203 (process_assert_insertions): Sort using compare_assert_loc<false>
23204 instead of compare_assert_loc, later sort using
23205 compare_assert_loc<true> before calling process_assert_insertions_for
23206 in a loop. Use break instead of continue once seen NULL pointer.
23208 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23210 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23211 Cortex-R7 and Cortex-R8 processors.
23213 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23215 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
23216 uninitialized while src is not.
23218 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
23220 * common/config/arm/arm-common.c: Adjust include path for
23222 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
23223 (arm-cpu.h): Create in build directory. Adjust dependency rules.
23224 (arm-cpu-data.h): Likewise.
23225 (arm-cpu-cdata.h): Likewise.
23226 * config/arm/arm-cpu.h: Delete.
23227 * config/arm/arm-cpu-cdata.h: Delete.
23228 * config/arm/arm-cpu-data.h: Delete.
23230 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
23232 * config/arm/arm-cpus.in (cortex-a55): New.
23233 (cortex-a75): Likewise.
23234 (cortex-a75.cortex-a55): Likewise.
23235 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
23237 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
23238 * config/arm/arm-cpu-cdata.h: Regenerate.
23239 * config/arm/arm-cpu-data.h: Regenerate.
23240 * config/arm/arm-cpu.h: Regenerate.
23241 * config/arm/arm-tables.opt: Regenerate.
23242 * config/arm/arm-tune.md: Regenerate.
23244 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23246 * haifa-sched.c (sched_create_recovery_edges): Update profile.
23248 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23250 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
23253 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
23255 PR tree-optimization/81292
23256 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
23257 full_string_p, also call adjust_related_strinfos if the adjustment
23258 is simple, otherwise invalidate related strinfos.
23260 2017-07-04 Martin Liska <mliska@suse.cz>
23263 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
23264 newly created variable as DECL_IGNORED_P.
23266 2017-07-04 Martin Liska <mliska@suse.cz>
23269 * ipa-inline.c (inline_small_functions):
23270 Use xstrdup_for_dump.
23272 2017-07-04 Tom de Vries <tom@codesourcery.com>
23274 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
23276 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23279 * config/darwin.c (darwin_function_switched_text_sections):
23282 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
23284 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
23286 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23288 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
23290 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23292 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
23293 min_profitable_iters, and th as inclusive lower bounds.
23294 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
23295 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
23296 for min_profitable_iters and min_profitable_estimate.
23297 (vect_transform_loop): Treat th as an inclusive lower bound.
23298 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23300 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23303 * config/darwin.c (darwin_function_switched_text_sections):
23304 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
23305 in two pieces, and suppress the use of buf.
23307 2017-07-03 Nathan Sidwell <nathan@acm.org>
23309 * hash-table.h (hash_table_mod1): Fix indentation.
23311 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23313 PR middle-end/81290
23314 * predict.c (force_edge_cold): Be more careful about propagation
23316 * profile-count.h (profile_probability::guessed,
23317 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
23319 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
23321 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
23323 * doc/invoke.texi (rcpc architecture extension): Document it.
23325 2017-07-03 Richard Biener <rguenther@suse.de>
23327 PR tree-optimization/60510
23328 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
23329 the scalar reduction PHI and use it.
23330 (vectorizable_reduction): Properly guard the single_defuse_cycle
23331 path for non-SLP reduction chains where we cannot use it.
23332 Rework reduc_def/index and vector type deduction. Rework
23333 vector operand gathering during reduction op code-gen.
23334 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
23335 chains dissolve the chain and leave it to non-SLP reduction
23338 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23340 * tree-data-ref.h (dr_alignment): Declare.
23341 * tree-data-ref.c (dr_alignment): New function.
23342 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
23343 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23345 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
23347 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23349 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
23350 and base_misalignment fields.
23351 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
23352 * tree-data-ref.c: Include builtins.h.
23353 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
23354 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
23355 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
23356 * tree-vect-data-refs.c: Include tree-cfg.h.
23357 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
23358 fields instead of calculating an alignment here.
23359 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
23360 innermost_loop_behavior fields.
23362 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23364 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
23366 (DR_STEP_ALIGNMENT): New macro.
23367 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
23368 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
23369 (create_data_ref): Print it.
23370 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
23371 to tell whether the step preserves vector (mis)alignment.
23372 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23373 Move the check for an integer step and generalise to all INTEGER_CST.
23374 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
23375 Print the outer step alignment.
23377 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23379 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
23380 with offset_alignment.
23381 (DR_ALIGNED_TO): Delete.
23382 (DR_OFFSET_ALIGNMENT): New macro.
23383 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
23384 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
23385 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
23386 (create_data_ref): Likewise.
23387 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23388 (vect_analyze_data_refs): Likewise.
23389 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
23390 creating dummy innermost behavior.
23392 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23394 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
23395 with a "innermost_loop_behavior *" and refeence tree.
23396 * tree-data-ref.c (dr_analyze_innermost): Likewise.
23397 (create_data_ref): Update call accordingly.
23398 * tree-predcom.c (find_looparound_phi): Likewise.
23400 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23402 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
23403 fields with dr_wrt_vec_loop.
23404 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
23405 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
23406 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
23407 (vect_dr_behavior): New function.
23408 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23409 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
23410 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
23411 track whether the step preserves the misalignment.
23412 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23413 Use vect_dr_behavior.
23414 (vect_setup_realignment): Update call accordingly.
23415 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
23416 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
23417 call to vect_create_addr_base_for_vector_ref.
23418 (vect_create_cond_for_align_checks): Likewise.
23419 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
23420 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
23421 (vect_recog_mask_conversion_pattern): Likewise.
23422 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
23423 (new_stmt_vec_info): Remove redundant zeroing.
23425 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23427 * common/config/arm/arm-common.c (arm_be8_option): New function.
23428 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
23429 (ISA_ARMv6): Add isa_bit_be8.
23430 * config/arm/arm.h (arm_be8_option): Add prototype.
23431 (BE8_SPEC_FUNCTION): New define.
23432 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
23433 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
23434 (mlittle-endian): Similarly.
23435 (mbe8, mbe32): New options.
23436 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
23437 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
23439 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23441 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
23443 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23445 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
23446 (cleanup_tree_cfg_bb): Use it.
23447 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
23449 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
23451 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23454 * loop-doloop.c (add_test): Update profile.
23456 2017-07-03 Martin Liska <mliska@suse.cz>
23459 * sanopt.c (rewrite_usage_of_param): New function.
23460 (sanitize_rewrite_addressable_params): Likewise.
23461 (pass_sanopt::execute): Call rewrite_usage_of_param.
23463 2017-07-03 Richard Biener <rguenther@suse.de>
23465 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
23466 back to using VIEW_CONVERT_EXPR.
23468 2017-07-03 Martin Liska <mliska@suse.cz>
23471 * doc/extend.texi: Document when a resolver function is
23472 generated for target_clones.
23474 2017-07-03 Martin Liska <mliska@suse.cz>
23476 * asan.c (asan_emit_stack_protection): Unpoison just red zones
23477 and shadow memory of auto variables which are subject of
23478 use-after-scope sanitization.
23479 (asan_expand_mark_ifn): Add do set only when is_poison.
23481 2016-07-03 Richard Biener <rguenther@suse.de>
23483 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
23485 (vect_force_simple_reduction): Record reduction def -> phi mapping.
23486 (vectorizable_reduction): Perform reduction PHI creation when
23487 visiting a reduction PHI and adjust and simplify code generation
23488 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
23489 (vect_transform_loop): Visit reduction PHIs.
23490 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
23491 defs into the SLP tree.
23492 (vect_build_slp_tree): Reduction defs terminate the recursion.
23493 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
23495 (vect_get_vec_defs_for_stmt_copy): Export.
23496 (vect_get_vec_defs): Likewise.
23497 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
23499 (vect_get_vec_defs_for_stmt_copy): Declare.
23500 (vect_get_vec_defs): Likewise.
23502 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23504 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
23505 parameter with a "loop" parameter and use it instead of the
23506 loop containing DR_STMT. Don't check simple_iv when doing
23507 BB analysis. Describe the two analysis modes in the comment.
23509 2017-07-03 Tom de Vries <tom@codesourcery.com>
23511 PR tree-optimization/69468
23512 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
23513 (find_same_succ_bb): Handle ignore_edge_flags.
23515 2017-07-03 Tom de Vries <tom@codesourcery.com>
23517 PR tree-optimization/81192
23518 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
23520 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
23522 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
23524 2017-07-03 Tom de Vries <tom@codesourcery.com>
23526 PR tree-optimization/81192
23527 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
23528 BB_SAME_SUCC (bb) == NULL.
23530 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23532 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
23535 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23537 * dumpfile.c: Include profile-count.h
23538 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
23540 (insert_cond_bb): Update profile.
23541 * tree-cfg.h (insert_cond_bb): Update prototype.
23542 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
23543 * tree-dump.c: Do not include tree-cfg.
23545 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23547 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
23549 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23551 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
23554 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23556 * tree-complex.c (expand_complex_div_wide): update profile.
23558 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23559 Alan Hayward <alan.hayward@arm.com>
23560 David Sherwood <david.sherwood@arm.com>
23562 * Makefile.in (MACHMODE_H): Remove insn-modes.h
23563 (CORETYPES_H): New define.
23564 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
23565 (insn-modes-inline.h, s-modes-inline-h): New rules.
23566 (generated_files): Add insn-modes-inline.h.
23567 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
23568 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
23569 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
23570 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
23571 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
23572 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
23573 (build/gencodes.o, build/genconditions.o): Likewise.
23574 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
23575 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
23576 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
23577 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
23578 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
23579 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
23580 * coretypes.h: Include everything up to real.h for generators.
23581 Include insn-modes.h first. Include wide-int-print.h after
23582 wide-int.h. Include insn-modes-inline.h and then machmode.h.
23583 * machmode.h: Don't include insn-modes.h here.
23584 * function-tests.c: Remove includes of signop.h, machmode.h,
23585 double-int.h and wide-int.h.
23587 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
23589 * optc-save-gen.awk: Likewise.
23590 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
23591 * godump.c: Remove include of wide-int-print.h.
23592 * pretty-print.h: Likewise.
23593 * wide-int-print.cc: Likewise.
23594 * wide-int.cc: Likewise.
23595 * hash-map-tests.c: Remove include of signop.h.
23596 * hash-set-tests.c: Likewise.
23597 * rtl-tests.c: Likewise.
23598 * mkconfig.sh: Remove include of machmode.h.
23599 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
23601 (emit_insn_modes_inline_h): ...this new function. Emit the code
23602 into an insn-modes-inline.h header file, adding appropriate
23603 include guards and end comments.
23604 (emit_insn_modes_c_header): Remove include of machmode.h.
23605 (emit_min_insn_modes_c_header): Include coretypes.h rather than
23607 (main): Handle -i flag and call emit_insn_modes_inline_h when
23610 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23612 * tree-ssa-strlen.c (strinfo): Rename the length field to
23613 nonzero_chars. Add a full_string_p field.
23614 (compare_nonzero_chars, zero_length_string_p): New functions.
23615 (get_addr_stridx): Add an offset_out parameter.
23616 Use compare_nonzero_chars.
23617 (get_stridx): Update accordingly. Use compare_nonzero_chars.
23618 (new_strinfo): Update after above changes to strinfo.
23619 (set_endptr_and_length): Set full_string_p.
23620 (get_string_length): Update after above changes to strinfo.
23621 (unshare_strinfo): Update call to new_strinfo.
23622 (maybe_invalidate): Likewise.
23623 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
23624 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
23625 as a uhwi instead of an shwi. Update after above changes to
23626 strinfo and new_strinfo.
23627 (zero_length_string): Assert that chainsi contains full strings.
23628 Use zero_length_string_p. Update call to new_strinfo.
23629 (adjust_related_strinfos): Update after above changes to strinfo.
23630 Copy full_string_p from origsi.
23631 (adjust_last_stmt): Use zero_length_string_p.
23632 (handle_builtin_strlen): Update after above changes to strinfo and
23633 new_strinfo. Install the lhs as the string length if the previous
23634 entry didn't describe a full string.
23635 (handle_builtin_strchr): Update after above changes to strinfo
23637 (handle_builtin_strcpy): Likewise.
23638 (handle_builtin_strcat): Likewise.
23639 (handle_builtin_malloc): Likewise.
23640 (handle_pointer_plus): Likewise.
23641 (handle_builtin_memcpy): Likewise. Track nonzero characters
23642 that aren't necessarily followed by a nul terminator.
23643 (handle_char_store): Likewise.
23645 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23647 PR tree-optimization/80769
23648 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
23649 for malloc and calloc. Document the new invariant that all related
23650 strinfos have delayed lengths or none do.
23651 (verify_related_strinfos): Move earlier in file.
23652 (set_endptr_and_length): New function, split out from...
23653 (get_string_length): ...here. Also set the lengths of related
23655 (zero_length_string): Assert that chainsi has known (rather than
23657 (adjust_related_strinfos): Likewise.
23659 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23661 PR tree-optimization/81136
23662 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
23663 assert that two references with the same misalignment have the same
23664 compile-time misalignment if those compile-time misalignments
23667 2017-07-01 Andi Kleen <ak@linux.intel.com>
23669 * print-tree.c (print_node): Print all attributes.
23671 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23673 * cfg.c (scale_bbs_frequencies): New function.
23674 * cfg.h (scale_bbs_frequencies): Declare it.
23675 * cfgloopanal.c (single_likely_exit): Cleanup.
23676 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
23678 (scale_loop_profile): Likewise.
23679 (loop_version): Likewise.
23680 (create_empty_loop_on_edge): Update.
23681 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
23682 scale_loop_frequencies, scale_loop_profile, loopify,
23683 loop_version): Update prototypes.
23684 * modulo-sched.c (sms_schedule): Update.
23685 * predict.c (unlikely_executed_edge_p): Also check probability.
23686 (probably_never_executed_edge_p): Fix typo.
23687 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23688 * tree-parloops.c (gen_parallel_loop): Update.
23689 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
23690 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23691 * tree-ssa-loop-split.c (split_loop): Update.
23692 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23693 * tree-vect-loop-manip.c (vect_do_peeling): Update.
23694 (vect_loop_versioning): Update.
23695 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23697 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23699 * trans-mem.c (split_bb_make_tm_edge): Update profile.
23701 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23703 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
23704 to keep profile consistent.
23706 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23708 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
23709 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
23710 * profile-count.h (max_safe_multiplier): Make unsigned.
23711 (profile_count::guessed_zero): New.
23713 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23715 * bb-reorder.c (fix_up_crossing_landing_pad,
23716 fix_crossing_conditional_branches): Use make_single_succ_edge
23717 to keep profile consistent.
23719 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23721 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
23724 2017-07-01 Jakub Jelinek <jakub@redhat.com>
23727 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
23728 the right scopes, make sure cond_jump isn't preserved between multiple
23729 iterations. Search for fallthru edge whenever there are 3+ edges and
23730 use find_fallthru_edge for it.
23732 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23734 Patch by Alexander Monakov <amonakov@ispras.ru>
23735 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
23736 probabilities consistently.
23738 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23740 * pa.c (pa_expand_compare_and_swap_loop): Update call of
23741 emit_cmp_and_jump_insns.
23743 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23746 * tree-inline.c (expand_call_inline): Combine profile statuses.
23748 2017-06-30 Andrew Pinski <apinski@cavium.com>
23750 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
23751 fold_stmt returned true.
23753 2017-06-30 Nathan Sidwell <nathan@acm.org>
23755 * ggc.h (empty_string): Delete.
23756 * cfgexpand.c (expand_asm_stmt): Use plain "".
23757 * optabs.c (expand_asm_memory_barrier): Likewise.
23758 * stringpool.c (empty_string): Delete.
23759 (digit_vector, digit_string): Delete.
23760 (ggc_alloc_string): Use plain "", don't optimize single digit
23761 strings. Use ggc_alloc_atomic.
23763 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
23765 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
23766 comparison set and one other set, use the cost of the non-comparison
23769 2017-06-30 Nathan Sidwell <nathan@acm.org>
23771 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
23774 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
23776 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
23777 loops. Remove now unneeded calls to gimple_switch_set_label() that
23778 just set removed labels to NULL_TREE.
23780 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
23782 * tree-ssanames.c (set_range_info_raw): Abstract from ...
23783 (set_range_info): ...here. Only call set_range_info_raw if domain
23785 (set_nonzero_bits): Call set_range_info_raw.
23786 * tree-ssanames.h (set_range_info_raw): New.
23788 2017-06-30 Jakub Jelinek <jakub@redhat.com>
23791 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
23792 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
23793 of nonimmediate_operand and <store_mask_constraint> instead of m
23794 for the input operand. For V8FI iterator, always split if input
23795 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
23796 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
23797 <store_mask_predicate> instead of register_operand and
23798 <store_mask_constraint> instead of v for the input operand. Make
23799 sure both operands aren't MEMs for if not <mask_applied>.
23801 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
23803 * lto-wrapper.c (copy_file) Close both file descriptors before
23806 2017-06-30 Martin Liska <mliska@suse.cz>
23809 * multiple_target.c (create_dispatcher_calls): Make ifunc
23810 also for function that don't have calls or are not referenced.
23812 2017-06-30 Richard Biener <rguenther@suse.de>
23814 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
23815 analyze the first scalar stmt. Move vector type computation
23816 for the BB case here from ...
23817 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
23818 live operation processing in the SLP case properly.
23820 2017-06-30 Richard Biener <rguenther@suse.de>
23822 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
23824 2017-06-30 Martin Liska <mliska@suse.cz>
23827 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
23828 before BUILT_IN_UNWIND_RESUME when ASAN is used.
23830 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
23832 * doc/invoke.texi (AArch64): Add missing options and remove redundant
23835 2017-06-30 Richard Biener <rguenther@suse.de>
23837 PR tree-optimization/81249
23838 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
23839 condition reduction result to original scalar type.
23841 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23843 * profile-count.h (enum profile_quality): Fix typos and whitespace
23846 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23848 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
23849 type for branch probabilities.
23851 2017-06-29 Julian Brown <julian@codesourcery.com>
23852 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23854 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
23855 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
23856 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
23857 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
23859 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23861 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
23862 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
23863 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
23864 CC usage from generic code to here.
23865 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
23866 CC usage into the target macros.
23868 2017-06-29 Maya Rashish <coypu@sdf.org>
23870 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
23873 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23875 * arm/arm-builtins.c: Include profile-count.h
23876 * except.c (sjlj_emit_function_enter): Use
23877 profile_probability::unlikely.
23879 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23881 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
23882 and tocrel_offset be pointer args rather than implicitly using
23884 (legitimate_constant_pool_address_p, rs6000_emit_move,
23885 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
23886 tocrel_offset and use in toc_relative_expr_p call.
23887 (print_operand, print_operand_address): Use static tocrel_base_oac
23888 and tocrel_offset_oac.
23889 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
23892 2017-06-29 Maya Rashish <coypu@sdf.org>
23894 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
23896 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
23898 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
23899 objects, take into account only the alignment of 'op0' and 'mode1' if
23902 2017-06-29 Steve Ellcey <sellcey@cavium.com>
23904 * ccmp.c (ccmp_tree_comparison_p): New function.
23905 (ccmp_candidate_p): Update to use above function.
23906 (get_compare_parts): New function.
23907 (expand_ccmp_next): Update to use new functions.
23908 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
23910 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
23911 take mode as argument.
23912 * ccmp.h (expand_ccmp_expr): Add mode as argument.
23913 * expr.c (expand_expr_real_1): Pass mode as argument.
23915 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
23917 * combine.c (combine_instructions): Print insns to dump_file, together
23920 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23922 * asan.c (asan_emit_stack_protection): Update.
23923 (create_cond_insert_point): Update.
23924 * auto-profile.c (afdo_propagate_circuit): Update.
23925 * basic-block.h (struct edge_def): Turn probability to
23926 profile_probability.
23927 (EDGE_FREQUENCY): Update.
23928 * bb-reorder.c (find_traces_1_round): Update.
23929 (better_edge_p): Update.
23930 (sanitize_hot_paths): Update.
23931 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23932 (make_single_succ_edge): Update.
23933 (check_bb_profile): Update.
23934 (dump_edge_info): Update.
23935 (update_bb_profile_for_threading): Update.
23936 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23937 probabilitycount to 0.
23938 * cfgbuild.c (compute_outgoing_frequencies): Update.
23939 * cfgcleanup.c (try_forward_edges): Update.
23940 (outgoing_edges_match): Update.
23941 (try_crossjump_to_edge): Update.
23942 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23943 (expand_gimple_tailcall): Update.
23944 (construct_init_block): Use make_single_succ_edge.
23945 (construct_exit_block): Use make_single_succ_edge.
23946 * cfghooks.c (verify_flow_info): Update.
23947 (redirect_edge_succ_nodup): Update.
23948 (split_edge): Update.
23949 (account_profile_record): Update.
23950 * cfgloopanal.c (single_likely_exit): Update.
23951 * cfgloopmanip.c (scale_loop_profile): Update.
23952 (set_zero_probability): Remove.
23953 (duplicate_loop_to_header_edge): Update.
23954 * cfgloopmanip.h (loop_version): Update prototype.
23955 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23956 (force_nonfallthru_and_redirect): Update.
23957 (update_br_prob_note): Update.
23958 (rtl_verify_edges): Update.
23959 (purge_dead_edges): Update.
23960 (rtl_lv_add_condition_to_bb): Update.
23961 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23962 * cgraphunit.c (init_lowered_empty_function): Update.
23963 (cgraph_node::expand_thunk): Update.
23964 * cilk-common.c: Include profile-count.h
23965 * dojump.c (inv): Remove.
23966 (jumpifnot): Update.
23967 (jumpifnot_1): Update.
23968 (do_jump_1): Update.
23970 (do_jump_by_parts_greater_rtx): Update.
23971 (do_compare_rtx_and_jump): Update.
23972 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23973 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23974 * dwarf2cfi.c: Include profile-count.h
23975 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23976 (sjlj_emit_dispatch_table): Likewise.
23977 * explow.c: Include profile-count.h
23978 * expmed.c (emit_store_flag_force): Update.
23979 (do_cmp_and_jump): Update.
23980 * expr.c (compare_by_pieces_d::generate): Update.
23981 (compare_by_pieces_d::finish_mode): Update.
23982 (emit_block_move_via_loop): Update.
23983 (store_expr_with_bounds): Update.
23984 (store_constructor): Update.
23985 (expand_expr_real_2): Update.
23986 (expand_expr_real_1): Update.
23987 * expr.h (try_casesi, try_tablejump): Update prototypes.
23988 * gimple-pretty-print.c (dump_probability): Update.
23989 (dump_profile): New.
23990 (dump_gimple_label): Update.
23991 (dump_gimple_bb_header): Update.
23992 * graph.c (draw_cfg_node_succ_edges): Update.
23993 * hsa-gen.c (convert_switch_statements): Update.
23994 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23995 (find_if_case_1): Update.
23996 (find_if_case_2): Update.
23997 * internal-fn.c (expand_arith_overflow_result_store): Update.
23998 (expand_addsub_overflow): Update.
23999 (expand_neg_overflow): Update.
24000 (expand_mul_overflow): Update.
24001 (expand_vector_ubsan_overflow): Update.
24002 * ipa-cp.c (good_cloning_opportunity_p): Update.
24003 * ipa-split.c (split_function): Use make_single_succ_edge.
24004 * ipa-utils.c (ipa_merge_profiles): Update.
24005 * loop-doloop.c (add_test): Update.
24006 (doloop_modify): Update.
24007 * loop-unroll.c (compare_and_jump_seq): Update.
24008 (unroll_loop_runtime_iterations): Update.
24009 * lra-constraints.c (lra_inheritance): Update.
24010 * lto-streamer-in.c (input_cfg): Update.
24011 * lto-streamer-out.c (output_cfg): Update.
24012 * mcf.c (adjust_cfg_counts): Update.
24013 * modulo-sched.c (sms_schedule): Update.
24014 * omp-expand.c (expand_omp_for_init_counts): Update.
24015 (extract_omp_for_update_vars): Update.
24016 (expand_omp_ordered_sink): Update.
24017 (expand_omp_for_ordered_loops): Update.
24018 (expand_omp_for_generic): Update.
24019 (expand_omp_for_static_nochunk): Update.
24020 (expand_omp_for_static_chunk): Update.
24021 (expand_cilk_for): Update.
24022 (expand_omp_simd): Update.
24023 (expand_omp_taskloop_for_outer): Update.
24024 (expand_omp_taskloop_for_inner): Update.
24025 * omp-simd-clone.c (simd_clone_adjust): Update.
24026 * optabs.c (expand_doubleword_shift): Update.
24027 (expand_abs): Update.
24028 (emit_cmp_and_jump_insn_1): Update.
24029 (expand_compare_and_swap_loop): Update.
24030 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
24031 * predict.c (predictable_edge_p): Update.
24032 (edge_probability_reliable_p): Update.
24033 (set_even_probabilities): Update.
24034 (combine_predictions_for_insn): Update.
24035 (combine_predictions_for_bb): Update.
24036 (propagate_freq): Update.
24037 (estimate_bb_frequencies): Update.
24038 (force_edge_cold): Update.
24039 * profile-count.c (profile_count::dump): Add missing space into dump.
24040 (profile_count::debug): Add newline.
24041 (profile_count::differs_from_p): Explicitly convert to unsigned.
24042 (profile_count::stream_in): Update.
24043 (profile_probability::dump): New member function.
24044 (profile_probability::debug): New member function.
24045 (profile_probability::differs_from_p): New member function.
24046 (profile_probability::differs_lot_from_p): New member function.
24047 (profile_probability::stream_in): New member function.
24048 (profile_probability::stream_out): New member function.
24049 * profile-count.h (profile_count_quality): Rename to ...
24050 (profile_quality): ... this one.
24051 (profile_probability): New.
24052 (profile_count): Update.
24053 * profile.c (compute_branch_probabilities): Update.
24054 * recog.c (peep2_attempt): Update.
24055 * sched-ebb.c (schedule_ebbs): Update.
24056 * sched-rgn.c (find_single_block_region): Update.
24057 (compute_dom_prob_ps): Update.
24058 (schedule_region): Update.
24059 * sel-sched-ir.c (compute_succs_info): Update.
24060 * stmt.c (struct case_node): Update.
24061 (do_jump_if_equal): Update.
24062 (get_outgoing_edge_probs): Update.
24063 (conditional_probability): Update.
24064 (emit_case_dispatch_table): Update.
24065 (expand_case): Update.
24066 (expand_sjlj_dispatch_table): Update.
24067 (emit_case_nodes): Update.
24068 * targhooks.c: Update.
24069 * tracer.c (better_p): Update.
24070 (find_best_successor): Update.
24071 * trans-mem.c (expand_transaction): Update.
24072 * tree-call-cdce.c: Update.
24073 * tree-cfg.c (gimple_split_edge): Upate.
24074 (move_sese_region_to_fn): Upate.
24075 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
24076 * tree-eh.c (lower_resx): Upate.
24077 (cleanup_empty_eh_move_lp): Upate.
24078 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24079 * tree-inline.c (copy_edges_for_bb): Update.
24080 (copy_cfg_body): Update.
24081 * tree-parloops.c (gen_parallel_loop): Update.
24082 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
24083 (gimple_gen_time_profiler): Update.
24084 * tree-ssa-dce.c (remove_dead_stmt): Update.
24085 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24086 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
24087 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
24088 (unloop_loops): Update.
24089 (try_peel_loop): Update.
24090 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24091 * tree-ssa-loop-split.c (connect_loops): Update.
24092 (split_loop): Update.
24093 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24094 (hoist_guard): Update.
24095 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
24096 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
24097 (value_replacement): Update.
24098 * tree-ssa-reassoc.c (branch_fixup): Update.
24099 * tree-ssa-tail-merge.c (replace_block_by): Update.
24100 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
24101 (create_edge_and_update_destination_phis): Update.
24102 (compute_path_counts): Update.
24103 (recompute_probabilities): Update.
24104 (update_joiner_offpath_counts): Update.
24105 (freqs_to_counts_path): Update.
24106 (duplicate_thread_path): Update.
24107 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
24108 (struct switch_conv_info): Update.
24109 (gen_inbound_check): Update.
24110 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24111 (vect_do_peeling): Update.
24112 (vect_loop_versioning): Update.
24113 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24114 (optimize_mask_stores): Update.
24115 * ubsan.c (ubsan_expand_null_ifn): Update.
24116 * value-prof.c (gimple_divmod_fixed_value): Update.
24117 (gimple_divmod_fixed_value_transform): Update.
24118 (gimple_mod_pow2): Update.
24119 (gimple_mod_pow2_value_transform): Update.
24120 (gimple_mod_subtract): Update.
24121 (gimple_mod_subtract_transform): Update.
24122 (gimple_ic): Update.
24123 (gimple_stringop_fixed_value): Update.
24124 (gimple_stringops_transform): Update.
24125 * value-prof.h: Update.
24127 2017-06-29 Carl Love <cel@us.ibm.com>
24129 * config/rs6000/rs6000-c.c: Add support for built-in functions
24130 vector signed int vec_signed (vector float);
24131 vector signed long long vec_signed (vector double);
24132 vector signed int vec_signed2 (vector double, vector double);
24133 vector signed int vec_signede (vector double);
24134 vector signed int vec_signedo (vector double);
24135 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
24136 instruction generator.
24137 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24138 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
24139 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
24141 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
24142 vunsignede_v2df): Add define_expands.
24143 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
24144 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
24145 VEC_UNSIGNEDO): Add definitions.
24146 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24147 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
24148 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
24149 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
24150 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
24151 * config/rs6000/altivec.h (vec_signed, vec_signed2,
24152 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
24153 vec_unsignede, vec_unsignedo): Add builtin defines.
24154 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
24156 * doc/extend.texi: Update the built-in documentation file for the
24157 new built-in functions.
24159 2017-06-29 Richard Biener <rguenther@suse.de>
24161 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
24162 reduction chains to LOOP_VINFO_REDUCTIONS.
24163 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
24164 SLP reductions after processing reduction chains.
24166 2017-06-29 Nathan Sidwell <nathan@acm.org>
24168 * builtins.c (fold_builtin_FUNCTION): Use
24169 lang_hooks.decl_printable_name.
24171 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
24173 PR middle-end/81194
24174 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
24175 with only one label.
24176 * stmt.c (expand_case): Assert NCASES is greater than one.
24178 2017-06-29 Richard Biener <rguenther@suse.de>
24180 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
24182 (group_case_labels): Likewise.
24183 (find_taken_edge): Push sanity checking on val to workers...
24184 (find_taken_edge_cond_expr): ... here
24185 (find_taken_edge_switch_expr): ... and here, handle cases
24186 with just a default label.
24187 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
24188 (group_case_labels): Likewise.
24189 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
24190 group_case_labels does anything cleanup the CFG again.
24192 2017-06-29 Bin Cheng <bin.cheng@arm.com>
24194 PR tree-optimization/81196
24195 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
24196 exit condition comparing two IVs.
24198 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
24200 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
24201 profile to the dummy entry at the end of the list of architectures.
24202 * config/arm/arm-cpu-cdata.h: Regenerated.
24204 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24205 Michael Collison <michael.collison@arm.com>
24208 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
24210 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
24211 (*aarch64_reg_<mode>3_minus_mask): New pattern.
24212 (*aarch64_<optab>_reg_di3_mask2): New pattern.
24213 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
24214 of shift when the shift amount is masked with constant equal to
24215 the size of the mode.
24216 * config/aarch64/predicates.md (subreg_lowpart_operator): New
24219 2017-06-29 Martin Liska <mliska@suse.cz>
24221 * config/i386/i386.opt: Change range from [1,5] to [0,5].
24223 2017-06-29 Yury Gribov <tetra2005@gmail.com>
24226 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
24229 (edge_growth_cache_entry::edge_growth_cache_entry): New
24231 (reset_edge_growth_cache): Update to use constructor.
24233 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24235 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
24236 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24237 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24239 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
24241 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
24242 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
24244 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
24246 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
24247 (*-linux-uclibc*): Add t-uclibc tmake_file.
24248 * config/t-musl: New.
24249 * config/t-uclibc: New.
24251 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
24253 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
24255 (gen_comm_data): Emit architectural setting of arch_prof.
24256 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
24258 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
24259 (armv8-m.base, armv8-m.main): Likewise.
24260 * arm-protos.h (arm_build_target): Add profile field.
24261 (arch_option): Likewise.
24262 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
24264 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
24265 arm_active_target.profile.
24267 2017-06-28 Richard Biener <rguenther@suse.de>
24269 PR middle-end/81227
24270 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
24271 TYPE_OVERFLOW_WRAPS.
24272 * match.pd (negate_expr_p): Likewise.
24273 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
24274 fold_build2, not fold_binary.
24276 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24278 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
24279 Convert memory address to Pmode.
24280 (aarch64_print_operand): Assert MEM operands are always Pmode.
24282 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24285 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
24286 Remove redundant if.
24287 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
24288 * config/arm/aarch-common-protos.h
24289 (aarch_forward_to_shift_is_not_shifted_re): Remove.
24290 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
24292 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
24295 * multiple_target.c (create_dispatcher_calls): Set the default
24296 clone to be static, not public.
24298 2017-06-28 Richard Biener <rguenther@suse.de>
24300 * tree-vect-loop.c (vectorizable_reduction): Move special
24301 cond reduction IV var creation ...
24302 (vect_create_epilog_for_reduction): ... here. Remove induction_index
24303 parameter. Use STMT_VINFO_VECTYPE.
24304 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
24307 2017-06-28 Martin Liska <mliska@suse.cz>
24310 * ipa-visibility.c (non_local_p): Handle visibility.
24312 2017-06-28 Martin Liska <mliska@suse.cz>
24315 * common.opt: Add IntegerRange to various options.
24316 * opt-functions.awk (integer_range_info): New function.
24317 * optc-gen.awk: Add integer_range_info to cl_options struct.
24318 * opts-common.c (decode_cmdline_option): Handle
24319 CL_ERR_INT_RANGE_ARG.
24320 (cmdline_handle_error): Likewise.
24321 * opts.c (print_filtered_help): Show valid interval in
24322 when --help is provided.
24323 * opts.h (struct cl_option): Add range_min and range_max fields.
24324 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
24326 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
24328 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
24329 (x * C EQ/NE y * C): New transformation.
24331 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
24333 * genmultilib (combination_space): Accept '+' in option names.
24335 2017-06-28 Martin Liska <mliska@suse.cz>
24338 * asan.c (instrument_derefs): Bail out inner references
24339 that are hard register variables.
24341 2017-06-28 Jakub Jelinek <jakub@redhat.com>
24344 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
24345 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
24347 2017-06-28 Richard Biener <rguenther@suse.de>
24349 * tree-vectorizer.h (vect_get_vec_defs): Remove.
24350 (vect_get_slp_defs): Adjust.
24351 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
24353 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
24355 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
24356 get_initial_defs_for_reduction instead of vect_get_vec_defs.
24357 (vectorizable_reduction): Adjust.
24358 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
24360 (vect_get_slp_defs): Likewise.
24361 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
24362 (vectorizable_bswap): Adjust.
24363 (vectorizable_call): Likewise.
24364 (vectorizable_conversion): Likewise.
24365 (vectorizable_assignment): Likewise.
24366 (vectorizable_shift): Likewise.
24367 (vectorizable_operation): Likewise.
24368 (vectorizable_store): Likewise.
24369 (vectorizable_condition): Likewise.
24370 (vectorizable_comparison): Likewise.
24372 2017-06-28 Michael Collison <michael.collison@arm.com>
24375 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
24377 (arm_gen_movmemqi): Removed unused variable 'i'.
24378 Convert 'for' loop into 'while' loop.
24379 (arm_expand_prologue): Remove last unnecessary set of insn.
24380 (thumb_pop): Remove unused variable 'pushed_words'.
24381 (thumb_exit): Remove last unnecessary set of regs_to_pop.
24383 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24385 * config/s390/predicates.md: Use s390_rel_address_ok_p.
24386 * config/s390/s390-protos.h: Add prototype of
24387 s390_rel_address_ok_p.
24388 * config/s390/s390.c (s390_got_symbol): New function.
24389 (s390_rel_address_ok_p): New function.
24390 (legitimize_pic_address): Use s390_rel_address_ok_p.
24391 (s390_load_got): Use s390_got_symbol.
24392 (s390_option_override): Issue error if
24393 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
24394 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
24396 * config/s390/s390.opt: New option mpic-data-is-text-relative.
24398 2017-06-27 Andrew Pinski <apinski@cavium.com>
24400 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
24401 (X * copysign (1.0, X)): New pattern.
24402 (X * copysign (1.0, -X)): New pattern.
24403 (copysign (-1.0, CST)): New pattern.
24405 2017-06-27 Joseph Myers <joseph@codesourcery.com>
24407 * genmultilib (combination_space): Remove variable.
24408 Validate reuse rules against regular expression for any sequence
24409 of multilib options in any order.
24411 2017-06-27 Michael Collison <michael.collison@arm.com>
24413 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
24414 call aarch64_split_simd_combine.
24415 * (aarch64_combine_internal<mode>): Delete pattern.
24416 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
24417 Allow register and subreg operands.
24419 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24421 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
24422 specific need, just fallback on defaults.
24423 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
24425 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24426 Olivier Hainque <hainque@adacore.com>
24428 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
24430 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
24431 targets. Pick a default if no particular attempt applied.
24432 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
24435 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24437 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
24438 (x86_64-wrs-vxworks7): Likewise.
24440 2017-06-27 Marek Polacek <polacek@redhat.com>
24443 * ubsan.c (instrument_null): Check get_base_address's result for null.
24445 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24447 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
24449 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24451 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
24452 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
24453 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
24454 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
24455 New function types.
24456 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
24457 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
24458 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
24459 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
24460 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
24461 BUILT_IN_FEUPDATEENV): New builtins.
24462 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
24463 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
24464 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
24465 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
24467 (builtin_structptr_types): Adjust size.
24468 * tree.c (builtin_structptr_types): Add four entries.
24470 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24471 Olivier Hainque <hainque@adacore.com>
24473 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
24474 (TLS_SYM): New local macro, forcing reference to __tls__ on
24475 link command lines for VxWorks 7 RTPs, triggering initialization
24477 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
24478 OS features TLS support, true for RTPs on VxWorks 7.
24479 * config/vxworks.c (vxworks_override_options): Setup emutls
24482 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24484 * predict.c (test_prediction_value_range): Use -1U instead of -1
24485 to avoid narrowing conversion warning.
24486 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
24487 to avoid narrowing conversion warning.
24488 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
24490 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
24492 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24494 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
24495 64bit configurations.
24496 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
24497 (SIZE_TYPE): Likewise.
24498 * config/vxworks.c (vxworks_emutls_var_fields): Use
24499 long_unsigned_type_node instead of unsigned_type_node as the offset
24500 field type, which is "pointer" mode in emutls.c.
24502 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24505 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
24507 PR middle-end/81207
24508 * gimple-fold.c (replace_call_with_call_and_fold): Handle
24509 gimple_vuse copying separately from gimple_vdef copying.
24511 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24513 * value-prof.c (free_hist): Remove call to memset and the enclosing if
24516 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24517 Olivier Hainque <hainque@adacore.com>
24519 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
24520 for all vxworks7 targets.
24521 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
24522 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
24523 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
24524 variations for VX6/VX7 and 32/64bits later on in ...
24525 (VXWORKS_LIB_SPEC): Leverage new macros.
24526 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
24527 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
24529 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24531 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
24532 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
24534 2017-06-26 Carl Love <cel@us.ibm.com>
24536 * config/rs6000/rs6000-c.c: Add support for built-in functions
24537 vector bool char vec_reve (vector bool char);
24538 vector signed char vec_reve (vector signed char);
24539 vector unsigned char vec_reve (vector unsigned char);
24540 vector bool int vec_reve (vector bool int);
24541 vector signed int vec_reve (vector signed int);
24542 vector unsigned int vec_reve (vector unsigned int);
24543 vector bool long long vec_reve (vector bool long long);
24544 vector signed long long vec_reve (vector signed long long);
24545 vector unsigned long long vec_reve (vector unsigned long long);
24546 vector bool short vec_reve (vector bool short);
24547 vector signed short vec_reve (vector signed short);
24548 vector double vec_reve (vector double);
24549 vector float vec_reve (vector float);
24550 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
24551 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
24552 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
24553 (altivec_vreve): New pattern.
24554 * config/rs6000/altivec.h (vec_reve): New define.
24555 * doc/extend.texi (vec_rev): Update the built-in documentation file
24556 for the new built-in functions.
24558 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24560 PR tree-optimization/71815
24561 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
24563 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
24565 (slsr_process_phi): Likewise.
24566 (replace_uncond_cands_and_profitable_phis): Don't replace a
24567 multiply candidate with a stride of 1 (copy or cast).
24568 (phi_incr_cost): Call uses_consumed_by_stmt rather than
24570 (lowest_cost_path): Likewise.
24571 (total_savings): Likewise.
24573 2017-06-26 Richard Biener <rguenther@suse.de>
24576 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
24577 Use def_builtin_pure for all gather builtins.
24579 2017-06-26 Richard Biener <rguenther@suse.de>
24581 PR tree-optimization/81203
24582 * tree-tailcall.c (find_tail_calls): Do not move stmts into
24583 non-dominating BBs.
24585 2017-06-26 Marek Polacek <polacek@redhat.com>
24588 * doc/invoke.texi: Document -Wmultistatement-macros.
24590 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
24592 * doc/sourcebuild.texi (ARM-specific attributes): Document new
24593 arm_neon_ok_no_float_abi effective target.
24595 2017-06-26 Richard Biener <rguenther@suse.de>
24597 PR tree-optimization/80928
24598 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
24599 (copy_bbs): Set BB_DUPLICATED flag early.
24600 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
24602 (execute_on_shrinking_pred): Likewise.
24603 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
24604 BB_DUPLICATED blocks.
24605 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
24606 iterate over all PHIs considering removal of *gsi.
24608 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
24610 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
24613 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24615 * config/rs6000/rs6000-string.c: (expand_block_clear,
24616 do_load_for_compare, select_block_compare_mode,
24617 compute_current_alignment, expand_block_compare,
24618 expand_strncmp_align_check, expand_strn_compare,
24619 expand_block_move, rs6000_output_load_multiple)
24620 Move functions related to string/block move/compare
24621 to a separate file.
24622 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
24623 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
24624 for this function which is now used in two files.
24625 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
24626 * config.gcc: Add rs6000-string.o to extra_objs for
24627 targets powerpc*-*-* and rs6000*-*-*.
24629 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
24632 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
24633 32-bit, since indexed is not valid for DImode.
24634 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
24635 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
24636 (define_peephole2 for Altivec d-form load): Add 32-bit support.
24637 (define_peephole2 for Altivec d-form store): Likewise.
24640 * multiple_target.c (create_dispatcher_calls): Only create the
24641 dispatcher call if the function is the default clone of a
24642 versioned function.
24644 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
24646 PR middle-end/80902
24647 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
24648 a call, force the call to not be a tail call.
24650 2017-06-23 Jeff Law <law@redhat.com>
24652 * doc/contrib.texi: Add entry for Steven Pemberton's work on
24655 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
24657 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
24658 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
24659 handling for early expansion of vector shifts (sl,sr,sra,rl).
24660 (builtin_function_type): Add vector shift right instructions
24661 to the unsigned argument list.
24663 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
24665 rtl-optimizatoin/79286
24666 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
24667 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
24668 trap. PIC register plus a const unspec without offset can never trap.
24670 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
24672 * tree.h (builtin_structptr_type): New type.
24673 (builtin_structptr_types): Declare new array.
24674 * tree.c (builtin_structptr_types): New array.
24675 (free_lang_data, build_common_tree_nodes): Use it.
24677 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
24680 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
24683 2017-06-22 Matt Turner <mattst88@gmail.com>
24685 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
24686 Lake models to skylake case. Assume skylake for unknown
24687 models with clflushopt.
24689 2017-06-22 Jeff Law <law@redhat.com>
24691 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
24692 frame sizes that do not satisfy aarch64_uimm12_shift.
24694 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
24696 * profile-count.h (apply_probability,
24697 apply_scale, probability_in): Fix checks for zero.
24699 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24701 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
24702 * doc/cppdiropts.texi (-I @var{dir}): Document it.
24704 2016-06-22 Richard Biener <rguenther@suse.de>
24706 * tree-vect-loop.c (vect_model_reduction_cost): Handle
24707 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
24708 REDUC_MAX_EXPR support.
24709 (vectorizable_reduction): Likewise.
24710 (vect_create_epilog_for_reduction): Likewise.
24712 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
24714 * match.pd (A / (1 << B) -> A >> B): New.
24715 * generic-match-head.c: Include optabs-tree.h.
24716 * gimple-match-head.c: Likewise.
24717 * optabs-tree.h (target_supports_op_p): New.
24718 * optabs-tree.c (target_supports_op_p): New.
24720 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24722 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
24723 $gcc_cv_ld --help output.
24724 (gcc_cv_ld_demangle): Likewise.
24725 (gcc_cv_ld_eh_frame_hdr): Likewise.
24726 (gcc_cv_ld_pie): Likewise.
24727 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
24728 (gcc_cv_ld_buildid): Likewise.
24729 (gcc_cv_ld_sysroot): Likewise.
24730 (ld_bndplt_support): Likewise.
24731 (ld_pushpopstate_support): Likewise.
24732 * configure: Regenerate.
24733 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
24735 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24738 * config/i386/sse.md (round<mode>2): Renumber match_dup and
24739 operands indexes to avoid gap between operands and match_dups.
24741 2017-06-21 Andrew Pinski <apinski@cavium.com>
24743 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
24744 Increment Arith_shift and Arith_shift_reg by 1.
24745 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
24747 * config/aarch64/aarch64.c (thunderx_tunings): Enable
24748 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
24749 (aarch64_strip_extend): Add new argument and test for it.
24750 (aarch64_cheap_mult_shift_p): New function.
24751 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
24752 add a cost if it is true.
24753 Update calls to aarch64_strip_extend.
24754 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
24756 2017-06-21 Andrew Pinski <apinski@cavium.com>
24758 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
24760 (thunderxt88): Likewise.
24761 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
24762 (thunderx_prefetch_tune): New variable.
24763 (thunderx2t99_prefetch_tune): Update for the correct values.
24764 (thunderxt88_tunings): New variable.
24765 (thunderx_tunings): Use thunderx_prefetch_tune instead of
24766 generic_prefetch_tune.
24767 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
24769 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24771 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
24772 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
24773 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
24774 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
24775 (aarch64_atomic_cas<mode>, GPI): Likewise.
24777 2017-06-21 Martin Liska <mliska@suse.cz>
24779 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
24780 statements on cold and hot labels.
24781 * predict.c (tree_estimate_probability_bb): Remove the
24782 prediction from this place.
24784 2017-06-21 Martin Liska <mliska@suse.cz>
24786 PR tree-optimization/79489
24787 * gimplify.c (maybe_add_early_return_predict_stmt): New
24789 (gimplify_return_expr): Call the function.
24790 * predict.c (tree_estimate_probability_bb): Remove handling
24792 * predict.def: Update comment about early return predictor.
24793 * gimple-predict.h (is_gimple_predict): New function.
24794 * predict.def: Change default value of early return to 66.
24795 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
24797 * passes.def: Put pass_strip_predict_hints to the beginning of
24800 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
24802 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
24803 FUNCTION_DECL declarations.
24804 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
24806 (dwaf2out_decl): Likewise.
24807 * godump.c (go_early_global_decl): Skip call to the real debug hook
24808 for FUNCTION_DECL declarations.
24809 * passes.c (rest_of_decl_compilation): Skip call to the
24810 early_global_decl debug hook for FUNCTION_DECL declarations, unless
24811 -fdump-go-spec is passed.
24813 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24815 * config/i386/i386.c (struct builtin_isa): New field pure_p.
24816 Reorder for compactness.
24817 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
24818 (def_builtin_pure, def_builtin_pure2): New functions.
24819 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
24821 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24823 * match.pd (nop_convert): New predicate.
24824 ((A +- CST1) +- CST2): Allow some NOP conversions.
24826 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24829 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
24830 with ctors/dtors if GOVD_SHARED is set.
24832 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24834 * config/aarch64/aarch64.md (movti_aarch64):
24835 Emit mov rather than orr.
24836 (movtf_aarch64): Likewise.
24837 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
24838 Emit mov rather than orr.
24840 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24842 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
24843 Swap alternatives, make integer dup more expensive.
24845 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24847 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24848 Return true for non-tls symbols.
24850 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
24852 * config/aarch64/aarch64-cores.def (cortex-a55): New.
24853 (cortex-a75): Likewise.
24854 (cortex-a75.cortex-a55): Likewise.
24855 * config/aarch64/aarch64-tune.md: Regenerate.
24856 * doc/invoke.texi (-mtune): Document new values for -mtune.
24858 2017-06-21 Tom de Vries <tom@codesourcery.com>
24860 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
24861 stack_size feature.
24862 (Effective-Target Keywords, Other attributes): Suggest using
24863 dg-add-options stack_size feature to get stack limit in stack_size
24864 effective target documentation.
24866 2017-06-21 Julian Brown <julian@codesourcery.com>
24867 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24869 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
24870 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
24871 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
24873 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
24874 attribute type list for neon_multiply.
24875 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
24876 attribute type list for neon_multiply.
24877 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
24878 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
24879 attribute type list for neon_multiply.
24880 * config/arm/types.md (crypto_pmull): Add.
24881 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
24882 attribute type list.
24884 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
24886 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
24889 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24891 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
24892 to make sure not to dereference a NULL cost_classes_ptr pointer.
24894 2017-06-20 Carl Love <cel@us.ibm.com>
24896 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24897 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24898 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24899 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24900 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24901 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24902 VMULOSW): New enum "unspec" values.
24903 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24904 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24905 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24906 altivec_vmulosw): New patterns.
24907 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24908 VMULOSW): Add definitions.
24910 2017-06-20 Julia Koval <julia.koval@intel.com>
24912 * config/i386/i386.c: Fix rounding expand for new pattern.
24913 * config/i386/subst.md: Fix pattern (parallel -> unspec).
24915 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24917 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
24918 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
24920 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24922 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
24925 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24927 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24928 architecture, then by implementer ID.
24929 * config/aarch64/aarch64-tune.md: Regenerate.
24931 2017-06-20 Richard Biener <rguenther@suse.de>
24933 PR middle-end/81097
24934 * fold-const.c (split_tree): Fold to type before negating.
24936 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24938 * diagnostic-show-locus.c
24939 (selftest::test_fixit_deletion_affecting_newline): New function.
24940 (selftest::diagnostic_show_locus_c_tests): Call it.
24942 2017-06-20 Andreas Schwab <schwab@suse.de>
24945 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24948 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24950 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24951 __ARM_FEATURE_COPROC according to support.
24953 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24955 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24956 Rewritten to avoid overflow for > 32-bit pointers.
24959 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24960 by removing enum keyword.
24961 (ubsan_type_descriptor): Likewise. Formatting fix.
24964 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24965 splitter): Require TARGET_SSE2 in the condition.
24967 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24970 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24971 for doing vector set of SFmode on ISA 3.0.
24972 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24973 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24975 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24976 SFmode value into a V4SF variable that was extracted from another
24977 V4SF variable without converting the element to double precision
24978 and back to single precision vector format.
24979 (vsx_insert_extract_v4sf_p9_2): Likewise.
24981 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24983 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24984 in UWHI to avoid undefined overflow.
24987 * ubsan.h (enum ubsan_encode_value_phase): New.
24988 (ubsan_encode_value): Change second argument to
24989 enum ubsan_encode_value_phase with default value of
24990 UBSAN_ENCODE_VALUE_GENERIC.
24991 * ubsan.c (ubsan_encode_value): Change second argument to
24992 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24993 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24994 create_tmp_var_raw instead of create_tmp_var and use a
24996 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24997 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24998 ubsan_encode_value callers.
25001 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
25002 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
25003 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
25005 2017-06-19 Richard Biener <rguenther@suse.de>
25007 PR middle-end/81118
25008 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
25009 estimates if we changed anything.
25011 2017-06-19 Richard Biener <rguenther@suse.de>
25013 PR tree-optimization/80887
25014 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
25015 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
25016 simplified lookups, then reset mprts_hook.
25017 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
25019 (try_to_simplify): Likewise.
25021 2017-06-19 Martin Liska <mliska@suse.cz>
25024 * gimplify.c (gimplify_switch_expr):
25025 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
25027 2017-06-19 Martin Liska <mliska@suse.cz>
25029 * doc/install.texi: Document that PGO runs in 4 stages.
25031 2017-06-19 Martin Liska <mliska@suse.cz>
25034 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
25035 to dispatcher function name.
25036 * multiple_target.c (replace_function_decl): New function.
25037 (create_dispatcher_calls): Redirect both edges and references.
25039 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
25041 * profile-count.c (profile_count::dump): Dump quality.
25042 (profile_count::differs_from_p): Update for unsigned val.
25043 * profile-count.h (profile_count_quality): New enum.
25044 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
25046 2017-06-19 Richard Biener <rguenther@suse.de>
25048 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
25049 struct function as arg.
25050 (estimate_numbers_of_iterations): Export overload with loop arg.
25051 (free_numbers_of_iterations_estimates_loop): Use an overload of
25052 free_numbers_of_iterations_estimates instead.
25053 * tree-cfg.c (remove_bb): Adjust.
25054 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
25055 * tree-parloops.c (gen_parallel_loop): Likewise.
25056 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
25058 (tree_unroll_loops_completely): Likewise.
25059 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
25060 Use an overload instead and export.
25061 (estimated_loop_iterations): Adjust.
25062 (max_loop_iterations): Likewise.
25063 (likely_max_loop_iterations): Likewise.
25064 (estimate_numbers_of_iterations): Take struct function as arg
25066 (loop_exits_before_overflow): Adjust.
25067 (free_numbers_of_iterations_estimates_loop): Use an overload.
25068 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
25069 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
25071 2017-06-19 Richard Biener <rguenther@suse.de>
25074 * ipa-prop.c (find_constructor_constant_at_offset): Handle
25075 RANGE_EXPR conservatively.
25077 2017-06-16 Carl Love <cel@us.ibm.com>
25079 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25080 definitions for vec_float, vec_float2, vec_floato,
25081 vec_floate built-ins.
25082 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
25083 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
25085 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
25086 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
25087 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
25088 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
25089 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
25090 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
25091 vec_floato): Add builtin defines.
25092 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
25093 Update the built-in documentation file for the new built-in
25096 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25098 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
25099 (mthumb): Mark as the negative of -marm.
25101 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25103 * doc/invoke.texi (ARM Options, -mcpu): Document supported
25105 (ARM Options, -mtune): Document that this accepts the same
25106 extension options as -mcpu.
25107 (ARM Options, -mfpu): Document addition of -mfpu=auto.
25109 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25111 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
25112 permitted extensions.
25114 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25116 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
25117 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
25118 (armv8-m.main): Add option +nodsp.
25119 * config/arm/arm-cpu-cdata.h: Regenerated.
25121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25123 * config/arm/t-fuchsia: New file.
25124 * config.gcc (arm*-*-fuchsia*): Use it.
25126 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25128 * config/arm/t-symbian: Rewrite for new option infrastructure.
25130 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25132 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
25133 (MULTILIB_REQUIRED): Likewise.
25135 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25137 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
25138 (MULTILIB_RESUE): Likewise.
25139 (MULTILIB_MATCHES): Likewise.
25140 (MULTLIB_REQUIRED): Likewise.
25142 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25144 * config/arm/t-rtems: Rewrite for new option framework.
25146 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25148 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
25149 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
25150 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
25151 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
25152 * config/arm/t-multilib: ... here.
25153 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
25154 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
25155 armv7-a and armv8*-a when A-profile libraries have not been built.
25156 * config/arm/t-rmprofile: Rewrite.
25158 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25160 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
25161 with a backslash. Remove the backslash after substituting unescaped
25163 * doc/fragments.texi (MULTILIB_REUSE): Document it.
25165 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25167 * config.gcc: (arm*-*-*): When building a-profile libraries, force
25168 the driver to pass through the default setting of -mfloat-abi.
25169 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
25171 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
25172 (all_feat_combs): New rule.
25173 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
25175 * config/arm/t-aprofile: Rewrite.
25177 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25179 * config/arm/arm.h (FPUTYPE_AUTO): Define.
25180 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
25181 fpu is not specified by the user/command-line.
25182 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
25183 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
25184 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
25185 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
25186 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
25187 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
25189 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25191 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
25192 * config/arm/t-arm-elf: Rewritten.
25194 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25196 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
25197 have some floating-point instructions.
25198 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
25199 (TARGET_MAYBE_HARD_FLOAT): New macro.
25200 * config/arm/arm-builtins.c (arm_init_builtins): Use
25201 TARGET_MAYBE_HARD_FLOAT.
25202 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
25204 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25206 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
25207 (configargs.h): Include it.
25208 (arm_print_hint_for_fpu_option): New function.
25209 (arm_parse_fpu_option): New function.
25210 (candidate_extension): New class.
25211 (arm_canon_for_multilib): New function.
25212 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
25213 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
25214 (ARCH_CANONICAL_SPECS): New macro.
25215 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
25217 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25219 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
25220 are set after handling multilib fragments. Set target_cpu_default2
25223 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25225 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
25227 (arm*-*-*): Set target_cpu_default2 to a quoted string.
25228 * config/arm/parsecpu.awk (check_cpu): Validate any extension
25230 (check_arch): Likewise.
25231 * config/arm/arm.c (arm_configure_build_target): Handle
25232 TARGET_CPU_DEFAULT being a string constant. Scan any feature
25233 options in the default.
25235 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25237 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
25238 when an option is an alias of another.
25239 * config/arm/parsecpu.awk (optalias): New parser token.
25240 (gen_comm_data): Mark non-alias options as such. Emit entries
25241 for extension aliases.
25242 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
25243 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
25244 (armv6kz, armv6zk, armv6t2): Likewise.
25245 (armv7): Make vfpv3-d16 an alias.
25246 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
25248 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
25249 Sort in canonical order.
25250 (armv8-a): Sort in canonical order.
25251 (armv8.1-a, armv8.2-a): Likewise.
25252 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
25254 (cortex-a9): Sort in canonical order.
25255 * config/arm/arm.c (selftests.h): Include it.
25256 (arm_test_cpu_arch_data): New function.
25257 (arm_run_self_tests): New function.
25258 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
25259 (targetm): Move declaration to the end of the file.
25260 * arm-cpu-cdata.h: Regenerated.
25262 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25264 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
25265 call to target_mode_check describing the type of option passed.
25266 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
25267 (arm_target_thumb_only): Use arm_parse_arch_option_name or
25268 arm_parse_cpu_option_name to match parameters against list of
25270 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
25271 arm_arch_core_flags data structure.
25272 * config/arm/arm-cpu_cdata.h: Regenerated.
25274 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25276 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
25278 (arm_print_hint_for_cpu_option): Likewise.
25279 (arm_print_hint_for_arch_option): Likewise.
25280 (arm_parse_cpu_option_name): Likewise.
25281 (arm_parse_arch_option_name): Likewise.
25282 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
25283 of entries in the all_fpus list.
25284 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
25285 (arm_parse_cpu_option_name): Declare.
25286 (arm_parse_arch_option_name): Declare.
25287 (arm_parse_option_features): Declare.
25288 (arm_intialize_isa): Declare.
25289 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
25291 (gen_comm_data): ... here. Make definitions non-static.
25292 * config/arm/arm-cpu-data.h: Regenerated.
25293 * config/arm/arm-cpu-cdata.h: Regenerated.
25295 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25297 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
25298 (cpu_arch_extension): New structure.
25299 (cpu_arch_option, arch_option, cpu_option): New structures.
25300 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
25301 architecture types.
25302 (gen_data): Generate new format data tables.
25303 * config/arm/arm.c (cpu_tune): New structure.
25304 (cpu_option, processors): Delete.
25305 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
25306 (arm_print_hint_for_cpu_option): ... this and ...
25307 (arm_print_hint_for_arch_option): ... this.
25308 (arm_parse_arch_cpu_name): Delete. Replace with ...
25309 (arm_parse_cpu_option_name): ... this and ...
25310 (arm_parse_arch_option_name): ... this.
25311 (arm_unrecognized_feature): Change type of target parameter to
25313 (arm_parse_arch_cpu_features): Delete. Replace with ...
25314 (arm_parse_option_features): ... this.
25315 (arm_configure_build_target): Rework to use new configuration data
25317 (arm_print_tune_info): Rework for new configuration data tables.
25318 * config/arm/arm-cpu-data.h: Regenerated.
25319 * config/arm/arm-cpu.h: Regenerated.
25321 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25323 * Makefile.in (OBJS): Move sbitmap.o from here ...
25324 (OBJS-libcommon): ... to here.
25326 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25328 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
25329 (ISA_ALL_CRYPTO): New macro.
25330 (ISA_ALL_SIMD): New macro
25331 (ISA_ALL_FP): New macro.
25332 * config/arm/arm.c (fpu_bitlist): Update initializer.
25333 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
25335 (arm9e): Add fpu. Add option for nofp
25336 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
25337 (arm926ej-s, arm1026ej-s): Likewise.
25338 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
25339 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
25340 neon-fp16, neon-vfpv4, nofp and nosimd.
25341 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
25342 (cortex-a8): Add fpu. Add option for nofp.
25343 (cortex-a9): Add fpu. Add options for nosimd and nofp.
25344 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
25345 (cortex-r4f): Add fpu.
25346 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
25347 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
25349 (cortex-r8): Likewise.
25350 (cortex-m4): Add fpu. Add option for nofp.
25351 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
25352 (cortex-a17.cortex-a7): Likewise.
25353 (cortex-a32): Add fpu. Add options for crypto and nofp.
25354 (cortex-a35, cortex-a53): Likewise.
25355 (cortex-a57): Add fpu. Add option for crypto.
25356 (cortex-a72, cortex-a73): Likewise.
25357 (exynos-m1): Likewise.
25358 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
25359 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
25360 (cortex-m33): Add fpu. Add option for nofp.
25361 * config/arm/arm-cpu-cdata.h: Regenerated
25362 * config/arm/arm-cpu-data.h: Regenerated.
25364 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25366 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
25367 (armv5te, armv5tej): Likewise.
25368 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
25369 (armv7): Add options fp and vfpv3-d16.
25370 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
25371 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
25373 (armv7ve): Likewise.
25374 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
25375 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
25376 (armv8-a): Add nocrypto option.
25377 (armv8.1-a, armv8.2-a): Likewise.
25378 (armv8-m.main): add options fp, fp.dp and nofp.
25380 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25382 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
25384 (armv8-a+crc): Delete.
25385 (armv8.1-a): Add options simd, crypto and nofp.
25386 (armv8.2-a): Add options fp16, simd, crypto and nofp.
25387 (armv8.2-a+fp16): Delete.
25388 (armv8-m.main): Add option dsp.
25389 (armv8-m.main+dsp): Delete.
25390 (cortex-a8): Add fpu. Add nofp option.
25391 (cortex-a9): Add fpu. Add nofp and nosimd options.
25392 * config/arm/parsecpu.awk (gen_data): Generate option tables and
25393 link to main cpu and architecture data structures.
25394 (gen_comm_data): Only put isa attributes from the main architecture
25396 (option): New statement for architecture and CPU entries.
25397 * arm.c (struct cpu_option): New structure.
25398 (struct processors): Add entry for options.
25399 (arm_unrecognized_feature): New function.
25400 (arm_parse_arch_cpu_name): Ignore any characters after the first
25402 (arm_parse_arch_cpu_feature): New function.
25403 (arm_configure_build_target): Separate out any CPU and architecture
25404 features and parse separately. Don't error out if -mfpu=auto is
25405 used with only an architecture string.
25406 (arm_print_asm_arch_directives): New function.
25407 (arm_file_start): Call it.
25408 * config/arm/arm-cpu-cdata.h: Regenerated.
25409 * config/arm/arm-cpu-data.h: Likewise.
25410 * config/arm/arm-tables.opt: Likewise.
25412 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25414 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
25415 assembler when it is not -mfpu=auto.
25417 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25419 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
25420 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
25421 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
25422 (ASM_CPU_SPEC): Rewrite.
25423 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
25424 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
25425 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
25426 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
25427 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
25428 copied string is NUL-terminated. Also strip any characters prefixed
25430 (arm_rewrite_selected_arch): New function.
25431 (arm_rewrite_march): New function.
25433 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25435 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
25436 (x_arm_cpu_string, x_arm_tune_string): Likewise.
25437 (march, mcpu, mtune): Convert to string-based options.
25438 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
25439 (arm_parse_arch_cpu_name): New function.
25440 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
25441 identify selected architecture or CPU.
25442 (arm_option_save): New function.
25443 (TARGET_OPTION_SAVE): Redefine.
25444 (arm_option_restore): Restore string options.
25445 (arm_option_print): Print string options.
25447 2017-06-16 Martin Sebor <msebor@redhat.com>
25449 PR tree-optimization/80933
25450 PR tree-optimization/80934
25451 * builtins.c (fold_builtin_3): Do not handle bcmp here.
25452 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
25453 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
25454 (gimple_fold_builtin): Call them.
25456 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25458 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
25459 as unlikely; update profile.
25461 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25463 * predict.c (force_edge_cold): Handle declaring edges impossible
25466 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25468 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
25470 (try_unroll_loop_completely): Fix reporting.
25472 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25474 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
25476 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
25479 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
25480 if given a non-constant argument for an intrinsic which requires a
25483 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25485 * profile.c (compare_freqs): New function.
25486 (branch_prob): Sort edge list.
25487 (find_spanning_tree): Assume that the list is priority sorted.
25489 2017-06-16 Richard Biener <rguenther@suse.de>
25491 PR tree-optimization/81090
25492 * passes.def (pass_record_bounds): Remove.
25493 * tree-pass.h (make_pass_record_bounds): Likewise.
25494 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
25495 make_pass_record_bounds): Likewise.
25496 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
25497 not free niter estimates at the beginning but at the end.
25498 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
25500 2017-06-16 Richard Biener <rguenther@suse.de>
25502 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
25503 initializer to workaround ICE in host GCC 4.8.
25505 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25507 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
25509 (clone_inlined_nodes): Update.
25511 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25513 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
25514 prefetch settings, and enable prefetching by default at -O3.
25516 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25518 * config/aarch64/aarch64.c (aarch64_override_options_internal):
25519 Set flag_prefetch_loop_arrays according to tuning data.
25521 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25523 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
25524 New tune structure.
25525 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
25526 [Unrelated to main purpose of the patch] Place the pointer field last
25527 to enable type checking errors when tune structure are wrongly merged.
25528 * config/aarch64/aarch64.c (generic_prefetch_tune,)
25529 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
25530 (thunderx2t99_prefetch_tune): New tune constants.
25531 (tune_params *_tunings): Update all tunings (no functional change).
25532 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
25533 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
25534 from tunings structures.
25536 2017-06-16 Jakub Jelinek <jakub@redhat.com>
25539 * ubsan.c (instrument_null): Add T argument, use it instead
25540 of computing it based on IS_LHS.
25541 (instrument_object_size): Likewise.
25542 (pass_ubsan::execute): Adjust instrument_null and
25543 instrument_object_size callers to pass gimple_get_lhs or
25544 gimple_assign_rhs1 result to it. Use instrument_null instead of
25545 calling get_base_address and instrument_mem_ref. Handle
25546 aggregate call arguments for object-size sanitization.
25548 2017-06-16 Yury Gribov <tetra2005@gmail.com>
25550 PR tree-optimization/81089
25551 * tree-vrp.c (is_masked_range_test): Validate operands of
25554 2017-06-15 Martin Sebor <msebor@redhat.com>
25557 * dumpfile.c (dump_register): Avoid calling memset to initialize
25558 a class with a default ctor.
25559 * gcc.c (struct compiler): Remove const qualification.
25560 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
25561 * hash-table.h: Ditto.
25562 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
25564 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
25565 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
25567 * params.h (struct param_info): Make struct members non-const.
25568 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
25569 with default initialization.
25570 * vec.h (vec_copy_construct, vec_default_construct): New helper
25572 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
25573 with vec_copy_construct.
25574 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
25575 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
25576 * doc/invoke.texi (-Wclass-memaccess): Document.
25578 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25580 * emit-rtl.h (is_leaf): Update comment about local
25581 register allocator.
25583 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25586 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
25587 for a variable to have a section before checking if the section has a
25589 Set section to.persistent if persistent attribute is set.
25590 Warn if .persistent attribute is used on an automatic variable.
25592 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
25594 PR rtl-optimization/80474
25595 * reorg.c (update_block): Do not ignore instructions in a delay slot.
25597 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
25599 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
25602 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
25604 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
25605 (casesi): Emit bounds checking as RTL.
25606 (casesi_internal_mips16_<mode>): Remove bounds checking.
25608 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
25610 * config/xtensa/xtensa.c (xtensa_option_override): Append
25611 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
25612 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
25613 xtensa_doloop_hooks): Define unconditionally.
25614 (xtensa_reorg_loops): Only call reorg_loops in the presence of
25616 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
25617 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
25618 for it in xtensa_option_override.
25619 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
25620 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
25622 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
25624 * doc/cppopts.texi: Document '-' special value to -MF.
25626 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
25628 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
25629 (cortex_a53_fconst): Likewise.
25630 (cortex_a53_fpmul): Likewise.
25631 (cortex_a53_f_load_64): Likewise.
25632 (cortex_a53_f_load_many): Likewise.
25633 (cortex_a53_advsimd_alu): Likewise.
25634 (cortex_a53_advsimd_alu_q): Likewise.
25635 (cortex_a53_advsimd_mul): Likewise.
25636 (cortex_a53_advsimd_mul_q): Likewise.
25637 (fpmac bypass): Add new bypass for fpmac-fpmac case.
25638 Add missing fmul, r2f_cvt and fconst cases.
25640 2017-06-14 Richard Biener <rguenther@suse.de>
25642 PR middle-end/81088
25643 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
25645 (fold_binary_loc): When associating do not treat pre-existing
25646 TREE_OVERFLOW on literal constants as a reason to allow
25647 TREE_OVERFLOW on associated literal constants.
25649 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25651 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
25652 (MASK_FEATURES): New macro.
25653 * config/sparc/sparc.c (sparc_option_override): Remove the special
25654 handling of -mfpu and generalize it to all MASK_FEATURES switches.
25656 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25658 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
25659 a division of 0 if non-call exceptions are enabled.
25661 2017-06-14 Andrew Pinski <apinski@cavium.com>
25662 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25665 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
25666 Improve vector initialization code gen for only variable case.
25668 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25670 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
25672 2017-06-14 Richard Biener <rguenther@suse.de>
25674 PR tree-optimization/81083
25675 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
25678 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25680 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
25681 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
25682 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
25683 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
25684 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
25685 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
25687 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25689 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
25690 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
25692 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25694 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
25696 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25698 * config/rs6000/t-rtems: Don't handle SPE.
25700 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25702 * config/rs6000/t-linux: Don't handle SPE.
25704 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25706 * config/rs6000/eabispe.h: Delete file.
25708 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25710 * config/rs6000/t-spe: Delete file.
25712 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25714 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
25715 (rs6000_legitimate_offset_address_p): Return false for anything in
25716 V2SImode or V2SFmode.
25718 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25720 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
25721 except V2SF and V2SI. Rearrange the vector modes, and add comments.
25722 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
25724 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
25725 (rs6000_legitimate_offset_address_p): Ditto.
25726 (rs6000_emit_move): Ditto.
25727 (rs6000_init_builtins): Remove V4HI_type_node.
25729 2017-06-13 Martin Liska <mliska@suse.cz>
25732 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
25733 (gate_asan): Likewise.
25734 * asan.h (asan_no_sanitize_address_p): Remove the function.
25735 (sanitize_flags_p): New function.
25736 * builtins.def: Fix coding style.
25737 * common.opt: Use renamed enum value.
25738 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
25739 * doc/extend.texi: Document no_sanitize attribute.
25740 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
25741 to SANITIZE_UNDEFINED_NONDEFAULT.
25742 * gcc.c (sanitize_spec_function): Use the renamed enum value.
25743 * gimple-fold.c (optimize_atomic_compare_exchange_p):
25744 Use sanitize_flags_p.
25745 * gimplify.c (gimplify_function_tree): Likewise.
25746 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
25747 * opts.c (parse_no_sanitize_attribute): New function.
25748 (common_handle_option): Use renamed enum value.
25749 * opts.h (parse_no_sanitize_attribute): Declare.
25750 * tree.c (sanitize_flags_p): New function.
25751 * tree.h: Declared here.
25752 * tsan.c: Use sanitize_flags_p.
25753 * ubsan.c (ubsan_expand_null_ifn): Likewise.
25754 (instrument_mem_ref): Likewise.
25755 (instrument_bool_enum_load): Likewise.
25756 (do_ubsan_in_current_function): Remove the function.
25757 (pass_ubsan::execute): Use sanitize_flags_p.
25758 * ubsan.h: Remove do_ubsan_in_current_function
25759 * tree-cfg.c (print_no_sanitize_attr_value): New function.
25760 (dump_function_to_file): Use it here.
25762 2017-06-13 Martin Jambor <mjambor@suse.cz>
25764 PR tree-optimization/80803
25765 PR tree-optimization/81063
25766 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
25767 (propagate_subaccesses_across_link): Enqueue subtree whenever
25768 necessary instead of relying on the caller.
25770 2017-06-13 Martin Jambor <mjambor@suse.cz>
25772 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
25773 that have a first_link.
25774 (sort_and_splice_var_accesses): Do not check first_link before
25776 (subtree_mark_written_and_enqueue): Likewise.
25777 (propagate_all_subaccesses): Likewise and do not stop at first
25778 parent with a first_link.
25780 2017-06-13 Martin Jambor <mjambor@suse.cz>
25782 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
25785 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25787 * match.pd: New pattern.
25789 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25791 * tree-vrp.c (is_masked_range_test): New function.
25792 (register_edge_assert_for): Determine ranges for
25795 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25797 PR tree-optimization/67328
25798 * fold-const.c (maskable_range_p): New function.
25799 (build_range_check): Generate bittests if possible.
25801 2017-06-13 Martin Liska <mliska@suse.cz>
25803 * gimple-pretty-print.c (dump_probability): Add new argument.
25804 (dump_edge_probability): Dump both probability and count.
25805 (dump_gimple_label): Likewise.
25806 (dump_gimple_bb_header): Likewise.
25808 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
25811 * config/avr/avr-devices.c: Fix indentation.
25812 * config/avr/gen-avr-mmcu-specs.c: Dito.
25814 2017-06-13 Richard Biener <rguenther@suse.de>
25816 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
25817 instead get vector type from stmt_info.
25818 (vectorizable_reduction): Adjust. Remove dead code.
25820 2017-06-13 Richard Biener <rguenther@suse.de>
25822 PR middle-end/81065
25823 * fold-const.c (extract_muldiv_1): Remove bogus distribution
25824 case of C * (x * C2 + C3).
25825 (fold_addr_of_array_ref_difference): Properly fold index difference.
25827 2017-06-12 David S. Miller <davem@davemloft.net>
25830 * config/sparc/sparc.md (return expander): Emit frame blockage if
25831 function uses alloca.
25833 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
25835 * combine.c (make_field_assignment): Check len rather than the mode
25836 precision when calling force_to_mode.
25838 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
25840 Support multilibs and devices that see flash in RAM address range.
25843 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
25844 (avr_mcu_t) <flash_pm_offset>: New field.
25845 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
25846 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
25847 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
25848 (AVR_TINY_PM_OFFSET): Remove macro.
25849 * config/avr/avr.opt (-mshort-calls): New option.
25850 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
25851 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
25852 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
25853 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
25854 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
25855 instead of avr_arch->have_jmp_call.
25856 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
25857 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
25858 avr_arch->flash_pm_offset to define.
25859 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
25860 new field flash_pm_offset. Add entry for avrxmega3.
25861 (avr_texinfo): Add entry for avrxmega3.
25862 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
25863 attiny212, attiny214,
25864 attiny412, attiny414, attiny416, attiny417,
25865 attiny814, attiny816, attiny817,
25866 attiny1614, attiny1616, attiny1617,
25867 attiny3214, attiny3216, attiny3217.
25868 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
25869 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
25870 (avr_print_operand_address) [AVR_TINY]: Same.
25871 (avr_asm_init_sections) <readonly_data_section>: Only patch
25872 callback if avr_arch->flash_pm_offset = 0.
25873 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
25874 for rodata if avr_arch->flash_pm_offset != 0.
25875 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
25876 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
25877 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
25878 (m_options): Append opt_rcall.
25879 (m_dirnames): Append dir_rcall.
25880 * config/avr/t-multilib: Regenerate.
25882 * configure.ac [target=avr]: Check whether avrxmega3 default
25883 linker description file works as needed.
25884 * configure: Regenerate.
25885 * doc/avr-mmcu.texi: Regenerate.
25886 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
25887 <__AVR_ARCH__>: Document avrxmega3 and 103.
25888 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
25889 <__AVR_SHORT_CALLS__>: Document it.
25890 <__AVR_PM_BASE_ADDRESS__>: Document it.
25891 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
25892 (AVR Variable Attributes) <progmem>: Document this is
25893 not needed for avrxmega3.
25894 (AVR Named Address Spaces) <__flash>: Dito.
25896 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
25898 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
25900 2017-06-12 Doug Rupp <rupp@adacore.com>
25902 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
25903 Append vxworks-stdint.h to the tm_file list.
25904 * config/vxworks-stdint.h: New file.
25906 2017-06-12 Martin Liska <mliska@suse.cz>
25908 PR tree-optimization/81041
25909 * tree-profile.c (gimple_gen_ic_func_profiler):
25910 Create an extra BB in profile-generate
25911 (gimple_gen_time_profiler): Likewise.
25913 2017-06-12 Jakub Jelinek <jakub@redhat.com>
25915 PR tree-optimization/81003
25916 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
25917 (update_range_test): Use it instead of force_gimple_operand_gsi.
25919 2017-06-12 Richard Biener <rguenther@suse.de>
25921 PR tree-optimization/81053
25922 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
25923 with backedge value not defined in loop. Simplify def stmt
25926 2017-06-11 Tom de Vries <tom@codesourcery.com>
25929 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25931 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25932 nvptx_cannot_force_const_mem.
25934 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25936 * opts.c (finish_options): Move test for flag_split_stack after
25937 it has been initialized.
25939 2017-06-11 Jason Merrill <jason@redhat.com>
25941 * tree.h (id_equal): New.
25942 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25943 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25944 instead of strcmp of IDENTIFIER_POINTER.
25946 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25948 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25949 (mark_all_inlined_calls_cdtor): Fix formating.
25950 (inline_transform): Rescale profile before inlining.
25952 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25954 * cgraph.h (cgraph_edge::clone): Update prototype.
25955 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25956 (cgraph_node::create_clone): Update.
25957 (cgraph_node::create_version_clone): Update.
25958 * tree-inline.c (copy_bb): Update.
25959 (expand_call_inline): Update.
25961 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25963 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25964 factored out from ...
25965 (rs6000_emit_prologue): ... here.
25967 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25969 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25970 factored out from ...
25971 (rs6000_emit_prologue): ... here.
25973 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25975 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25977 (handle_missing_profiles): Fix computation of tp_first_run.
25978 (counts_to_freqs): Do not touch freqs when count is 0.
25980 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25982 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25985 2017-06-10 Tom de Vries <tom@codesourcery.com>
25987 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25988 attributes): Document signal effective target.
25990 2017-06-10 Tom de Vries <tom@codesourcery.com>
25992 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25993 Document effective target stack_size.
25995 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25997 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25998 to the edit_context if they can be auto-applied.
26000 2017-06-9 Ian Lance Taylor <iant@golang.org>
26002 * opts.c (finish_options): If -fsplit-stack, disable implicit
26003 -forder-blocks-and-partition.
26004 * doc/invoke.texi (Optimize Options): Document that when using
26005 -fsplit-stack -forder-blocks-and-partition is not implicitly
26008 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26010 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
26011 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
26012 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
26013 * builtins.def (abort, trap, unreachable): Declare cold.
26014 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
26015 * tree-core.h (ECF_COLD): New.
26016 * tree.c (set_call_expr_flags): Handle ECF_COLD.
26017 (build_common_builtin_nodes): Mark unreachable and abort as cold.
26019 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26021 * predict.c (unlikely_executed_stmt_p): Cleanup.
26023 2017-06-09 Richard Biener <rguenther@suse.de>
26025 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
26026 model if the ref is always written to.
26028 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26030 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
26032 2017-06-09 Tamar Christina <tamar.christina@arm.com>
26034 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
26037 2017-06-09 Tom de Vries <tom@codesourcery.com>
26040 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
26041 "target cannot support label values" when encountering LABEL_REF.
26043 2017-06-09 Martin Liska <mliska@suse.cz>
26045 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
26046 (gimple_gen_ic_func_profiler): Emit direct comparison
26047 of __gcov_indirect_call_callee with NULL.
26048 (gimple_gen_time_profiler): Change probability from
26049 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
26051 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26053 * profile.c (edge_gcov_counts): Turn to pointer.
26054 (compute_branch_probabilities, compute_branch_probabilities): Update.
26055 (branch_prob): Do not clear edge_gcov_count.
26056 * profile.h (edge_gcov_counts): Turn to pointer.
26057 (edge_gcov_count): Update.
26059 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
26061 * gimple.h (gimple_check_failed): Mark cold.
26063 2017-06-09 Richard Biener <rguenther@suse.de>
26065 PR tree-optimization/66623
26066 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
26067 refactor check_reduction into two parts, properly computing
26068 whether we have to check reduction validity for outer loop
26071 2017-06-09 Richard Biener <rguenther@suse.de>
26073 PR tree-optimization/79483
26074 * graphite-scop-detection.c (order): New global.
26075 (get_order): Compute bb to order mapping that satisfies code
26076 generation constraints.
26077 (cmp_pbbs): New helper.
26078 (build_scops): Start domwalk at entry block, sort generated
26081 2017-06-09 Richard Biener <rguenther@suse.de>
26083 PR middle-end/81007
26084 * ipa-polymorphic-call.c
26085 (ipa_polymorphic_call_context::restrict_to_inner_class):
26086 Skip FIELD_DECLs with error_mark_node type.
26087 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
26090 2017-06-09 Martin Liska <mliska@suse.cz>
26092 * predict.c (struct branch_predictor): New struct.
26093 (test_prediction_value_range): New test.
26094 (predict_c_tests): New function.
26095 * selftest-run-tests.c (selftest::run_tests): Run the function.
26096 * selftest.h: Declare new tests.
26098 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
26101 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
26102 gen_add3_insn did not fail.
26103 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
26104 r0, construct that number in a temporary reg and add that reg to r0.
26105 If asked to put the result in r0 as well, fail.
26107 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
26109 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26110 for early expansion of vec_eqv.
26112 2017-06-08 Jakub Jelinek <jakub@redhat.com>
26114 PR middle-end/81005
26115 * ubsan.c (instrument_null): Avoid pointless code temporary.
26116 (pass_ubsan::execute): Instrument aggregate arguments of calls.
26118 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
26122 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
26125 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
26128 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26130 * predict.c (unlikely_executed_edge_p): Move ahead.
26131 (probably_never_executed_edge_p): Use it.
26133 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26135 PR middle-end/79988
26136 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
26137 gimple_call_builtin_p call.
26139 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26141 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
26142 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
26143 rtl_check_failed_type2, rtl_check_failed_code1,
26144 rtl_check_failed_code2, rtl_check_failed_code_mode,
26145 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
26146 rtvec_check_failed_bounds, rtl_check_failed_flag,
26147 _fatal_insn_not_found, _fatal_insn): Likewise.
26148 * tree.h (tree_contains_struct_check_failed,
26149 tree_check_failed, tree_not_check_failed,
26150 tree_class_check_failed, tree_range_check_failed,
26151 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
26152 tree_vec_elt_check_failed, phi_node_elt_check_failed,
26153 tree_operand_check_failed, omp_clause_check_failed,
26154 omp_clause_operand_check_failed, omp_clause_range_check_failed):
26157 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26159 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
26160 flag_branch_probabilities.
26161 * ipa-inline.c (edge_badness): Likewise.
26162 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
26163 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
26164 * predict.c (maybe_hot_frequency_p): Likewise.
26165 (probably_never_executed): Likewise.
26166 * sched-ebb.c (schedule_ebbs): Likewise.
26167 * sched-rgn.c (find_single_block_region): Likewise.
26168 * tracer.c (tail_duplicate): Likewise.
26170 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26172 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
26173 longer requires x_flag_profile_use.
26175 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26177 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
26178 instead of flag_reorder_blocks_and_partition.
26179 * dbxout.c (dbxout_function_end): Likewise.
26180 * dwarf2out.c (gen_subprogram_die): Likewise.
26181 * haifa-sched.c (sched_create_recovery_edges): Likewise.
26182 * hw-doloop.c (reorg_loops): Likewise.
26183 * varasm.c (assemble_start_function,
26184 assemble_end_function): Likewise.
26185 (decide_function_section): Do not check for
26186 flag_reorder_blocks_and_partition.
26188 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26190 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26192 (chkp_get_hard_register_fake_addr_expr): Ditto.
26193 (chkp_build_addr_expr): Add check for hard reg case.
26194 (chkp_parse_array_and_component_ref): Ditto.
26195 (chkp_find_bounds_1): Ditto.
26196 (chkp_process_stmt): Don't generate bounds store for
26199 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26201 * predict.c (maybe_hot_bb_p): Do not check profile status.
26202 (maybe_hot_edge_p): Likewise.
26203 (probably_never_executed): Check for zero counts even if profile
26205 (unlikely_executed_edge_p): New function.
26206 (unlikely_executed_stmt_p): New function.
26207 (unlikely_executed_bb_p): New function.
26208 (set_even_probabilities): Use unlikely predicates.
26209 (combine_predictions_for_bb): Likewise.
26210 (predict_paths_for_bb): Likewise.
26211 (predict_paths_leading_to_edge): Likewise.
26212 (determine_unlikely_bbs): New function.
26213 (estimate_bb_frequencies): Use it.
26214 (compute_function_frequency): Use zero counts even if profile is
26216 * profile-count.h: Fix typo.
26218 2017-08-08 Julia Koval <julia.koval@intel.com>
26220 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
26221 _mm512_mask_cvtsepi16_storeu_epi8,
26222 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
26223 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
26224 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
26225 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
26226 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
26227 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
26228 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
26229 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
26230 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
26231 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
26232 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
26233 __builtin_ia32_pmovuswb256mem_mask,
26234 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
26235 __builtin_ia32_pmovwb512mem_mask): New builtins.
26237 2017-08-08 Julia Koval <julia.koval@intel.com>
26239 PR target/73350,80862
26240 * config/i386/subst.md (round): Fix round pattern.
26241 * config/i386/i386.c (ix86_erase_embedded_rounding):
26242 Fix erasing rounding for the fixed pattern.
26244 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26246 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
26248 2017-06-08 Martin Liska <mliska@suse.cz>
26250 PR gcov-profile/80911
26251 * gcov.c (block_info::block_info): New constructor.
26253 2017-06-07 Carl Love <cel@us.ibm.com>
26255 * config/rs6000/rs6000-c: The return type of the following
26256 built-in functions was implemented as int not long long. Fix sign
26257 of return value for the unsigned version of vec_mulo and vec_mule.
26258 vector unsigned long long vec_bperm (vector unsigned long long,
26259 vector unsigned char)
26260 vector signed long long vec_mule (vector signed int,
26262 vector unsigned long long vec_mule (vector unsigned int,
26263 vector unsigned int)
26264 vector signed long long vec_mulo (vector signed int,
26266 vector unsigned long long vec_mulo (vector unsigned int,
26267 vector unsigned int)
26268 * doc/extend.texi: Fix the documentation for the built-in
26271 2017-06-07 Carl Love <cel@us.ibm.com>
26274 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
26277 2017-06-07 Carl Love <cel@us.ibm.com>
26279 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
26280 support, Generate doublehv for signed int/float for BE case only.
26282 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
26284 * doc/invoke.texi (mcx16): Rewrite.
26286 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26288 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
26289 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
26290 *mov<mode>_softfloat, and an anonymous splitter): Use
26291 nonimmediate_operand instead of rs6000_nonimmediate_operand.
26293 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26295 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
26297 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26298 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
26299 (rs6000_debug_reg_global): Adjust.
26300 (rs6000_init_hard_regno_mode_ok): Adjust.
26301 (rs6000_dbx_register_number): Adjust.
26302 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
26303 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26304 Remove SPE_ACC and SPEFSCR.
26305 (REG_ALLOC_ORDER): Ditto.
26306 (FRAME_POINTER_REGNUM): Change to 111.
26307 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
26308 (REG_CLASS_NAMES): Ditto.
26309 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
26310 (REGISTER_NAMES): Ditto.
26311 (ADDITIONAL_REG_NAMES): Ditto.
26312 (rs6000_reg_names): Ditto.
26313 * config/rs6000/rs6000.md: Renumber some register number
26316 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26318 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
26320 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26321 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
26323 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
26324 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26325 Delete the SPE high registers.
26326 (REG_ALLOC_ORDER): Ditto.
26327 (enum reg_class): Remove SPE_HIGH_REGS.
26328 (REG_CLASS_NAMES): Ditto.
26329 (REG_CLASS_CONTENTS): Delete the SPE high registers.
26330 (REGISTER_NAMES): Ditto.
26331 (rs6000_reg_names): Ditto.
26332 * doc/tm.texi.in: Remove SPE as example.
26333 * doc/tm.texi: Regenerate.
26335 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26337 * config/rs6000/8540.md (ppc8540_brinc): Delete.
26338 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
26339 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
26340 * config/rs6000/rs6000.md (type): Remove "brinc".
26342 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26344 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
26345 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
26346 * config/rs6000/linuxspe.h: Delete file.
26347 * config/rs6000/rs6000.md: Don't include spe.md.
26348 * config/rs6000/spe.h: Delete file.
26349 * config/rs6000/spe.md: Delete file.
26350 * config/rs6000/t-rs6000: Remove spe.md.
26352 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26354 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
26355 (reg_or_none500mem_operand): Delete.
26356 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
26357 instead of reg_or_none500mem_operand.
26359 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26361 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26362 handling of SPE flags.
26363 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
26365 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26367 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
26369 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
26370 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
26371 paired_divv2sf3): Similar.
26372 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
26373 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
26374 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
26376 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
26377 Rename the paired_* instruction patterns.
26378 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
26380 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
26381 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
26382 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
26383 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
26384 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
26385 PAIRED_VECTOR_MODE.
26386 (struct machine_function): Delete field spe_insn_chain_scanned_p.
26387 (spe_func_has_64bit_regs_p): Delete.
26388 (spe_expand_predicate_builtin): Delete.
26389 (spe_expand_evsel_builtin): Delete.
26390 (TARGET_DWARF_REGISTER_SPAN): Do not define.
26391 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
26392 (invalid_e500_subreg): Delete.
26393 (rs6000_legitimize_address): Always force_reg op2 as well, for
26394 paired single memory accesses.
26395 (rs6000_member_type_forces_blk): Delete.
26396 (rs6000_spe_function_arg): Delete.
26397 (rs6000_expand_unop_builtin): Delete SPE handling.
26398 (rs6000_expand_binop_builtin): Ditto.
26399 (spe_expand_stv_builtin): Delete.
26400 (bdesc_2arg_spe): Delete.
26401 (spe_expand_builtin): Delete.
26402 (spe_expand_predicate_builtin): Delete.
26403 (spe_expand_evsel_builtin): Delete.
26404 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
26405 (spe_init_builtins): Delete.
26406 (spe_func_has_64bit_regs_p): Delete.
26407 (savres_routine_name): Delete "info" parameter. Adjust callers.
26408 (rs6000_emit_stack_reset): Ditto.
26409 (rs6000_dwarf_register_span): Delete.
26410 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
26411 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
26412 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
26414 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
26416 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
26417 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
26418 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
26419 mulv2sf3, divv2sf3): Delete expanders.
26421 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26423 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
26425 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26427 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
26428 * config/rs6000/rs6000.c: Ditto.
26430 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26432 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
26433 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
26434 comparison_operator.
26436 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26438 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
26439 * config/rs6000/rs6000.opt: Ditto.
26440 * config/rs6000/t-rtems: Ditto.
26442 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26444 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
26445 TARGET_E500_SINGLE by 0, simplify.
26446 * config/rs6000/rs6000.c: Ditto.
26447 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
26448 (spe_build_register_parallel): Delete.
26449 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
26450 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
26451 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
26452 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
26453 (E500_CONVERT): Delete.
26454 * config/rs6000/spe.md: Remove many patterns and all define_constants.
26456 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26458 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
26459 * config/rs6000/dfp.md: Ditto.
26460 (negdd2, *negdd2_fpr): Merge.
26461 (absdd2, *absdd2_fpr): Merge.
26462 (negtd2, *negtd2_fpr): Merge.
26463 (abstd2, *abstd2_fpr): Merge.
26464 * config/rs6000/e500.h: Delete file.
26465 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
26466 TARGET_FPRS by 1 and simplify.
26467 * config/rs6000/rs6000-c.c: Ditto.
26468 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
26469 TARGET_DF_SPE by 0.
26470 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
26472 * config/rs6000/rs6000.md: Ditto.
26473 (floatdidf2, *floatdidf2_fpr): Merge.
26474 (move_from_CR_gt_bit): Delete.
26475 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
26476 (E500_CR_IOR_COMPARE): Delete.
26477 (All patterns that require !TARGET_FPRS): Delete.
26478 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
26480 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26482 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
26484 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26486 * graphds.c (add_edge): Intitialize edge's attached data.
26487 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
26488 pointer parameter. Call pointed function on each edge during
26489 graph traversing. Skip traversing the edge when the function
26491 (graphds_dfs, graphds_scc): Ditto.
26492 (for_each_edge): New parameter. Pass the new parameter to callback
26494 * graphds.h (skip_edge_callback): New function pointer type.
26495 (graphds_dfs, graphds_scc): New function pointer parameter.
26496 (graphds_edge_callback, for_each_edge): New parameter.
26498 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26500 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
26501 out code checking if runtime alias check is possible to below ...
26502 Call the new function.
26503 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
26504 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
26506 2017-06-07 Marek Polacek <polacek@redhat.com>
26509 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
26510 TYPE_OVERFLOW_WRAPS check.
26512 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26514 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
26515 if versioning is required.
26516 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
26517 peeling with the check for versioning.
26519 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26521 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
26522 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
26523 Set true to new parameter if new ssa variable is defined.
26524 (vect_gen_vector_loop_niters): Refactor. Set range information
26525 for the new vector loop bound variable.
26526 (vect_do_peeling): Ditto.
26528 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26530 * tree-affine.c (ssa.h): Include header file.
26531 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
26532 has wrapping overflow behavior.
26534 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26536 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
26538 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26540 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
26541 (tree_to_aff_combination): ... here.
26543 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26545 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
26546 reg_pressure model function.
26547 (ivopts_global_cost_for_size): Delete.
26548 (determine_set_costs, iv_ca_recount_cost): Call new model function
26549 ivopts_estimate_reg_pressure.
26551 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26553 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
26554 expensive than udiv. Remove floating point cases from mod.
26556 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26558 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
26559 Increase idiv cost.
26561 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26563 * config/aarch64/aarch64.md
26564 (copysignsf3): Fix mask generation.
26566 2017-06-07 Jakub Jelinek <jakub@redhat.com>
26568 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
26570 (class dump_manager): Add register_dumps method.
26571 * dumpfile.c: Include langhooks.h.
26572 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
26573 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
26574 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
26575 (dump_manager::dump_register): Start with 512 entries instead of 32.
26576 (dump_manager::register_dumps): New method.
26577 * toplev.c (general_init): Instead of invoking register_dumps
26578 langhook, invoke register_dumps method on the dump manager.
26579 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
26582 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
26584 * doc/md.texi: Clarify the restrictions on a define_insn condition.
26585 Say that # requires an associated define_split to exist, and that
26586 the define_split must be suitable for use after register allocation.
26588 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26590 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
26591 (compute_outgoing_frequencies): Also initialize zero counts.
26592 (find_many_sub_basic_blocks): Do not produce uninitialized profile
26593 around loops; preserve more of profile when nothing changes.
26595 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
26597 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
26599 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
26600 * config/arm/arm-cpu-cdata.h: Regenerate.
26601 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
26602 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
26603 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
26604 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
26605 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
26607 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
26609 * config/arm/t-rmprofile: Likewise.
26610 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
26612 2017-06-06 David S. Miller <davem@davemloft.net>
26615 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
26616 blockage if function uses alloca.
26618 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26620 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
26621 New "uid" fields to hold pretty-print IDs of group and ref.
26622 Memory references are now identified as <group_id>:<ref_id>
26623 instead of using [random] addresses.
26624 (dump_mem_details): Simplify, no functional change.
26625 (dump_mem_ref): Simplify and make output more concise.
26626 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
26627 (find_or_create_group): Initialize group uid.
26628 (record_ref): Initialize ref uid. Improve debug output.
26629 (prune_group_by_reuse, should_issue_prefetch_p,)
26630 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
26631 (mark_nontemporal_store, determine_loop_nest_reuse):
26632 Improve debug output.
26634 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26636 * dbgcnt.def (prefetch): New debug counter.
26637 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
26638 (schedule_prefetches): Stop issueing prefetches if debug counter
26641 2017-06-06 Tom de Vries <tom@codesourcery.com>
26643 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
26644 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
26646 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26648 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
26649 Use aarch64_reg_or_zero predicate for operand 4.
26650 (aarch64_compare_and_swap<mode> define_insn_and_split):
26651 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
26652 (aarch64_store_exclusive<mode>): Likewise for operand 2.
26654 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
26656 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
26657 (arm_compute_save_core_reg_mask): This.
26658 (thumb1_compute_save_reg_mask): Rename into ...
26659 (thumb1_compute_save_core_reg_mask): This.
26660 (arm_compute_save_reg0_reg12_mask): Adapt comment.
26661 (arm_compute_frame_layout): Likewise.
26663 2017-06-06 Richard Biener <rguenther@suse.de>
26665 PR tree-optimization/80974
26666 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
26667 keep or clear leaders SSA info.
26669 2017-06-06 Tom de Vries <tom@codesourcery.com>
26671 * config/nvptx/nvptx.c (split_mode_p): New function.
26672 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
26674 2017-06-06 Tom de Vries <tom@codesourcery.com>
26676 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
26678 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26681 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
26684 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26686 * shrink-wrap.c (handle_simple_exit): Update profile.
26687 (try_shrink_wrapping): Upate profile.
26689 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26691 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
26692 (tree_guess_outgoing_edge_probabilities): New.
26693 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
26694 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
26696 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26698 * ipa-split.c (split_function): Initialize return bb profile.
26700 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26702 * profile.c (compute_branch_probabilities): Also initialize
26703 EXIT_BLOCK profile.
26705 2017-06-06 Richard Biener <rguenther@suse.de>
26707 PR tree-optimization/80928
26708 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
26709 (vect_analyze_loop_operations): Properly guard analysis for
26711 (vect_transform_loop): Likewise.
26712 (vect_analyze_loop_2): Also reset SLP type on PHIs.
26713 (vect_model_induction_cost): Do not cost for pure SLP.
26714 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
26715 of induction in inner loop vectorization.
26716 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
26717 (vect_get_and_check_slp_defs): Handle vect_induction_def.
26718 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
26720 (vect_analyze_slp_cost_1): Cost induction.
26721 (vect_detect_hybrid_slp_stmts): Handle PHIs.
26722 (vect_get_slp_vect_defs): Likewise.
26723 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
26724 (vect_transform_stmt): Handle SLP reductions.
26725 * tree-vectorizer.h (vectorizable_induction): Adjust.
26727 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26729 * config/rs6000/rs6000.c (make_resolver_func): Update
26730 init_lowered_empty_function call.
26732 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
26734 * doc/invoke.texi: Document the -fprofile-abs-path option.
26735 * common.opt (fprofile-abs-path): New option.
26736 * gcov-io.h (gcov_write_filename): Declare.
26737 * gcov-io.c (gcov_write_filename): New function.
26738 * coverage.c (coverage_begin_function): Use gcov_write_filename.
26739 * profile.c (output_location): Likewise.
26741 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26743 * shring-wrap.c: Revert accidental commit.
26745 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
26747 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
26749 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26751 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
26753 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
26754 profile in callgraph edge.
26755 * profile-count.h (apply_probability): If THIS is 0, then result is 0
26756 (apply_scale): Likewise.
26757 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
26758 Also scale profile when inlining function with zero profile.
26759 (initialize_cfun): Update exit block profile even when it is zero.
26760 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
26761 when profile is read.
26763 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26765 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
26766 (CLONE_*): New constants to define the processors we can generate
26767 code for with the target_clone attribute.
26768 (rs6000_clone_map): New array to identify which clone processors
26769 the current program is running on.
26770 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
26771 target_clone attribute.
26772 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
26773 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
26774 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
26775 (cpu_expand_builtin): Add support for target_clone attribute.
26776 (rs6000_valid_attribute_p): Allow "default" attribute.
26777 (get_decl_name): New debug function to simplify printing the
26778 current function name in debugging statements.
26779 (rs6000_clone_priority): New functions to support the target_clone
26780 attribute, and be able to generate code to switch between ISA 2.05
26781 through ISA 3.0 (power6 through power9).
26782 (rs6000_compare_version_priority): Likewise.
26783 (rs6000_get_function_versions_dispatcher): Likewise.
26784 (make_resolver_func): Likewise.
26785 (add_condition_to_bb): Likewise.
26786 (dispatch_function_versions): Likewise.
26787 (rs6000_generate_version_dispatcher_body): Likewise.
26788 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
26789 (fusion_gpr_load_p): Fix a spacing issue.
26790 * doc/extend.texi (Common Function Attributes): Document that the
26791 PowerPC supports the target_clone attribute.
26793 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26795 * config/arm/arm.h: explain F symbol found in description of ARM
26796 register allocation in its legend.
26798 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26800 * config/mips/frame-header-opt.c: Include profile-count.h.
26801 * config/riscv/riscv.c: Include profile-count.h
26803 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26805 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
26807 (sm_set_flag_if_changed): Add bbs field.
26808 (execute_sm_if_changed_flag_set): Pass BBS.
26809 (execute_sm): Update.
26811 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26813 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
26816 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26818 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
26819 (peephole2): New peephole2 to emit the above.
26820 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
26822 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26824 * config/aarch64/aarch64.c (define_peephole2 above
26825 *sub_<shift>_<mode>): New peephole.
26827 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26829 * config/i386/i386.c (make_resolver_func): Update.
26830 * Makefile.in: Add profile-count.h and profile-count.o
26831 * auto-profile.c (afdo_indirect_call): Update to new API.
26832 (afdo_set_bb_count): Update.
26833 (afdo_propagate_edge): Update.
26834 (afdo_propagate_circuit): Update.
26835 (afdo_calculate_branch_prob): Update.
26836 (afdo_annotate_cfg): Update.
26837 * basic-block.h: Include profile-count.h
26838 (struct edge_def): Turn count to profile_count.
26839 (struct basic_block_def): Likewie.
26840 (REG_BR_PROB_BASE): Move to profile-count.h
26841 (RDIV): Move to profile-count.h
26842 * bb-reorder.c (max_entry_count): Turn to profile_count.
26843 (find_traces): Update.
26844 (rotate_loop):Update.
26845 (connect_traces):Update.
26846 (sanitize_hot_paths):Update.
26847 * bt-load.c (migrate_btr_defs): Update.
26848 * cfg.c (RDIV): Remove.
26849 (init_flow): Use alloc_block.
26850 (alloc_block): Uninitialize count.
26851 (unchecked_make_edge): Uninitialize count.
26852 (check_bb_profile): Update.
26853 (dump_edge_info): Update.
26854 (dump_bb_info): Update.
26855 (update_bb_profile_for_threading): Update.
26856 (scale_bbs_frequencies_int): Update.
26857 (scale_bbs_frequencies_gcov_type): Update.
26858 (scale_bbs_frequencies_profile_count): New.
26859 * cfg.h (update_bb_profile_for_threading): Update.
26860 (scale_bbs_frequencies_profile_count): Declare.
26861 * cfgbuild.c (compute_outgoing_frequencies): Update.
26862 (find_many_sub_basic_blocks): Update.
26863 * cfgcleanup.c (try_forward_edges): Update.
26864 (try_crossjump_to_edge): Update.
26865 * cfgexpand.c (expand_gimple_tailcall): Update.
26866 (construct_exit_block): Update.
26867 * cfghooks.c (verify_flow_info): Update.
26868 (dump_bb_for_graph): Update.
26869 (split_edge): Update.
26870 (make_forwarder_block): Update.
26871 (duplicate_block): Update.
26872 (account_profile_record): Update.
26873 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
26874 (get_estimated_loop_iterations): Update.
26875 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
26876 (single_likely_exit): Update.
26877 * cfgloopmanip.c (scale_loop_profile): Update.
26879 (set_zero_probability): Update.
26880 (lv_adjust_loop_entry_edge): Update.
26881 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
26882 (purge_dead_edges): Update.
26883 (rtl_account_profile_record): Update.
26884 * cgraph.c (cgraph_node::create): Uninitialize count.
26885 (symbol_table::create_edge): Uninitialize count.
26886 (cgraph_update_edges_for_call_stmt_node): Update.
26887 (cgraph_edge::dump_edge_flags): Update.
26888 (cgraph_node::dump): Update.
26889 (cgraph_edge::maybe_hot_p): Update.
26890 * cgraph.h: Include profile-count.h
26891 (create_clone), create_edge, create_indirect_edge): Update.
26892 (cgraph_node): Turn count to profile_count.
26893 (cgraph_edge0: Likewise.
26894 (make_speculative, clone): Update.
26895 (create_edge): Update.
26896 (init_lowered_empty_function): Update.
26897 * cgraphclones.c (cgraph_edge::clone): Update.
26898 (duplicate_thunk_for_node): Update.
26899 (cgraph_node::create_clone): Update.
26900 * cgraphunit.c (cgraph_node::analyze): Update.
26901 (cgraph_node::expand_thunk): Update.
26902 * final.c (dump_basic_block_info): Update.
26903 * gimple-streamer-in.c (input_bb): Update.
26904 * gimple-streamer-out.c (output_bb): Update.
26905 * graphite.c (print_global_statistics): Update.
26906 (print_graphite_scop_statistics): Update.
26907 * hsa-brig.c: Include basic-block.h.
26908 * hsa-dump.c: Include basic-block.h.
26909 * hsa-gen.c (T sum_slice): Update.
26910 (convert_switch_statements):Update.
26911 * hsa-regalloc.c: Include basic-block.h.
26912 * ipa-chkp.c (chkp_produce_thunks): Update.
26913 * ipa-cp.c (struct caller_statistics): Update.
26914 (init_caller_stats): Update.
26915 (gather_caller_stats): Update.
26916 (ipcp_cloning_candidate_p): Update.
26917 (good_cloning_opportunity_p): Update.
26918 (get_info_about_necessary_edges): Update.
26919 (dump_profile_updates): Update.
26920 (update_profiling_info): Update.
26921 (update_specialized_profile): Update.
26922 (perhaps_add_new_callers): Update.
26923 (decide_about_value): Update.
26924 (ipa_cp_c_finalize): Update.
26925 * ipa-devirt.c (struct odr_type_warn_count): Update.
26926 (struct decl_warn_count): Update.
26927 (struct final_warning_record): Update.
26928 (possible_polymorphic_call_targets): Update.
26929 (ipa_devirt): Update.
26930 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26931 * ipa-icf.c (sem_function::merge): Update.
26932 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26933 * ipa-inline.c (compute_uninlined_call_time): Update.
26934 (compute_inlined_call_time): Update.
26935 (want_inline_small_function_p): Update.
26936 (want_inline_self_recursive_call_p): Update.
26937 (edge_badness): Update.
26938 (lookup_recursive_calls): Update.
26939 (recursive_inlining): Update.
26940 (inline_small_functions): Update.
26941 (dump_overall_stats): Update.
26942 (dump_inline_stats): Update.
26943 * ipa-profile.c (ipa_profile_generate_summary): Update.
26944 (ipa_propagate_frequency): Update.
26945 (ipa_profile): Update.
26946 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26947 * ipa-utils.c (ipa_merge_profiles): Update.
26948 * loop-doloop.c (doloop_modify): Update.
26949 * loop-unroll.c (report_unroll): Update.
26950 (unroll_loop_runtime_iterations): Update.
26951 * lto-cgraph.c (lto_output_edge): Update.
26952 (lto_output_node): Update.
26953 (input_node): Update.
26954 (input_edge): Update.
26955 (merge_profile_summaries): Update.
26956 * lto-streamer-in.c (input_cfg): Update.
26957 * lto-streamer-out.c (output_cfg): Update.
26958 * mcf.c (create_fixup_graph): Update.
26959 (adjust_cfg_counts): Update.
26960 (sum_edge_counts): Update.
26961 * modulo-sched.c (sms_schedule): Update.
26962 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26963 * predict.c (maybe_hot_count_p): Update.
26964 (probably_never_executed): Update.
26965 (dump_prediction): Update.
26966 (combine_predictions_for_bb): Update.
26967 (propagate_freq): Update.
26968 (handle_missing_profiles): Update.
26969 (counts_to_freqs): Update.
26970 (rebuild_frequencies): Update.
26971 (force_edge_cold): Update.
26972 * predict.h: Include profile-count.h
26973 (maybe_hot_count_p, counts_to_freqs): UPdate.
26974 * print-rtl-function.c: Do not include cfg.h
26975 * print-rtl.c: Include basic-block.h
26976 * profile-count.c: New file.
26977 * profile-count.h: New file.
26978 * profile.c (is_edge_inconsistent): Update.
26979 (correct_negative_edge_counts): Update.
26980 (is_inconsistent): Update.
26981 (set_bb_counts): Update.
26982 (read_profile_edge_counts): Update.
26983 (compute_frequency_overlap): Update.
26984 (compute_branch_probabilities): Update; Initialize and deinitialize
26986 (branch_prob): Update.
26987 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26988 (edge_gcov_count): New.
26989 (bb_gcov_count): New.
26990 * shrink-wrap.c (try_shrink_wrapping): Update.
26991 * tracer.c (better_p): Update.
26992 * trans-mem.c (expand_transaction): Update.
26993 (ipa_tm_insert_irr_call): Update.
26994 (ipa_tm_insert_gettmclone_call): Update.
26995 * tree-call-cdce.c: Update.
26996 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26997 (gimple_duplicate_sese_tail): Update.
26998 (gimple_account_profile_record): Update.
26999 (execute_fixup_cfg): Update.
27000 * tree-inline.c (copy_bb): Update.
27001 (copy_edges_for_bb): Update.
27002 (initialize_cfun): Update.
27003 (freqs_to_counts): Update.
27004 (copy_cfg_body): Update.
27005 (expand_call_inline): Update.
27006 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
27007 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
27008 (try_unroll_loop_completely): Update.
27009 (try_peel_loop): Update.
27010 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
27011 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
27012 * tree-ssa-loop-split.c (connect_loops): Update.
27013 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
27014 * tree-ssa-reassoc.c (branch_fixup): Update.
27015 * tree-ssa-tail-merge.c (replace_block_by): Update.
27016 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
27017 (compute_path_counts): Update.
27018 (update_profile): Update.
27019 (recompute_probabilities): Update.
27020 (update_joiner_offpath_counts): Update.
27021 (estimated_freqs_path): Update.
27022 (freqs_to_counts_path): Update.
27023 (clear_counts_path): Update.
27024 (ssa_fix_duplicate_block_edges): Update.
27025 (duplicate_thread_path): Update.
27026 * tree-switch-conversion.c (case_bit_test_cmp): Update.
27027 (struct switch_conv_info): Update.
27028 * tree-tailcall.c (decrease_profile): Update.
27029 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
27030 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
27031 * value-prof.c (check_counter): Update.
27032 (gimple_divmod_fixed_value): Update.
27033 (gimple_mod_pow2): Update.
27034 (gimple_mod_subtract): Update.
27035 (gimple_ic_transform): Update.
27036 (gimple_stringop_fixed_value): Update.
27037 * value-prof.h (gimple_ic): Update.
27039 2017-06-02 Carl Love <cel@us.ibm.com>
27041 * config/rs6000/rs6000-c: Add support for built-in functions
27042 vector double vec_doublee (vector signed int);
27043 vector double vec_doublee (vector unsigned int);
27044 vector double vec_doublee (vector float);
27045 vector double vec_doubleh (vector signed int);
27046 vector double vec_doubleh (vector unsigned int);
27047 vector double vec_doubleh (vector float);
27048 vector double vec_doublel (vector signed int);
27049 vector double vec_doublel (vector unsigned int);
27050 vector double vec_doublel (vector float);
27051 vector double vec_doubleo (vector signed int);
27052 vector double vec_doubleo (vector unsigned int);
27053 vector double vec_doubleo (vector float);.
27054 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
27055 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
27057 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
27058 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
27059 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
27061 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
27062 vec_doublel, vec_doubleh.
27063 * doc/extend.texi: Update the built-in documentation file for the
27064 new built-in functions.
27066 2017-06-02 David Malcolm <dmalcolm@redhat.com>
27069 * ipa-inline-analysis.c (free_growth_caches): Set
27070 edge_removal_hook_holder to NULL after removing it.
27072 2017-06-02 Sudakshina Das <sudi.das@arm.com>
27074 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
27075 comparision with zero.
27077 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27078 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27079 for early expansion of vec_min and vec_max builtins.
27080 (builtin_function_type): Add min/max unsigned variants to those
27081 identified as having unsigned arguments.
27083 2017-06-02 Olivier Hainque <hainque@adacore.com>
27085 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
27087 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27089 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
27090 Use VALL_F16 iterator rather than VALL.
27092 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27094 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
27095 Emit CBNZ inside loop when doing a strong exchange and comparing
27096 against zero. Generate the CC flags after the loop.
27098 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
27100 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
27101 (dl_section_ref): New.
27102 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
27103 On AIX, append an expression to subtract the size of the
27104 section length to dl_section_ref.
27106 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27108 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27109 for early expansion of vector absolute builtins.
27111 2017-06-02 Richard Biener <rguenther@suse.de>
27113 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
27114 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
27116 2017-06-02 Richard Biener <rguenther@suse.de>
27118 PR tree-optimization/80948
27119 * tree-tailcall.c (find_tail_calls): Track stmts to move in
27120 stmt order as well.
27122 2017-06-02 Richard Biener <rguenther@suse.de>
27124 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
27126 * tree-vect-stmts.c (process_use): Do not mark backedge defs
27127 for inductions as relevant.
27129 2017-06-02 Richard Biener <rguenther@suse.de>
27131 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
27132 (vectorizable_induction): ... this. Remove dead code.
27134 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
27136 * builtins. (expand_builtin_alloca): Remove second parameter and
27137 infer its value from the first parameter instead.
27138 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
27140 2017-06-02 Jakub Jelinek <jakub@redhat.com>
27142 PR rtl-optimization/80903
27143 * loop-doloop.c (add_test): Unshare sequence.
27145 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
27147 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
27149 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
27151 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
27153 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
27154 xlogue_layout::get_instance, logue_layout::xlogue_layout,
27155 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
27156 (xlogue_layout::get_stub_rtx): Make static.
27157 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
27158 (xlogue_layout::compute_stub_managed_regs): Rename to...
27159 (xlogue_layout::count_stub_managed_regs): ...this.
27160 (xlogue_layout::is_stub_managed_reg): New function.
27161 (xlogue_layout::m_stub_names): Rename to...
27162 (xlogue_layout::s_stub_names): ...this, make static.
27163 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
27164 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
27165 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
27166 xlogue_layout::s_stub_names): Instantiate statics.
27167 (stub_managed_regs): Remove.
27168 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
27169 (disable_call_ms2sysv_xlogues): Rename to...
27170 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
27171 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
27173 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
27174 change after reload_completed.
27175 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
27177 (ix86_expand_prologue): Likewise.
27178 (ix86_expand_epilogue): Likewise.
27179 (ix86_expand_split_stack_prologue): Likewise.
27180 (ix86_compute_frame_layout): Remove frame parameter ...
27181 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
27182 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
27184 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
27185 (ix86_frame): Move from here ...
27186 * config/i386/i386.h (ix86_frame): ... to here.
27187 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
27188 complete ix86_frame data structure instead. Remove some_ld_name.
27190 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27192 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
27193 symbols that hold a DECL_VALUE_EXPR.
27195 2017-06-01 Martin Jambor <mjambor@suse.cz>
27197 PR tree-optimization/80898
27198 * tree-sra.c (process_subtree_disqualification): Removed.
27199 (disqualify_candidate): Do not acll
27200 process_subtree_disqualification.
27201 (subtree_mark_written_and_enqueue): New function.
27202 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
27203 RHS has been disqualified and re-queue LHS if necessary. Apart
27204 from that, ignore disqualified RHS.
27206 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27208 * config/s390/s390.c (s390_emit_epilogue): Disable early return
27209 address fetch for z10 or later.
27211 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27213 * config/arc/arc.md (tst_movb): Add guard when splitting.
27215 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27217 * config/arc/arc.c (arc_can_eliminate): Test against
27218 arc_frame_pointer_needed.
27220 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27222 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
27223 to prevent store reordering.
27224 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
27225 (type): Add block type.
27226 (stack_tie): Define special instruction to be used in
27229 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27231 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
27232 constraint. It is not valid for the pattern.
27233 (noncommutative_binary_comparison): Likewise.
27235 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27237 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
27240 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27242 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
27243 be used by the reg-alloc.
27245 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27247 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
27248 reg-alloc when having mul64 or mul32x16 instructions.
27249 (mulsidi3): Likewise.
27250 (umulsidi3): Likewise.
27251 (mulsi32x16): New pattern.
27252 (mulsi64): Likewise.
27253 (mulsidi64): Likewise.
27254 (umulsidi64): Likewise.
27255 (MUL32x16_REG): Define.
27256 (mul64_600): Use MUL32x16_REG.
27257 (mac64_600): Likewise.
27258 (umul64_600): Likewise.
27259 (umac64_600): Likewise.
27261 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27263 * config/arc/arc.md (mulsi3_700): Make it commutative.
27265 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
27267 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
27269 (*sign_extendsidi2_insn): Likewise for movstosw.
27271 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27273 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
27274 the type of the input discriminant value. Convert the
27275 discriminant value of signedness vary.
27277 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
27279 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
27280 Add to -Wall section.
27282 2017-06-01 Richard Biener <rguenther@suse.de>
27284 PR middle-end/66313
27285 * fold-const.c (fold_plusminus_mult_expr): If the factored
27286 factor may be zero use a wrapping type for the inner operation.
27287 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
27288 and handle moved defs.
27289 (process_assignment): Properly guard the unary op case. Return a
27290 tri-state indicating that moving the stmt before the call may allow
27291 to continue. Pass through to_move.
27292 (find_tail_calls): Handle moving unrelated defs before
27295 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
27298 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
27299 splitter result in the canonical way.
27301 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
27303 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
27304 also for 32bit target. Update insn attributes.
27305 (zero-extendsidi2 splitter): Allow all registers for operand 1.
27307 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
27309 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
27310 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
27311 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
27312 (_mm_maskz_min_ss): New intrinsics.
27314 2017-05-31 Martin Liska <mliska@suse.cz>
27316 * tree-vect-loop.c (vect_create_epilog_for_reduction):
27317 Change comment style to one we normally use.
27318 (vectorizable_reduction): Likewise.
27319 (vectorizable_induction): Likewise.
27320 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
27321 (vectorizable_call): Likewise.
27322 (vectorizable_simd_clone_call): Likewise.
27323 (vectorizable_conversion): Likewise.
27324 (vectorizable_assignment): Likewise.
27325 (vectorizable_shift): Likewise.
27326 (vectorizable_operation): Likewise.
27327 (vectorizable_store): Likewise.
27328 (vectorizable_load): Likewise.
27329 * tree-vectorizer.h: Likewise.
27331 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
27333 * passes.c (emergency_dump_function): New.
27334 * tree-pass.h (emergency_dump_function): Declare.
27335 * plugin.c (plugins_internal_error_function): Remove.
27336 * plugin.h (plugins_internal_error_function): Remove declaration.
27337 * toplev.c (internal_error_function): New static function. Use it...
27338 (general_init): ...here.
27340 2017-05-31 Graham Markall <graham.markall@embecosm.com>
27342 * config/arc/arc.c (arc_print_operand): Handle constant operands.
27343 (arc_rtx_costs): Add costs for new patterns.
27344 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
27345 * config/arc/predicates.md: Add _1_2_3_operand predicate.
27347 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27349 * tree-ssa-strlen.c (get_next_strinfo): New function.
27350 (get_stridx_plus_constant): Use it.
27351 (zero_length_string): Likewise.
27352 (adjust_related_strinfos): Likewise.
27353 (adjust_last_stmt): Likewise.
27355 2017-05-31 Richard Biener <rguenther@suse.de>
27358 * config/i386/i386.c (ix86_expand_builtin): Remove assert
27359 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
27361 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27363 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
27364 loop_vinfo argument and use of dependence distance vectors.
27365 Check instead whether the two references differ only in their
27366 initial value and assume that they have the same alignment if the
27367 difference is a multiple of the vector alignment.
27368 (vect_analyze_data_refs_alignment): Update call accordingly.
27370 2017-05-31 Martin Liska <mliska@suse.cz>
27373 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
27375 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27377 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
27378 (create_intersect_range_checks): Move from ...
27379 * tree-data-ref.c (create_intersect_range_checks_index)
27380 (create_intersect_range_checks): ... to here.
27381 (create_runtime_alias_checks): New function factored from ...
27382 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
27383 here. Call above function.
27384 * tree-data-ref.h (create_runtime_alias_checks): New function.
27386 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27388 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
27389 segment length for dr_b and compute it in wide_int.
27391 2017-05-31 Richard Biener <rguenther@suse.de>
27393 PR tree-optimization/80906
27394 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
27395 and pass through iv_map.
27396 (copy_bb_and_scalar_dependences): Adjust.
27397 (translate_pending_phi_nodes): Likewise.
27398 (copy_loop_close_phi_args): Handle code-generating IVs instead
27401 2017-05-30 David Malcolm <dmalcolm@redhat.com>
27403 * diagnostic-color.c (color_dict): Add "type-diff".
27404 (parse_gcc_colors): Update comment.
27405 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
27406 -fdiagnostics-show-template-tree and -fno-elide-type.
27407 (GCC_COLORS): Add type-diff to example.
27409 (-fdiagnostics-show-template-tree): New.
27410 (-fno-elide-type): New.
27411 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
27412 the pp_format_decoder callback. Call any m_format_postprocessor's
27414 (pretty_printer::pretty_printer): Initialize
27415 m_format_postprocessor.
27416 (pretty_printer::~pretty_printer): Delete any
27417 m_format_postprocessor.
27418 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
27419 (class format_postprocessor): New class.
27420 (struct pretty_printer::format_decoder): Document the new parameters.
27421 (struct pretty_printer::m_format_postprocessor): New field.
27422 * tree-diagnostic.c (default_tree_printer): Update for new
27423 bool and const char ** params.
27424 * tree-diagnostic.h (default_tree_printer): Likewise.
27426 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
27428 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
27429 (lwa_operand): Delete rs6000_gen_cell_microcode test.
27430 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27431 rs6000_gen_cell_microcode code.
27432 (rs6000_final_prescan_insn): Delete.
27433 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
27434 "warn-cell-microcode" entries.
27435 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
27436 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
27437 throughout. Change cc_reg_not_micro_cr0_operand to
27438 cc_reg_not_cr0_operand throughout.
27439 (*extendhi<mode>2_noload): Delete.
27440 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
27441 (mwarn-cell-microcode): Delete.
27442 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
27443 -mgen-cell-microcode and -mwarn-cell-microcode.
27445 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
27448 * config/i386/constraints.md (Yd): New constraint.
27450 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
27451 and (?Yd, r) alternatives. Update insn attributes.
27452 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
27453 and (?*Yd, r) alternatives. Update insn attributes.
27454 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
27456 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
27458 * gimplify.c (gimplify_modify_expr): Don't create a
27459 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
27462 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
27464 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
27466 2017-05-30 Richard Biener <rguenther@suse.de>
27468 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
27469 and reduc_def fields.
27470 (STMT_VINFO_REDUC_TYPE): New define.
27471 (STMT_VINFO_REDUC_DEF): Likewise.
27472 (vect_force_simple_reduction): Adjust prototype.
27473 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
27474 (vect_is_simple_reduction): Remove check_reduction argument.
27475 (vect_force_simple_reduction): Adjust and set
27476 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27477 (vectorizable_reduction): Do not re-do reduction analysis
27478 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27479 * tree-parloops.c (gather_scalar_reductions): Adjust.
27481 2017-05-30 Richard Biener <rguenther@suse.de>
27483 PR middle-end/80901
27484 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
27487 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
27489 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
27490 Introduce unknown_misalignment parameter and remove vf.
27491 (vect_peeling_hash_get_lowest_cost):
27492 Pass unknown_misalignment parameter.
27493 (vect_enhance_data_refs_alignment):
27494 Fix unsupportable data ref treatment.
27496 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27498 * tree-vect-data-refs.c (vect_get_data_access_cost):
27499 Workaround for SLP handling.
27500 (vect_enhance_data_refs_alignment):
27501 Compute costs for doing no peeling at all, compare to the best
27502 peeling costs so far and avoid peeling if cheaper.
27504 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27506 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
27507 Return peeling info and set costs to zero for unlimited cost
27509 (vect_enhance_data_refs_alignment): Also inspect all datarefs
27510 with unknown misalignment. Compute and costs for unknown
27511 misalignment, compare them to the costs for known misalignment
27512 and choose the cheapest for peeling.
27514 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27516 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
27517 (vect_get_peeling_costs_all_drs): Create function.
27518 (vect_peeling_hash_get_lowest_cost):
27519 Use vect_get_peeling_costs_all_drs.
27520 (vect_peeling_supportable): Create function.
27521 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
27523 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27525 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
27526 DR_HAS_NEGATIVE_STEP.
27527 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
27528 (vect_enhance_data_refs_alignment): Use.
27529 (vect_duplicate_ssa_name_ptr_info): Use.
27530 * tree-vectorizer.h (dr_misalignment): Use.
27531 (known_alignment_for_access_p): Use.
27533 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27536 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
27538 (has_section_name): New function.
27540 2017-05-30 Martin Liska <mliska@suse.cz>
27543 * auto-profile.c (get_function_decl_from_block): Fix
27546 2017-05-30 Richard Biener <rguenther@suse.de>
27548 PR middle-end/80876
27549 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
27551 2017-05-30 Martin Liska <mliska@suse.cz>
27553 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
27554 * dumpfile.h (struct dump_file_info): Remove ctors.
27556 2017-05-30 Martin Liska <mliska@suse.cz>
27558 * predict.def: Fix GNU coding style.
27560 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27562 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27563 Mark 'to' argument with ATTRIBUTE_UNUSED.
27565 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27567 * config/xtensa/xtensa.c (xtensa_emit_call): Use
27568 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
27569 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
27572 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
27574 * doc/install.texi (Options specification): Restore entry of
27575 --enable-sjlj-exceptions.
27577 2017-05-27 Michael Eager <eager@eagercon.com>
27580 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
27582 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
27584 * config/microblaze/microblaze.h
27585 (FIXED_REGISTERS): Update in macro.
27586 (CALL_USED_REGISTERS): Update in macro.
27588 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
27590 * doc/install.texi: Add links to macOS binary distributions.
27592 2017-05-27 Jakub Jelinek <jakub@redhat.com>
27596 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27598 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27600 2017-05-26 Martin Liska <mliska@suse.cz>
27602 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
27604 2017-05-26 Martin Liska <mliska@suse.cz>
27606 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
27607 always leading ';; '.
27608 (dump_bb_info): Likewise.
27609 (brief_dump_cfg): Likewise.
27610 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
27611 * dumpfile.c: Remove usage of TDF_VERBOSE.
27612 * dumpfile.h (enum dump_kind): Likewise.
27613 (dump_gimple_bb_header): Do not use TDF_COMMENT.
27614 * print-tree.c (debug_verbose): Remove.
27615 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
27616 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
27617 * tree-diagnostic.c (default_tree_printer): Replace
27618 TDF_DIAGNOSTIC with TDF_SLIM.
27620 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27622 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
27623 in parameter loop, rather than loop_vinfo.
27624 (create_intersect_range_checks): Ditto.
27625 (vect_create_cond_for_alias_checks): Update call to above functions.
27627 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27629 PR tree-optimization/80815
27630 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
27631 for merging runtime alias checks. Handle negative DR_STEPs.
27633 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27635 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
27637 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
27638 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
27639 out code pruning runtime alias checks.
27640 * tree-data-ref.c (prune_runtime_alias_test_list): New function
27641 factored out from above.
27642 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27644 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27646 (prune_runtime_alias_test_list): New decalaration.
27648 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27650 * tree-vect-data-refs.c (compare_tree): Rename and move ...
27651 * tree-data-ref.c (data_ref_compare_tree): ... to here.
27652 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
27653 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
27654 (operator==, comp_dr_with_seg_len_pair): Ditto.
27655 (vect_prune_runtime_alias_test_list): Ditto.
27657 2017-05-26 Martin Liska <mliska@suse.cz>
27660 * params.def: Bound partial-inlining-entry-probability param.
27662 2017-05-26 Marek Polacek <polacek@redhat.com>
27665 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
27668 2017-05-26 Richard Biener <rguenther@suse.de>
27670 PR tree-optimization/80842
27671 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
27674 2017-05-26 Richard Biener <rguenther@suse.de>
27676 PR tree-optimization/80844
27677 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
27679 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
27681 * doc/md.texi (Machine Constraints): Update x86 family
27682 machine constraints section to match 'config/i386/constraints.md'.
27684 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
27686 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
27688 2017-05-25 Nathan Sidwell <nathan@acm.org>
27690 * doc/invoke.texi (--enable-languages): Update documentation.
27692 2017-05-25 Martin Liska <mliska@suse.cz>
27694 * dumpfile.c: Add TDF_FOLDING.
27695 * dumpfile.h (enum dump_kind): Likewise.
27696 * genmatch.c (dt_simplify::gen_1): Use it.
27698 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27700 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
27702 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27704 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27705 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
27707 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27709 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
27710 * match.pd (X == C): Rewrite it here.
27711 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
27712 with_certain_nonzero_bits2): New predicates.
27713 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
27715 2017-05-24 Nathan Sidwell <nathan@acm.org>
27717 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
27720 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
27723 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27725 * config/powerpcspe: New port. Files are copied from the rs6000
27726 port, with "rs6000" in filenames replaced by "powerpcspe".
27728 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
27730 PR rtl-optimization/80754
27731 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
27733 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
27735 * config/sparc/sparc.md (length): Return the correct value for -mflat
27736 sibcalls to match output_sibcall.
27738 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27742 * config/rs6000/rs6000.c (struct machine_function): Add new field
27744 (rs6000_get_separate_components): Init that field, use it.
27745 (rs6000_components_for_bb): Use the field.
27747 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
27749 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
27751 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
27753 PR middle-end/80823
27754 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
27756 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27759 * config/s390/s390.c (s390_check_qrst_address): Check incoming
27760 address against address_operand predicate.
27761 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
27763 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
27765 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
27766 parameters passed indirectly.
27768 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
27770 * config/i386/i386.md (*movdi_internal): Remove SSE4
27771 alternative 18 (?r, *v). Update insn attributes.
27772 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
27773 Update insn attributes.
27774 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
27775 Update insn attributes.
27776 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
27777 alternative 1 (r, v). Remove isa attribute.
27778 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
27779 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
27780 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
27782 2017-05-23 Tom de Vries <tom@codesourcery.com>
27784 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
27787 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27789 * cgraphunit.c (symbol_table::process_new_functions): Update.
27790 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
27791 (inline_generate_summary): Rename to ...
27792 (ipa_fn_summary_generate): ... this one.
27793 (inline_read_summary): Rename to ...
27794 (ipa_fn_summary_read): ... this one.
27795 (inline_write_summary): Rename to ...
27796 (ipa_fn_summary_write): ... this one.
27797 (inline_free_summary): Rename to ...
27798 (ipa_free_fn_summary): ... this one.
27799 (pass_data_local_fn_summary, pass_local_fn_summary,
27800 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
27801 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
27802 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
27803 make_pass_ipa_fn_summary): New.
27804 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
27805 inline_write_summary, inline_free_summary): Remove.
27806 (ipa_free_fn_summary) : New.
27807 * ipa-inline.c (ipa_inline): Update.
27808 (pass_ipa_inline): Do not generate summaries.
27809 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
27811 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
27812 and add pass_ipa_fn_summary.
27813 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
27815 (make_pass_inline_parameters): Remove.
27817 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
27819 * omp-low.c (struct omp_context): Remove "default_kind" member.
27822 * omp-offload.c (execute_oacc_device_lower): Remove the
27823 parallelism dimensions function attributes for unparallelized
27824 OpenACC kernels constructs.
27826 2017-05-23 Martin Liska <mliska@suse.cz>
27828 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
27830 (cgraph_edge::make_speculative): Likewise.
27831 (cgraph_edge::resolve_speculation): Likewise.
27832 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27833 (cgraph_node::dump): Likewise.
27834 * cgraph.h: Likewise.
27835 * cgraphunit.c (analyze_functions): Likewise.
27836 (symbol_table::compile): Likewise.
27837 * ipa-cp.c (print_all_lattices): Likewise.
27838 (determine_versionability): Likewise.
27839 (initialize_node_lattices): Likewise.
27840 (ipcp_verify_propagated_values): Likewise.
27841 (estimate_local_effects): Likewise.
27842 (update_profiling_info): Likewise.
27843 (create_specialized_node): Likewise.
27844 (perhaps_add_new_callers): Likewise.
27845 (decide_about_value): Likewise.
27846 (decide_whether_version_node): Likewise.
27847 (identify_dead_nodes): Likewise.
27848 (ipcp_store_bits_results): Likewise.
27849 * ipa-devirt.c (dump_targets): Likewise.
27850 (ipa_devirt): Likewise.
27851 * ipa-icf.c (sem_item::dump): Likewise.
27852 (sem_function::equals): Likewise.
27853 (sem_variable::equals): Likewise.
27854 (sem_item_optimizer::read_section): Likewise.
27855 (sem_item_optimizer::execute): Likewise.
27856 (congruence_class::dump): Likewise.
27857 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
27858 (dump_inline_summary): Likewise.
27859 (estimate_node_size_and_time): Likewise.
27860 (inline_analyze_function): Likewise.
27861 * ipa-inline-transform.c (inline_call): Likewise.
27862 * ipa-inline.c (report_inline_failed_reason): Likewise.
27863 (want_early_inline_function_p): Likewise.
27864 (edge_badness): Likewise.
27865 (update_edge_key): Likewise.
27866 (inline_small_functions): Likewise.
27867 * ipa-profile.c (ipa_profile): Likewise.
27868 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
27869 (ipa_make_edge_direct_to_target): Likewise.
27870 (remove_described_reference): Likewise.
27871 (ipa_impossible_devirt_target): Likewise.
27872 (propagate_controlled_uses): Likewise.
27873 (ipa_print_node_params): Likewise.
27874 (ipcp_transform_function): Likewise.
27875 * ipa-pure-const.c (pure_const_read_summary): Likewise.
27876 (propagate_pure_const): Likewise.
27877 * ipa-reference.c (generate_summary): Likewise.
27878 (read_write_all_from_decl): Likewise.
27879 (propagate): Likewise.
27880 (ipa_reference_read_optimization_summary): Likewise.
27881 * ipa-utils.c (ipa_merge_profiles): Likewise.
27882 * ipa.c (walk_polymorphic_call_targets): Likewise.
27883 (symbol_table::remove_unreachable_nodes): Likewise.
27884 (ipa_single_use): Likewise.
27885 * passes.c (execute_todo): Likewise.
27886 * predict.c (drop_profile): Likewise.
27887 * symtab.c (symtab_node::get_dump_name): New function.
27888 (symtab_node::dump_name): Likewise.
27889 (symtab_node::dump_asm_name): Likewise.
27890 (symtab_node::dump_references): Likewise.
27891 (symtab_node::dump_referring): Likewise.
27892 (symtab_node::dump_base): Likewise.
27893 (symtab_node::debug_symtab): Likewise.
27894 * tree-sra.c (convert_callers_for_node): Likewise.
27895 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
27896 * value-prof.c (init_node_map): Likewise.
27898 2017-05-23 Martin Liska <mliska@suse.cz>
27900 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
27901 and symtab_node::debug_symtab to symbol_table::debug.
27902 * cgraphunit.c (analyze_functions): Use the renamed function.
27903 (symbol_table::compile): Likewise.
27904 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
27905 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
27906 * passes.c (execute_todo): Likewise.
27907 * symtab.c (symbol_table::dump): New function.
27908 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
27910 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27912 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
27913 that nonconst implies exec.
27915 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27917 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27918 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27919 (inline_edge_summary_vec): Turn into ...
27920 (ipa_call_summaries): ... this one.
27921 (redirect_to_unreachable, edge_set_predicate,
27922 evaluate_properties_for_edge, inline_summary_alloc,
27923 reset_ipa_call_summary, reset_inline_summary,
27924 inline_summary_t::duplicate): Update.
27925 (inline_edge_duplication_hook): Turn to ...
27926 (ipa_call_summary_t::duplicate): ... this one.
27927 (inline_edge_removal_hook): Turn to ...
27928 (ipa_call_summary_t::remove): ... this one.
27929 (dump_inline_edge_summary): Turn to ...
27930 (dump_ipa_call_summary): ... this one.
27931 (estimate_function_body_sizes): Update.
27932 (inline_update_callee_summaries): Update.
27933 (remap_edge_change_prob): Update.
27934 (remap_edge_summaries): Update.
27935 (inline_merge_summary): Update.
27936 (do_estimate_edge_time): Update.
27937 (inline_generate_summary): Update.
27938 (inline_read_section): Update.
27939 (inline_read_summary): Update.
27940 (inline_free_summary): Update.
27941 * ipa-inline.c (can_inline_edge_p): Update.
27942 (compute_inlined_call_time): Update.
27943 (want_inline_small_function_p): Update.
27944 (edge_badness): Update.
27945 (early_inliner): Update.
27946 * ipa-inline.h (inline_edge_summary): Turn to ...
27947 (ipa_call_summary): ... this one.
27948 (ipa_call_summary_t): New class.
27949 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27950 (ipa_call_summaries): New.
27951 (inline_edge_summary): Remove.
27952 (estimate_edge_growth): Update.
27953 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27954 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27955 * ipa-split.c (execute_split_functions): Update.
27956 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27958 2017-05-23 Tom de Vries <tom@codesourcery.com>
27960 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27961 attributes): Document rdrand effective target.
27963 2017-05-23 Tom de Vries <tom@codesourcery.com>
27965 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27966 attributes): Sort alphabetically.
27968 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27970 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27972 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27975 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27976 V2DF/V2DI splat into two separate patterns, one that handles
27977 registers, and the other that only handles memory. Drop support
27978 for splatting from a GPR on ISA 2.07 and then splitting the
27979 splat into direct move and splat.
27980 (vsx_splat_<mode>_reg): Likewise.
27981 (vsx_splat_<mode>_mem): Likewise.
27983 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27985 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27987 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27989 PR middle-end/80809
27990 * omp-low.c (finish_taskreg_remap): New function.
27991 (finish_taskreg_scan): If unit size of ctx->record_type
27992 is non-constant, unshare the size expression and replace
27993 decls in it with possible outer var refs.
27995 PR middle-end/80809
27996 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27997 GOVD_SHARED rather than GOVD_PRIVATE with it.
27998 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27999 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
28001 PR middle-end/80853
28002 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
28003 as last argument to build_outer_var_ref for pointer bases of array
28004 section reductions.
28006 2017-05-19 Martin Sebor <msebor@redhat.com>
28008 * print-tree.c (print_node): Print DECL_READ_P flag.
28010 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28012 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
28013 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
28014 * cgraph.c: Likewise.
28015 * cgraphunit.c: Likewise.
28016 * gengtype.c: Likewise.
28017 * ipa-cp.c: Likewise.
28018 * ipa-devirt.c: Likewise.
28019 * ipa-icf.c: Likewise.
28020 * ipa-predicate.c: Likewise.
28021 * ipa-profile.c: Likewise.
28022 * ipa-prop.c: Likewise.
28023 * ipa-split.c: Likewise.
28025 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
28026 edge_predicate_pool, dump_inline_hints,
28027 inline_summary::account_size_time, redirect_to_unreachable,
28028 edge_set_predicate, set_hint_predicate,
28029 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
28030 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
28031 inline_summary_t::remove, remap_hint_predicate_after_duplication,
28032 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
28033 ipa_call_summary_t::remove, initialize_growth_caches,
28034 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
28035 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
28036 mark_modified, unmodified_parm_1, unmodified_parm,
28037 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
28038 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
28039 compute_bb_predicates, will_be_nonconstant_expr_predicate,
28040 will_be_nonconstant_predicate, record_modified_bb_info,
28041 get_minimal_bb, record_modified, param_change_prob,
28042 phi_result_unknown_predicate, predicate_for_phi_result,
28043 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
28044 estimate_function_body_sizes, compute_inline_parameters,
28045 compute_inline_parameters_for_curren, pass_data_inline_parameters,
28046 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
28047 inline_update_callee_summaries, remap_edge_change_prob,
28048 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
28049 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
28050 inline_analyze_function, inline_summary_t::insert,
28051 inline_generate_summary, read_ipa_call_summary, inline_read_section,
28052 inline_read_summary, write_ipa_call_summary, inline_write_summary,
28053 inline_free_summary): Move to ipa-fnsummary.h
28054 (predicate_t): Remove.
28055 * ipa-fnsummary.c: New file.
28056 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
28057 (enum inline_hints_vals, inline_hints, agg_position_info,
28058 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
28059 inline_summaries, ipa_call_summary, ipa_call_summary_t,
28060 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
28061 dump_inline_summary, dump_inline_hints, inline_generate_summary,
28062 inline_read_summary, inline_write_summary, inline_free_summary,
28063 inline_analyze_function, initialize_inline_failed,
28064 inline_merge_summary, inline_update_overall_summary,
28065 compute_inline_parameters): Move to ipa-fnsummary.h
28066 * ipa-fnsummary.h: New file.
28067 * ipa-inline-transform.h: Include ipa-inline.h.
28068 * ipa-inline.c: LIkewise.
28070 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28072 * ipa-inline.c (edge_badness): Use inlined_time instead of
28073 inline_summaries->get.
28075 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28077 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
28079 2017-05-22 Nathan Sidwell <nathan@acm.org>
28081 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
28082 (fdump-lang): Document 'raw' option.
28083 * dumpfile.h (TDI_tu): Delete.
28084 * dumpfile.c (dump_files): Remove translation-unit.
28085 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
28087 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
28089 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
28090 command option from $(AWK) call.
28091 * config/avr/genmultilib.awk: Simplify and rewrite so that it
28092 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
28093 [FORMAT]: Remove handling of variable.
28094 * config/avr/t-multilib: Regenerate.
28096 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28098 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
28100 (dump_inline_summary): Do not print self_time.
28101 (estimate_function_body_sizes): Do not set self_time.
28102 (compute_inline_parameters): Likewise.
28103 (inline_read_section, inline_write_summary): Do not stream self_time.
28104 * ipa-inline.h (inline_summary): Drop self_time.
28106 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28108 * ipa-inline-analysis.c (account_size_time): Rename to ...
28109 (inline_summary::account_size_time): ... this one.
28110 (reset_ipa_call_summary): Turn to ...
28111 (ipa_call_summary::reset): ... this one.
28112 (reset_inline_summary): Turn to ...
28113 (inline_summary::reset): ... this one.
28114 (inline_summary_t::remove): Update.
28115 (inline_summary_t::duplicate): Update.
28116 (ipa_call_summary_t::remove): Update.
28117 (dump_inline_summary): Update.
28118 (estimate_function_body_sizes): Update.
28119 (compute_inline_parameters): Update.
28120 (estimate_node_size_and_time): Update.
28121 (inline_merge_summary): Update.
28122 (inline_update_overall_summary): Update.
28123 (inline_read_section): Update.
28124 (inline_write_summary): Update.
28125 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
28126 add account_size_time and reset member functions.
28127 (ipa_call_summary): Add reset function.
28128 * ipa-predicate.h (predicate::operator &): Constify.
28130 2017-05-22 Richard Biener <rguenther@suse.de>
28132 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
28134 2017-05-19 Jason Merrill <jason@redhat.com>
28136 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
28138 2017-05-19 Marek Polacek <polacek@redhat.com>
28141 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
28142 TYPE_OVERFLOW_WRAPS checks.
28144 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
28146 * tree-core.h (enum omp_clause_default_kind): Add
28147 "OMP_CLAUSE_DEFAULT_PRESENT".
28148 * tree-pretty-print.c (dump_omp_clause): Handle it.
28149 * gimplify.c (enum gimplify_omp_var_data): Add
28150 "GOVD_MAP_FORCE_PRESENT".
28151 (gimplify_adjust_omp_clauses_1): Map it to
28152 "GOMP_MAP_FORCE_PRESENT".
28153 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
28155 * gimplify.c (oacc_default_clause): Clarify.
28157 2017-05-19 Nathan Sidwell <nathan@acm.org>
28159 LANG_HOOK_REGISTER_DUMPS
28160 * toplev.c (general_init): Call register dump lang hook.
28161 * doc/invoke.texi: Document -fdump-lang option family.
28162 * dumpfile.c (dump_files): Remove class dump here.
28163 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
28164 * dumpfile.h (tree_dump_index): Remove TDI_class.
28165 * langhooks-def.h (lhd_register_dumps): Declare.
28166 (LANG_HOOKS_REGISTER_DUMPS): Define.
28167 (LANG_HOOKS_INITIALIZER): Add it.
28168 * langhooks.c (lhd_register_dumps): Define.
28169 * langhooks.h (struct lang_hooks): Add register_dumps.
28171 2017-05-19 Nathan Sidwell <nathan@acm.org>
28173 * context.h (context::set_passes): New.
28174 * context.c (context::context): Do not create pass manager.
28175 * toplev.c (general_init): Create pass manager here.
28177 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
28179 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
28180 use this splitter if two add or or instructions would also work for
28181 the constant we want to generate.
28183 2017-05-19 Richard Biener <rguenther@suse.de>
28186 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
28187 predicate evaluation.
28189 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28191 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
28193 * ipa-inline.c (want_inline_small_function_p): Do not cast to
28196 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28198 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28199 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28200 (inline_edge_summary_vec): Turn into ...
28201 (ipa_call_summaries): ... this one.
28202 (redirect_to_unreachable, edge_set_predicate,
28203 evaluate_properties_for_edge, inline_summary_alloc,
28204 reset_ipa_call_summary, reset_inline_summary,
28205 inline_summary_t::duplicate): Update.
28206 (inline_edge_duplication_hook): Turn to ...
28207 (ipa_call_summary_t::duplicate): ... this one.
28208 (inline_edge_removal_hook): Turn to ...
28209 (ipa_call_summary_t::remove): ... this one.
28210 (dump_inline_edge_summary): Turn to ...
28211 (dump_ipa_call_summary): ... this one.
28212 (estimate_function_body_sizes): Update.
28213 (inline_update_callee_summaries): Update.
28214 (remap_edge_change_prob): Update.
28215 (remap_edge_summaries): Update.
28216 (inline_merge_summary): Update.
28217 (do_estimate_edge_time): Update.
28218 (inline_generate_summary): Update.
28219 (inline_read_section): Update.
28220 (inline_read_summary): Update.
28221 (inline_free_summary): Update.
28222 * ipa-inline.c (can_inline_edge_p): Update.
28223 (compute_inlined_call_time): Update.
28224 (want_inline_small_function_p): Update.
28225 (edge_badness): Update.
28226 (early_inliner): Update.
28227 * ipa-inline.h (inline_edge_summary): Turn to ...
28228 (ipa_call_summary): ... this one.
28229 (ipa_call_summary_t): New class.
28230 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28231 (ipa_call_summaries): New.
28232 (inline_edge_summary): Remove.
28233 (estimate_edge_growth): Update.
28234 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28235 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28236 * ipa-split.c (execute_split_functions): Update.
28237 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28239 2017-05-19 Richard Biener <rguenther@suse.de>
28241 PR middle-end/80764
28242 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
28244 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
28246 * config/rs6000/rs6000.c (struct machine_function): Add field
28247 fpr_is_wrapped_separately.
28248 (rs6000_get_separate_components): Use 64 components. Handle the
28249 new FPR components.
28250 (rs6000_components_for_bb): Handle the FPR components.
28251 (rs6000_emit_prologue_components): Handle the FPR components.
28252 (rs6000_emit_epilogue_components): Handle the FPR components.
28253 (rs6000_set_handled_components): Handle the FPR components.
28254 (rs6000_emit_prologue): Don't output prologue code for those FPRs
28255 that are already separately shrink-wrapped.
28256 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
28257 that are already separately shrink-wrapped.
28259 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28262 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
28265 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
28266 (define_peephole2 for Altivec d-form load): Add peepholes to catch
28267 cases where the register allocator uses a move and an offsettable
28268 memory operation to/from a FPR register on ISA 2.06/2.07.
28269 (define_peephole2 for Altivec d-form store): Likewise.
28271 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
28274 * config/i386/mmx.md (*mov<mode>_internal): Enable
28275 alternatives 11, 12, 13 and 14 also for 32bit targets.
28276 Remove alternatives 15, 16, 17 and 18.
28277 * config/i386/sse.md (vec_concatv2di): Change
28278 alternative (!x, *y) to (x, ?!*Yn).
28280 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
28282 * dumpfile.h (enum dump_kind): Remove stray comma.
28284 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28286 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
28287 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
28288 predicate::num_conditions
28289 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
28290 (CHANGED): turn into predicate::changed.
28291 (agg_position_info): Move to ipa-predicate.h
28292 (add_condition, predicate::add_clause, predicate::operator &=,
28293 predicate::or_with, predicate::evaluate, predicate::probability,
28294 dump_condition, dump_clause, predicate::dump,
28295 predicate::remap_after_duplication, predicate::remap_after_inlining,
28296 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
28297 (evaluate_conditions_for_known_args): Update.
28298 (set_cond_stmt_execution_predicate): Update.
28299 * ipa-inline.h: Include ipa-predicate.h
28300 (condition, inline_param_summary, conditions, agg_position_info,
28301 predicate): Move to ipa-predicate.h
28302 * ipa-predicate.c: New file.
28303 * ipa-predicate.h: New file.
28305 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
28307 * final.c (leaf_function_p): Check we are not in a sequence.
28309 2017-05-18 Martin Liska <mliska@suse.cz>
28311 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
28312 * dumpfile.c (dump_register): Use new enum dump_kind.
28313 (get_dump_file_name): Likewise.
28314 (dump_enable_all): Likewise.
28315 (dump_switch_p_1): Likewise.
28316 (enable_rtl_dump_file): Remove usage of TDF_RTL.
28317 * dumpfile.h (enum dump_kind): New enum type.
28318 (struct dump_file_info): Create constructor and
28319 format fields and comments.
28320 * passes.c (pass_manager::register_one_dump_file):
28322 * statistics.c (statistics_early_init): Likewise.
28323 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
28324 TDF_TREE with TDF_SLIM.
28325 (gather_memory_references_ref): Likewise.
28327 2017-05-18 Martin Liska <mliska@suse.cz>
28329 * vec.h (struct vnull): Use it.
28331 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28333 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
28334 (true_predicate, false_predicate, true_predicate_p,
28335 false_predicate_p): Remove.
28336 (single_cond_predicate, not_inlined_predicate): Turn to member function
28338 (add_condition): Update.
28339 (add_clause): Turn to...
28340 (predicate::add_clause): ... this one; update; allow passing NULL
28342 (and_predicates): Turn to ...
28343 (predicate::operator &=): ... this one.
28344 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
28345 (or_predicates): Turn to ...
28346 (predicate::or_with): ... this one.
28347 (evaluate_predicate): Turn to ...
28348 (predicate::evaluate): ... this one.
28349 (predicate_probability): Turn to ...
28350 (predicate::probability): ... this one.
28351 (dump_condition): Update.
28352 (dump_predicate): Turn to ...
28353 (predicate::dump): ... this one.
28354 (account_size_time): Update.
28355 (edge_set_predicate): Update.
28356 (set_hint_predicate): UPdate.
28357 (evaluate_conditions_for_known_args): Update.
28358 (evaluate_properties_for_edge): Update.
28359 (remap_predicate_after_duplication): Turn to...
28360 (predicate::remap_after_duplication): ... this one.
28361 (remap_hint_predicate_after_duplication): Update.
28362 (inline_summary_t::duplicate): UPdate.
28363 (dump_inline_edge_summary): Update.
28364 (dump_inline_summary): Update.
28365 (set_cond_stmt_execution_predicate): Update.
28366 (set_switch_stmt_execution_predicate): Update.
28367 (compute_bb_predicates): Update.
28368 (will_be_nonconstant_expr_predicate): Update.
28369 (will_be_nonconstant_predicate): Update.
28370 (phi_result_unknown_predicate): Update.
28371 (predicate_for_phi_result): Update.
28372 (array_index_predicate): Update.
28373 (estimate_function_body_sizes): Update.
28374 (estimate_node_size_and_time): Update.
28375 (estimate_ipcp_clone_size_and_time): Update.
28376 (remap_predicate): Rename to ...
28377 (predicate::remap_after_inlining): ... this one.
28378 (remap_hint_predicate): Update.
28379 (inline_merge_summary): Update.
28380 (inline_update_overall_summary): Update.
28381 (estimate_size_after_inlining): Update.
28382 (read_predicate): Rename to ...
28383 (predicate::stream_in): ... this one.
28384 (read_inline_edge_summary): Update.
28385 (write_predicate): Rename to ...
28386 (predicate::stream_out): ... this one.
28387 (write_inline_edge_summary): Update.
28388 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
28389 (clause_t): Turn to uint32_t
28390 (predicate): Turn to class; implement constructor and operators
28392 (size_time_entry): Update.
28393 (inline_summary): Update.
28394 (inline_edge_summary): Update.
28396 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
28398 * fold-const.c (fold_binary_loc): Move transformation...
28399 * match.pd (C - X CMP X): ... here.
28401 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
28403 * config/sparc/sparc.c (sparc_option_override): Set function
28404 alignment for -mcpu=niagara7 to 64 to match the I$ line.
28405 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
28407 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
28409 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
28411 2017-05-18 Marek Polacek <polacek@redhat.com>
28414 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
28415 (pass_ubsan::execute): Call gimple_assign_single_p instead of
28416 gimple_assign_load_p.
28418 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
28420 PR middle-end/80692
28421 * real.c (do_compare): Give decimal_do_compare preference over
28422 comparing just the signs.
28424 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
28426 * doc/md.texi (Canonicalization of Instructions): Describe the
28427 canonical form of instructions that inherently set a condition
28430 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
28432 PR middle-end/80775
28433 * tree-cfg.c: Move deletion of unreachable case statements to after
28434 the merging of consecutive case labels.
28436 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
28438 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
28439 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
28440 restoring of callee-saved registers.
28442 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
28444 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
28445 * config/visium/visium.c (single_set_and_flags): Likewise.
28446 * config/visium/visium.md (Substitutions): Likewise.
28448 2017-05-17 Martin Liska <mliska@suse.cz>
28450 * cfg.c: Introduce dump_flags_t type and
28451 use it instead of int type.
28453 * cfghooks.c: Likewise.
28454 * cfghooks.h (struct cfg_hooks): Likewise.
28455 * cfgrtl.c: Likewise.
28456 * cfgrtl.h: Likewise.
28457 * cgraph.c (cgraph_node::get_body): Likewise.
28458 * coretypes.h: Likewise.
28459 * domwalk.c: Likewise.
28460 * domwalk.h: Likewise.
28461 * dumpfile.c (struct dump_option_value_info): Likewise.
28462 (dump_enable_all): Likewise.
28463 (dump_switch_p_1): Likewise.
28464 (opt_info_switch_p): Likewise.
28465 * dumpfile.h (enum tree_dump_index): Likewise.
28466 (struct dump_file_info): Likewise.
28467 * genemit.c: Likewise.
28468 * generic-match-head.c: Likewise.
28469 * gengtype.c (open_base_files): Likewise.
28470 * gimple-pretty-print.c: Likewise.
28471 * gimple-pretty-print.h: Likewise.
28472 * graph.c (print_graph_cfg): Likewise.
28473 * graphite-scop-detection.c (dot_all_sese): Likewise.
28474 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
28475 * loop-unroll.c (report_unroll): Likewise.
28476 * passes.c (pass_manager::register_one_dump_file): Likewise.
28477 * print-tree.c: Likewise.
28478 * statistics.c: Likewise.
28479 * tree-cfg.c: Likewise.
28480 * tree-cfg.h: Likewise.
28481 * tree-dfa.c: Likewise.
28482 * tree-dfa.h: Likewise.
28483 * tree-dump.c (dump_function): Likewise.
28484 * tree-dump.h (struct dump_info): Likewise.
28485 * tree-pretty-print.c: Likewise.
28486 * tree-pretty-print.h: Likewise.
28487 * tree-ssa-live.c: Likewise.
28488 * tree-ssa-live.h: Likewise.
28489 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
28490 * tree-vect-loop.c: Likewise.
28491 * tree-vect-slp.c: Likewise.
28493 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
28494 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28496 PR tree-optimization/80457
28497 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
28498 of all arguments to a statement as scalar_to_vec operations.
28499 (vectorizable_call): Adjust call to vect_model_simple_cost for
28501 (vectorizable_conversion): Likewise.
28502 (vectorizable_assignment): Likewise.
28503 (vectorizable_shift): Likewise.
28504 (vectorizable_operation): Likewise.
28505 (vectorizable_comparison): Likewise.
28506 (vect_is_simple_cond): Record the def types for operands.
28507 (vectorizable_condition): Likewise, call vect_model_simple_cost.
28508 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
28509 for statement argument count.
28511 2017-05-16 Carl Love <cel@us.ibm.com>
28513 * config/rs6000/rs6000-c: Add support for built-in functions
28514 vector unsigned long long vec_bperm (vector unsigned long long,
28515 vector unsigned char)
28516 vector signed long long vec_mule (vector signed int,
28518 vector unsigned long long vec_mule (vector unsigned int,
28519 vector unsigned int)
28520 vector signed long long vec_mulo (vector signed int,
28522 vector unsigned long long vec_mulo (vector unsigned int,
28523 vector unsigned int)
28524 vector signed char vec_sldw (vector signed char,
28525 vector signed char,
28527 vector unsigned char vec_sldw (vector unsigned char,
28528 vector unsigned char,
28530 vector signed short vec_sldw (vector signed short,
28531 vector signed short,
28533 vector unsigned short vec_sldw (vector unsigned short,
28534 vector unsigned short,
28536 vector signed int vec_sldw (vector signed int,
28539 vector unsigned int vec_sldw (vector unsigned int,
28540 vector unsigned int,
28542 vector signed long long vec_sldw (vector signed long long,
28543 vector signed long long,
28545 vector unsigned long long vec_sldw (vector unsigned long long,
28546 vector unsigned long long,
28548 * config/rs6000/rs6000-c: Add support for built-in functions
28549 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
28550 * config/rs6000/altivec.h: Add defintion for vec_sldw.
28551 * doc/extend.texi: Update the built-in documentation for the
28552 new built-in functions.
28554 2017-05-16 Marek Polacek <polacek@redhat.com>
28558 * tree.c (save_expr): Don't fold the expression.
28560 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
28562 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
28563 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
28564 and (?*y,m). Update insn attributes.
28566 2017-05-16 Martin Liska <mliska@suse.cz>
28568 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
28569 flags argument of print_gimple_stmt, print_gimple_expr,
28570 print_generic_stmt and print_generic_expr.
28571 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
28572 * coretypes.h: Likewise.
28573 * except.c (dump_eh_tree): Likewise.
28574 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
28575 * gimple-pretty-print.h: Likewise.
28576 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
28577 (backprop::push_to_worklist): Likewise.
28578 (backprop::pop_from_worklist): Likewise.
28579 (backprop::process_use): Likewise.
28580 (backprop::intersect_uses): Likewise.
28581 (note_replacement): Likewise.
28582 * gimple-ssa-store-merging.c
28583 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
28584 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
28585 (pass_store_merging::execute): Likewise.
28586 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
28587 (ssa_base_cand_dump_callback): Likewise.
28588 (dump_incr_vec): Likewise.
28589 (replace_refs): Likewise.
28590 (replace_mult_candidate): Likewise.
28591 (create_add_on_incoming_edge): Likewise.
28592 (create_phi_basis): Likewise.
28593 (insert_initializers): Likewise.
28594 (all_phi_incrs_profitable): Likewise.
28595 (introduce_cast_before_cand): Likewise.
28596 (replace_one_candidate): Likewise.
28597 * gimplify.c (gimplify_expr): Likewise.
28598 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
28599 (set_rename): Likewise.
28600 (rename_uses): Likewise.
28601 (copy_loop_phi_nodes): Likewise.
28602 (add_close_phis_to_merge_points): Likewise.
28603 (copy_loop_close_phi_args): Likewise.
28604 (copy_cond_phi_args): Likewise.
28605 (graphite_copy_stmts_from_block): Likewise.
28606 (translate_pending_phi_nodes): Likewise.
28607 * graphite-poly.c (print_pdr): Likewise.
28608 (dump_gbb_cases): Likewise.
28609 (dump_gbb_conditions): Likewise.
28610 (print_scop_params): Likewise.
28611 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
28612 (build_cross_bb_scalars_use): Likewise.
28613 (gather_bbs::before_dom_children): Likewise.
28614 * hsa-dump.c (dump_hsa_immed): Likewise.
28615 * ipa-cp.c (print_ipcp_constant_value): Likewise.
28616 (get_replacement_map): Likewise.
28617 * ipa-inline-analysis.c (dump_condition): Likewise.
28618 (estimate_function_body_sizes): Likewise.
28619 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
28620 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
28621 * ipa-prop.c (ipa_dump_param): Likewise.
28622 (ipa_print_node_jump_functions_for_edge): Likewise.
28623 (ipa_modify_call_arguments): Likewise.
28624 (ipa_modify_expr): Likewise.
28625 (ipa_dump_param_adjustments): Likewise.
28626 (ipa_dump_agg_replacement_values): Likewise.
28627 (ipcp_modif_dom_walker::before_dom_children): Likewise.
28628 * ipa-pure-const.c (check_stmt): Likewise.
28629 (pass_nothrow::execute): Likewise.
28630 * ipa-split.c (execute_split_functions): Likewise.
28631 * omp-offload.c (dump_oacc_loop_part): Likewise.
28632 (dump_oacc_loop): Likewise.
28633 * trans-mem.c (tm_log_emit): Likewise.
28634 (tm_memopt_accumulate_memops): Likewise.
28635 (dump_tm_memopt_set): Likewise.
28636 (dump_tm_memopt_transform): Likewise.
28637 * tree-cfg.c (gimple_verify_flow_info): Likewise.
28638 (print_loop): Likewise.
28639 * tree-chkp-opt.c (chkp_print_addr): Likewise.
28640 (chkp_gather_checks_info): Likewise.
28641 (chkp_get_check_result): Likewise.
28642 (chkp_remove_check_if_pass): Likewise.
28643 (chkp_use_outer_bounds_if_possible): Likewise.
28644 (chkp_reduce_bounds_lifetime): Likewise.
28645 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
28646 (chkp_mark_completed_bounds): Likewise.
28647 (chkp_register_incomplete_bounds): Likewise.
28648 (chkp_mark_invalid_bounds): Likewise.
28649 (chkp_maybe_copy_and_register_bounds): Likewise.
28650 (chkp_build_returned_bound): Likewise.
28651 (chkp_get_bound_for_parm): Likewise.
28652 (chkp_build_bndldx): Likewise.
28653 (chkp_get_bounds_by_definition): Likewise.
28654 (chkp_generate_extern_var_bounds): Likewise.
28655 (chkp_get_bounds_for_decl_addr): Likewise.
28656 * tree-chrec.c (chrec_apply): Likewise.
28657 * tree-data-ref.c (dump_data_reference): Likewise.
28658 (dump_subscript): Likewise.
28659 (dump_data_dependence_relation): Likewise.
28660 (analyze_overlapping_iterations): Likewise.
28661 * tree-inline.c (expand_call_inline): Likewise.
28662 (tree_function_versioning): Likewise.
28663 * tree-into-ssa.c (dump_defs_stack): Likewise.
28664 (dump_currdefs): Likewise.
28665 (dump_names_replaced_by): Likewise.
28666 (dump_update_ssa): Likewise.
28667 (update_ssa): Likewise.
28668 * tree-object-size.c (pass_object_sizes::execute): Likewise.
28669 * tree-parloops.c (build_new_reduction): Likewise.
28670 (try_create_reduction_list): Likewise.
28671 (ref_conflicts_with_region): Likewise.
28672 (oacc_entry_exit_ok_1): Likewise.
28673 (oacc_entry_exit_single_gang): Likewise.
28674 * tree-pretty-print.h: Likewise.
28675 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
28676 (get_scalar_evolution): Likewise.
28677 (add_to_evolution): Likewise.
28678 (get_loop_exit_condition): Likewise.
28679 (analyze_evolution_in_loop): Likewise.
28680 (analyze_initial_condition): Likewise.
28681 (analyze_scalar_evolution): Likewise.
28682 (instantiate_scev): Likewise.
28683 (number_of_latch_executions): Likewise.
28684 (gather_chrec_stats): Likewise.
28685 (final_value_replacement_loop): Likewise.
28686 (scev_const_prop): Likewise.
28687 * tree-sra.c (dump_access): Likewise.
28688 (disqualify_candidate): Likewise.
28689 (create_access): Likewise.
28690 (reject): Likewise.
28691 (maybe_add_sra_candidate): Likewise.
28692 (create_access_replacement): Likewise.
28693 (analyze_access_subtree): Likewise.
28694 (analyze_all_variable_accesses): Likewise.
28695 (sra_modify_assign): Likewise.
28696 (initialize_constant_pool_replacements): Likewise.
28697 (find_param_candidates): Likewise.
28698 (decide_one_param_reduction): Likewise.
28699 (replace_removed_params_ssa_names): Likewise.
28700 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
28701 * tree-ssa-copy.c (dump_copy_of): Likewise.
28702 (copy_prop_visit_cond_stmt): Likewise.
28703 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
28704 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
28705 (record_equivalences_from_stmt): Likewise.
28706 * tree-ssa-dse.c (compute_trims): Likewise.
28707 (delete_dead_call): Likewise.
28708 (delete_dead_assignment): Likewise.
28709 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
28710 (forward_propagate_into_cond): Likewise.
28711 (pass_forwprop::execute): Likewise.
28712 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
28713 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
28715 (move_computations_worker): Likewise.
28716 (execute_sm): Likewise.
28717 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
28718 (remove_exits_and_undefined_stmts): Likewise.
28719 (remove_redundant_iv_tests): Likewise.
28720 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
28721 (adjust_iv_update_pos): Likewise.
28722 * tree-ssa-math-opts.c (bswap_replace): Likewise.
28723 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
28724 (value_replacement): Likewise.
28725 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
28726 * tree-ssa-pre.c (print_pre_expr): Likewise.
28727 (get_representative_for): Likewise.
28728 (create_expression_by_pieces): Likewise.
28729 (insert_into_preds_of_block): Likewise.
28730 (eliminate_insert): Likewise.
28731 (eliminate_dom_walker::before_dom_children): Likewise.
28732 (eliminate): Likewise.
28733 (remove_dead_inserted_code): Likewise.
28734 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
28735 * tree-ssa-reassoc.c (get_rank): Likewise.
28736 (eliminate_duplicate_pair): Likewise.
28737 (eliminate_plus_minus_pair): Likewise.
28738 (eliminate_not_pairs): Likewise.
28739 (undistribute_ops_list): Likewise.
28740 (eliminate_redundant_comparison): Likewise.
28741 (update_range_test): Likewise.
28742 (optimize_range_tests_var_bound): Likewise.
28743 (optimize_vec_cond_expr): Likewise.
28744 (rewrite_expr_tree): Likewise.
28745 (rewrite_expr_tree_parallel): Likewise.
28746 (linearize_expr): Likewise.
28747 (break_up_subtract): Likewise.
28748 (linearize_expr_tree): Likewise.
28749 (attempt_builtin_powi): Likewise.
28750 (attempt_builtin_copysign): Likewise.
28751 (transform_stmt_to_copy): Likewise.
28752 (transform_stmt_to_multiply): Likewise.
28753 (dump_ops_vector): Likewise.
28754 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
28755 (print_scc): Likewise.
28756 (set_ssa_val_to): Likewise.
28757 (visit_reference_op_store): Likewise.
28758 (visit_use): Likewise.
28759 (sccvn_dom_walker::before_dom_children): Likewise.
28760 (run_scc_vn): Likewise.
28761 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
28763 (expr_hash_elt::print): Likewise.
28764 (const_and_copies::pop_to_marker): Likewise.
28765 (const_and_copies::record_const_or_copy_raw): Likewise.
28766 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
28767 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
28768 (dump_predicates): Likewise.
28769 (find_uninit_use): Likewise.
28770 (warn_uninitialized_phi): Likewise.
28771 (pass_late_warn_uninitialized::execute): Likewise.
28772 * tree-ssa.c (verify_vssa): Likewise.
28773 (verify_ssa): Likewise.
28774 (maybe_optimize_var): Likewise.
28775 * tree-vrp.c (dump_value_range): Likewise.
28776 (dump_all_value_ranges): Likewise.
28777 (dump_asserts_for): Likewise.
28778 (register_edge_assert_for_2): Likewise.
28779 (vrp_visit_cond_stmt): Likewise.
28780 (vrp_visit_switch_stmt): Likewise.
28781 (vrp_visit_stmt): Likewise.
28782 (vrp_visit_phi_node): Likewise.
28783 (simplify_cond_using_ranges_1): Likewise.
28784 (fold_predicate_in): Likewise.
28785 (evrp_dom_walker::before_dom_children): Likewise.
28786 (evrp_dom_walker::push_value_range): Likewise.
28787 (evrp_dom_walker::pop_value_range): Likewise.
28788 (execute_early_vrp): Likewise.
28790 2017-05-16 Richard Biener <rguenther@suse.de>
28792 * dwarf2out.c (loc_list_from_tree_1): Do not create
28793 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
28795 2017-05-16 Richard Biener <rguenther@suse.de>
28797 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
28799 (note_variable_value_in_expr): If we resolved the decl ref
28800 do not push to the stack.
28802 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
28804 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
28805 operations in fast-math mode.
28806 (vaddq_f16): Likewise.
28807 (vmul_f16): Likewise.
28808 (vmulq_f16): Likewise.
28809 (vsub_f16): Likewise.
28810 (vsubq_f16): Likewise.
28811 * config/arm/neon.md (add<mode>3): New.
28813 (fma:<VH:mode>3): New. Also remove outdated comment.
28816 2017-05-16 Martin Liska <mliska@suse.cz>
28820 * ipa-devirt.c (warn_types_mismatch): Fix typo.
28821 (odr_types_equivalent_p): Likewise.
28823 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
28825 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
28827 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
28830 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
28831 non-interunit SSE move alternatives with '?'.
28832 (zero-extendsidi peephole2): New peephole to skip intermediate
28833 general register in SSE zero-extend sequence.
28835 2017-05-15 Jeff Law <law@redhat.com>
28837 * reorg.c (relax_delay_slots): Create a new variable to hold
28838 the temporary target rather than clobbering TARGET_LABEL.
28840 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
28841 missing argument to extract_bit_field call.
28842 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
28844 2017-05-15 Martin Liska <mliska@suse.cz>
28847 * gcc.c (process_command): Do not allow empty argument of -o option.
28849 2017-05-15 Renlin Li <renlin.li@arm.com>
28851 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
28852 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
28853 * config/aarch64/constraints.md (Usf): Add long call check.
28854 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
28855 (call_value): Likewise.
28856 (sibcall): Likewise.
28857 (sibcall_value): Likewise.
28859 (call_value_insn): New.
28860 (sibcall_insn): Update rtx pattern.
28861 (sibcall_value_insn): Likewise.
28862 (call_internal): Remove.
28863 (call_value_internal): Likewise.
28864 (sibcall_internal): Likewise.
28865 (sibcall_value_internal): Likewise.
28866 (call_reg): Likewise.
28867 (call_symbol): Likewise.
28868 (call_value_reg): Likewise.
28869 (call_value_symbol): Likewise.
28871 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
28874 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
28876 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
28878 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
28879 compatible with CCGOCmode and with CCZmode.
28881 2017-05-14 Martin Sebor <msebor@redhat.com>
28883 PR middle-end/77671
28884 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
28885 (gimple_fold_builtin_snprintf): Same.
28886 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
28887 (gimple_fold_builtin_snprintf): Same.
28888 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
28889 of character types.
28890 (is_call_safe): New function.
28891 (try_substitute_return_value): Call it.
28892 (try_simplify_call): New function.
28893 (pass_sprintf_length::handle_gimple_call): Call it.
28895 2017-05-14 Martin Sebor <msebor@redhat.com>
28897 PR middle-end/80669
28898 * builtins.c (expand_builtin_stpncpy): Simplify.
28900 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
28902 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
28903 * config/i386/i386.h
28904 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
28905 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
28906 (struct machine_function): Add new members call_ms2sysv,
28907 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
28908 (struct machine_frame_state): New fields sp_realigned and
28909 sp_realigned_offset.
28910 * config/i386/i386.c
28911 (enum xlogue_stub): New enum.
28912 (enum xlogue_stub_sets): New enum.
28913 (class xlogue_layout): New class.
28914 (struct ix86_frame): New fields stack_realign_allocate_offset,
28915 stack_realign_offset and outlined_save_offset. Modify comments to
28916 detail stack layout when using out-of-line stubs.
28917 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
28918 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
28919 -mcall-ms2sysv-xlogues.
28920 (stub_managed_regs): New static variable.
28921 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
28922 registers managed by out-of-line stub.
28923 (disable_call_ms2sysv_xlogues): New function.
28924 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
28925 m->call_ms2sysv when appropriate and compute frame layout for
28927 (sp_valid_at, fp_valid_at): New inline functions.
28928 (choose_basereg): New function.
28929 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28931 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28932 Use align parameter of choose_baseaddr to generated aligned SSE movs
28934 (pro_epilogue_adjust_stack): Modify to track
28935 machine_frame_state::sp_realigned.
28936 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28937 (ix86_nsaved_sseregs): Likewise.
28938 (ix86_emit_save_regs): Likewise.
28939 (ix86_emit_save_regs_using_mov): Likewise.
28940 (ix86_emit_save_sse_regs_using_mov): Likewise.
28941 (get_scratch_register_on_entry): Likewise.
28942 (gen_frame_set): New function.
28943 (gen_frame_load): Likewise.
28944 (gen_frame_store): Likewise.
28945 (emit_outlined_ms2sysv_save): Likewise.
28946 (emit_outlined_ms2sysv_restore): Likewise.
28947 (ix86_expand_prologue): Modify stack re-alignment code and call
28948 emit_outlined_ms2sysv_save when appropriate.
28949 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28950 parameter rtx_insn *insn, which allows the function to be used to only
28951 generate the notes.
28952 (ix86_expand_epilogue): Modify validity checks of frame and stack
28953 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28954 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28955 * config/i386/predicates.md
28956 (save_multiple): New predicate.
28957 (restore_multiple): Likewise.
28958 * config/i386/sse.md
28959 (save_multiple<mode>): New pattern.
28960 (save_multiple_realign<mode>): Likewise.
28961 (restore_multiple<mode>): Likewise.
28962 (restore_multiple_and_return<mode>): Likewise.
28963 (restore_multiple_leave_return<mode>): Likewise.
28964 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28966 2017-05-14 Julia Koval <julia.koval@intel.com>
28968 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28969 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28970 (__builtin_ia32_xsetbv): New builtins.
28971 * config/i386/i386.c (ix86_expand_special_args_builtin):
28973 (ix86_expand_builtin): Special expand for new intrinsics.
28974 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28975 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28976 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28978 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28980 * cfganal.c (inverted_post_order_compute): Change argument type
28982 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28983 * df-core.c (rest_of_handle_df_initialize): Adjust.
28984 (rest_of_handle_df_finish): Likewise.
28985 (df_analyze_1): Likewise.
28986 (df_analyze): Likewise.
28987 (loop_inverted_post_order_compute): Change argument to be a vec *.
28988 (df_analyze_loop): Adjust.
28989 (df_get_n_blocks): Likewise.
28990 (df_get_postorder): Likewise.
28991 * df.h (struct df_d): Change field to be a vec.
28992 * lcm.c (compute_laterin): Adjust.
28993 (compute_available): Likewise.
28994 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28995 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28996 * tree-ssa-pre.c (compute_antic): Likewise.
28998 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29000 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
29001 (depth_first_search::depth_first_search): Change structure init
29002 function to this constructor.
29003 (depth_first_search::add_bb): Rename function to this member.
29004 (depth_first_search::execute): Likewise.
29005 (flow_dfs_compute_reverse_finish): Adjust.
29007 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29009 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
29010 (longest_simple_path): Likewise.
29011 * shrink-wrap.c (spread_components): Likewise.
29012 (disqualify_problematic_components): Likewise.
29013 (emit_common_heads_for_components): Likewise.
29014 (emit_common_tails_for_components): Likewise.
29015 (insert_prologue_epilogue_for_components): Likewise.
29017 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29019 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
29022 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29024 * df-core.c (df_set_blocks): Start using auto_bitmap.
29025 (df_compact_blocks): Likewise.
29026 * df-problems.c (df_rd_confluence_n): Likewise.
29027 * df-scan.c (df_insn_rescan_all): Likewise.
29028 (df_process_deferred_rescans): Likewise.
29029 (df_update_entry_block_defs): Likewise.
29030 (df_update_exit_block_uses): Likewise.
29031 (df_entry_block_bitmap_verify): Likewise.
29032 (df_exit_block_bitmap_verify): Likewise.
29033 (df_scan_verify): Likewise.
29034 * lra-constraints.c (lra_constraints): Likewise.
29035 (undo_optional_reloads): Likewise.
29036 (lra_undo_inheritance): Likewise.
29037 * lra-remat.c (calculate_gen_cands): Likewise.
29038 (do_remat): Likewise.
29039 * lra-spills.c (assign_spill_hard_regs): Likewise.
29040 (spill_pseudos): Likewise.
29041 * tree-ssa-pre.c (bitmap_set_and): Likewise.
29042 (bitmap_set_subtract_values): Likewise.
29044 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29046 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
29047 management with auto_bitmap.
29048 (fix_inter_tick): Likewise.
29049 (fix_recovery_deps): Likewise.
29050 * ira.c (add_store_equivs): Likewise.
29051 (find_moveable_pseudos): Likewise.
29052 (split_live_ranges_for_shrink_wrap): Likewise.
29053 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
29054 (rtx_reuse_manager::seen_def_p): Likewise.
29055 (rtx_reuse_manager::set_seen_def): Likewise.
29056 * print-rtl.h (class rtx_reuse_manager): Likewise.
29058 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29060 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
29062 (migrate_btr_def): Likewise.
29063 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
29064 * df-core.c (loop_post_order_compute): Likewise.
29065 (loop_inverted_post_order_compute): Likewise.
29066 * hsa-common.h: Likewise.
29067 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
29068 * init-regs.c (initialize_uninitialized_regs): Likewise.
29069 * ipa-inline.c (resolve_noninline_speculation): Likewise.
29070 (inline_small_functions): Likewise.
29071 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
29072 * ira.c (combine_and_move_insns): Likewise.
29073 (build_insn_chain): Likewise.
29074 * loop-invariant.c (find_invariants): Likewise.
29075 * lower-subreg.c (propagate_pseudo_copies): Likewise.
29076 * predict.c (tree_predict_by_opcode): Likewise.
29077 (predict_paths_leading_to): Likewise.
29078 (predict_paths_leading_to_edge): Likewise.
29079 (estimate_loops_at_level): Likewise.
29080 (estimate_loops): Likewise.
29081 * shrink-wrap.c (try_shrink_wrapping): Likewise.
29082 (spread_components): Likewise.
29083 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
29084 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
29085 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
29086 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
29087 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
29088 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
29089 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
29090 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
29091 (mark_threaded_blocks): Likewise.
29092 (thread_through_all_blocks): Likewise.
29093 * tree-ssa.c (verify_ssa): Likewise.
29094 (execute_update_addresses_taken): Likewise.
29095 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
29097 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29099 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
29101 (post_order_compute): Likewise.
29102 (inverted_post_order_compute): Likewise.
29103 (pre_and_rev_post_order_compute_fn): Likewise.
29105 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29107 * genrecog.c (int_set::int_set): Explicitly construct our
29108 auto_vec base class.
29109 * vec.h (auto_vec::auto_vec): New constructor.
29111 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29113 * bitmap.h (class auto_bitmap): New constructor taking
29114 bitmap_obstack * argument.
29116 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29118 * bitmap.h (class auto_bitmap): Change type of m_bits to
29119 bitmap_head, and adjust ctor / dtor and member operators.
29121 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
29123 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
29124 when returned register mode doesn't match original mode.
29126 2017-05-12 Jeff Law <law@redhat.com>
29127 Jakub Jelinek <jakub@redhat.com>
29129 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
29130 we look for cc setter after the compare-elim changes.
29131 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
29132 within the vector to match what compare-elim now expects.
29133 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
29134 (xorsi3_flags, one_cmplsi2_flags): Likewise.
29136 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
29137 after the compare-elim changes.
29138 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
29139 the vector to match what compare-elim now expects.
29140 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
29141 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
29142 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
29143 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
29144 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
29146 * config/visium/visium.c (single_set_and_flags): Fix where
29147 we look for cc setter after the compare-elim changes.
29148 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
29149 with the vector to match what compare-elim now expects.
29150 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
29151 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
29152 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
29153 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
29154 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
29155 (neg<mode>2_insn_set_overflow): Likewise.
29157 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
29159 PR middle-end/79794
29160 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
29161 maybe_expand_insn call, set ops[0].target. If still set after call,
29162 set alt_rtl. Add extra arg to recursive calls.
29163 (extract_bit_field): Add alt_rtl argument. Pass to
29165 * expmed.h (extract_bit_field): Fix prototype.
29166 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
29167 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
29168 to extract_bit_field_calls.
29169 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
29170 Pass alt_rtl to extract_bit_field calls.
29171 * calls.c (store_unaligned_arguments_into_psuedos)
29172 load_register_parameters): Pass extra NULL to extract_bit_field calls.
29173 * optabs.c (maybe_legitimize_operand): Clear op->target when call
29175 * optabs.h (struct expand_operand): Add target bitfield.
29177 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29179 * compare-elim.c (try_eliminate_compare): Canonicalize
29180 operation with embedded compare to
29181 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
29182 (set (reg) (operation)].
29184 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
29186 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29189 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
29190 cost of adding a carry flag for ADC instruction.
29191 [case MINUS]: Ignore the cost of subtracting a carry flag
29192 for SBB instruction.
29194 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
29196 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
29198 * config/rs6000/bmiintrin.h: New file.
29199 * config/rs6000/bmi2intrin.h: New file.
29200 * config/rs6000/x86intrin.h: New file.
29202 2017-05-12 Jeff Law <law@redhat.com>
29204 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
29207 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
29209 PR middle-end/80707
29210 * tree-cfg.c: Remove cfg edges of unreachable case statements.
29212 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29214 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29215 early expansion of vector divide builtins.
29216 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
29217 builtins identified as having unsigned arguments.
29219 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29221 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
29222 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
29223 expansion of vector logical operations (and, andc, or, xor,
29226 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29228 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
29229 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
29231 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29233 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29234 early GIMPLE expansion of vector multiplies.
29236 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29238 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
29239 TARGET_HAVE_MOVT conditional.
29240 (movt splitter): Likewise.
29242 2017-05-12 Richard Biener <rguenther@suse.de>
29244 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
29245 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29246 Fold all stmts not inplace.
29248 2017-05-12 Richard Biener <rguenther@suse.de>
29250 PR tree-optimization/80713
29251 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
29252 inserted_exprs bit for not removed stmts.
29254 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
29256 PR middle-end/69921
29257 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
29258 parallelized" attribute for parallelized OpenACC kernels.
29259 * omp-offload.c (execute_oacc_device_lower): Use it.
29261 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
29262 Set "oacc kernels" attribute.
29263 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
29264 parameter. Adjust all users.
29265 (oacc_fn_attrib_kernels_p): Remove function.
29266 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
29267 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
29268 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
29269 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
29270 assert "oacc kernels" attribute is set.
29272 2017-05-11 Carl Love <cel@us.ibm.com>
29274 * config/rs6000/rs6000-c: Add support for built-in functions
29275 vector unsigned char vec_popcnt (vector signed char)
29276 vector unsigned char vec_popcnt (vector unsigned char)
29277 vector unsigned short vec_popcnt (vector signed short)
29278 vector unsigned short vec_popcnt (vector unsigned short)
29279 vector unsigned int vec_popcnt (vector signed int)
29280 vector unsigned int vec_popcnt (vector unsigned int)
29281 vector unsigned long long vec_popcnt (vector signed long long)
29282 vector unsigned long long vec_popcnt (vector unsigned long long)
29283 vector signed long long vec_slo (vector signed long long,
29284 vector signed char)
29285 vector signed long long vec_slo (vector signed long long,
29286 vector unsigned char)
29287 vector unsigned long long vec_slo (vector unsigned long long,
29288 vector signed char)
29289 vector unsigned long long vec_slo (vector unsigned long long,
29290 vector unsigned char)
29291 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
29292 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
29293 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
29294 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
29295 * doc/extend.texi: Update the built-in documentation file for the
29296 new built-in functions.
29298 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
29300 * attribs.h (sorted_attr_string): Move machine independent
29301 functions for target clone support from the i386 port to common
29302 code. Rename ix86_function_versions to common_function_versions.
29303 Rename make_name to make_unique_name.
29304 (common_function_versions): Likewise.
29305 (make_unique_name): Likewise.
29306 (make_dispatcher_decl): Likewise.
29307 (is_function_default_version): Likewise.
29308 * attribs.c (attr_strcmp): Likewise.
29309 (sorted_attr_string): Likewise.
29310 (common_function_versions): Likewise.
29311 (make_unique_name): Likewise.
29312 (make_dispatcher_decl): Likewise.
29313 (is_function_default_version): Likewise.
29314 * config/i386/i386.c (attr_strcmp): Likewise.
29315 (sorted_attr_string): Likewise.
29316 (ix86_function_versions): Likewise.
29317 (make_name): Likewise.
29318 (make_dispatcher_decl): Likewise.
29319 (is_function_default_version): Likewise.
29320 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
29322 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29325 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
29326 Account for direct move costs for vec_construct of integer
29329 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
29332 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
29333 (UNSPEC_STX_ATOMIC): Ditto.
29334 (loaddi_via_sse): New insn.
29335 (storedi_via_sse): Ditto.
29336 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
29337 Update corresponding peephole2 patterns.
29338 (atomic_storedi_fpu): Ditto.
29340 2017-05-11 Julia Koval <julia.koval@intel.com>
29342 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
29343 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
29345 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
29346 (__builtin_ia32_rsqrt14ss_mask): New builtins.
29347 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
29349 2017-05-11 Nathan Sidwell <nathan@acm.org>
29351 * graphite-poly.c: Include dumpfile.h.
29353 * dumpfle.h (dump_function): Declare here ...
29354 * tree-dump.h (dump_function): ... not here.
29355 * dumpfile.c: #include tree-cfg.h.
29356 (dump_function): Move here from ...
29357 * tree-dump.c (dump_function): ... here.
29358 * gimplify.c: #include splay-tree.h, not tree-dump.h.
29359 * graphite-poly.c: Don't include tree-dump.h.
29360 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
29361 * print-tree.c: Likewise.
29362 * stor-layout.c: Likewise.
29363 * tree-nested.c: Likewise.
29365 * dumpfile.c (dump_start): Use TDF_FLAGS.
29366 (dump_enable_all): Fix TDF_KIND check thinko.
29368 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
29370 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29371 array entries to represent two legal parameterizations of the
29372 overloaded __builtin_cmpb function, as represented by the
29373 P6_OV_BUILTIN_CMPB constant.
29374 (altivec_resolve_overloaded_builtin): Add special case handling
29375 for the __builtin_cmpb function, as represented by the
29376 P6_OV_BUILTIN_CMPB constant.
29377 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
29378 (BU_P6_64BIT_2): New macro.
29379 (BU_P6_OVERLOAD_2): New macro
29380 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
29381 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
29382 (CMPB): Add overload support to represent both 32-bit and 64-bit
29383 compare-bytes function.
29384 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
29385 support for TARGET_CMPB.
29386 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
29387 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29388 documentation of the __builtin_cmpb overloaded built-in function.
29390 2017-05-11 Richard Biener <rguenther@suse.de>
29392 PR tree-optimization/80705
29393 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
29394 bases are not vectorizable.
29396 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29398 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
29399 when counting register pressure.
29401 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29403 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
29404 (iv_ca_more_deps): Renamed to ...
29405 (iv_ca_compare_deps): ... this.
29406 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
29408 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29410 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
29412 (determine_group_iv_costs): ... here.
29413 (find_inv_vars_cb): Record inv var if it's not recorded before.
29415 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29417 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
29418 (get_shiftadd_cost): Ditto.
29420 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29422 * tree-ssa-address.c: Include header file.
29423 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
29425 (add_to_parts): Refactor.
29426 (addr_to_parts): New parameter. Update use of move_hint_to_base.
29427 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
29430 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29432 PR tree-optimization/53090
29433 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
29435 (extract_cond_operands): Detect condition with IV on both sides
29436 and return COMP_IV_EXPR_2.
29437 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
29438 (rewrite_use_compare): Simplify by removing call to function
29439 extract_cond_operands.
29441 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29443 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
29444 (extract_cond_operands): Detect condition comparing against non-
29445 invariant bound and return appropriate enum value.
29446 (find_interesting_uses_cond): Update use of extract_cond_operands.
29447 Handle its return value accordingly.
29448 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
29450 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29452 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
29453 nonlinear iv_use computation in loop invariant sensitive way.
29455 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29457 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
29458 (find_iv_candidates): Call relate_compare_use_with_all_cands.
29460 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29462 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
29463 (dump_cand): Support iv_cand.inv_exprs.
29464 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
29466 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
29469 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29471 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
29473 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
29474 as local function. Include necessary header files.
29475 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
29477 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29479 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
29481 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29483 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
29484 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
29485 RSHIFT_EXPR and BIT_NOT_EXPR.
29487 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29489 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
29490 (adjust_setup_cost): New parameter supporting round up adjustment.
29491 (struct address_cost_data): Delete.
29492 (force_expr_to_var_cost): Don't bound cost with spill_cost.
29493 (split_address_cost, ptr_difference_cost): Delete.
29494 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
29495 (struct ainc_cost_data): New struct.
29496 (get_address_cost_ainc): New function.
29497 (get_address_cost, get_computation_cost): Reimplement.
29498 (determine_group_iv_cost_address): Record inv_expr for all uses of
29500 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
29501 (iv_ca_has_deps): Reimplemented to ...
29502 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
29504 (iv_ca_extend): Call iv_ca_more_deps.
29506 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29508 * tree-ssa-address.c (struct mem_address): Move to header file.
29509 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
29510 * tree-ssa-address.h (struct mem_address): Move from C file.
29511 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
29513 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29515 * tree-affine.h (aff_combination_type): New interface.
29516 (aff_combination_zero_p): Remove static.
29517 (aff_combination_const_p): New interface.
29518 (aff_combination_singleton_var_p): New interfaces.
29520 2017-05-11 Richard Biener <rguenther@suse.de>
29522 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29523 Skip unreachable blocks and destinations.
29524 (eliminate): Move stmt removal and fixup ...
29525 (fini_eliminate): ... here. Skip inserted exprs.
29526 (pass_pre::execute): Move fini_pre after fini_eliminate.
29527 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
29528 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
29529 PRE to get rid of dead code that has invalid SSA form and
29530 split critical edges again.
29532 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29534 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29536 2017-05-11 Richard Biener <rguenther@suse.de>
29538 * passes.c (execute_function_todo): Verify loops if they are
29539 said to be up-to-date.
29540 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
29541 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
29543 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
29546 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
29547 handle calling assemble_external ourself.
29550 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
29551 modes with zero size. Enhance comment.
29553 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29555 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
29556 built-ins for vec_xl and vec_xst with short and char pointer
29559 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29561 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
29562 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
29563 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
29564 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
29565 (_mm_maskz_min_round_ss): New intrinsics.
29566 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
29567 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
29568 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
29569 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
29570 (__builtin_ia32_minss_mask_round): New builtins.
29571 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29572 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29573 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
29575 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
29576 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
29578 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
29581 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29583 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
29584 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
29585 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
29586 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
29587 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
29588 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
29589 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
29590 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29591 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29592 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
29593 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
29594 (__builtin_ia32_mulss_mask_round): New builtins.
29595 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29596 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29597 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
29599 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
29600 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29602 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29605 2017-05-10 Julia Koval <julia.koval@intel.com>
29607 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
29608 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
29609 (_mm256_setr_m128i): New intrinsics.
29611 2017-05-10 Julia Koval <julia.koval@intel.com>
29613 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
29614 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
29615 (_mm_maskz_rcp14_ss): New intrinsics.
29616 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
29617 (__builtin_ia32_rcp14ss_mask): New builtins.
29618 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
29620 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
29622 PR tree-optimization/51513
29623 * tree-cfg.c (gimple_seq_unreachable_p): New function.
29624 (assert_unreachable_fallthru_edge_p): Use it.
29625 (group_case_labels_stmt): Likewise.
29626 * tree-cfg.h: Prototype it.
29627 * stmt.c: Include cfghooks.h and tree-cfg.h.
29628 (emit_case_dispatch_table) <gap_label>: New local variable.
29629 Use it to fill dispatch table gaps.
29630 Test for default_label before updating probabilities.
29631 (expand_case) <default_label>: Remove unneeded initialization.
29632 Test for unreachable default case statement and remove its edge.
29633 Set default_label accordingly.
29634 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
29636 2017-05-10 Carl Love <cel@us.ibm.com>
29638 * config/rs6000/rs6000-c: Add support for built-in functions
29639 vector signed char vec_neg (vector signed char)
29640 vector signed short int vec_neg (vector short int)
29641 vector signed int vec_neg (vector signed int)
29642 vector signed long long vec_neg (vector signed long long)
29643 vector float vec_neg (vector float)
29644 vector double vec_neg (vector double)
29645 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
29647 * config/rs6000/altivec.h: Add define for vec_neg
29648 * doc/extend.texi: Update the built-in documentation for the
29649 new built-in functions.
29651 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29653 PR tree-optimization/77644
29654 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
29656 2017-05-10 Nathan Sidwell <nathan@acm.org>
29658 * dumpfile.h (TDI_lang_all): New.
29659 (TDF_KIND): New. Renumber others
29660 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
29662 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
29664 (get_dump_file_name): Adjust suffix generation.
29665 (dump_enable_all): Use TDF_KIND.
29666 * doc/invoke.texi (-fdump-lang-all): Document.
29668 * dumpfile.h: Tabify.
29670 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
29673 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
29674 Move member access before delete.
29676 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
29678 * tree-inline.c (expand_call_inline): Split block at stmt
29681 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
29684 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
29685 are now unused after splitting mov{sf,sd}_hardfloat.
29686 (f32_lr2): Likewise.
29687 (f32_lm): Likewise.
29688 (f32_lm2): Likewise.
29689 (f32_li): Likewise.
29690 (f32_li2): Likewise.
29691 (f32_lv): Likewise.
29692 (f32_sr): Likewise.
29693 (f32_sr2): Likewise.
29694 (f32_sm): Likewise.
29695 (f32_sm2): Likewise.
29696 (f32_si): Likewise.
29697 (f32_si2): Likewise.
29698 (f32_sv): Likewise.
29699 (f32_dm): Likewise.
29700 (f32_vsx): Likewise.
29701 (f32_av): Likewise.
29702 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
29703 For movsf, order stores so the VSX stores occur before the GPR
29704 store which encourages the register allocator to use a traditional
29705 FPR instead of a GPR. For movsd, order the stores so that the GPR
29706 store comes before the VSX stores to allow the power6 to work.
29707 This is due to the power6 not having a 32-bit integer store
29708 instruction from a FPR.
29709 (movsf_hardfloat): Likewise.
29710 (movsd_hardfloat): Likewise.
29712 2017-05-09 Martin Sebor <msebor@redhat.com>
29714 PR translation/80280
29715 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
29718 PR translation/80280
29719 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
29720 data member added in r247778.
29721 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
29723 2017-05-09 Nathan Sidwell <nathan@acm.org>
29725 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
29727 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
29730 2017-05-09 Marek Polacek <polacek@redhat.com>
29732 * doc/invoke.texi: Fix typo.
29734 2017-05-09 Richard Biener <rguenther@suse.de>
29736 * tree-vrp.c (vrp_val_is_max): Adjust comment.
29737 (vrp_val_is_min): Likewise.
29738 (set_value_range_to_value): Likewise.
29739 (set_value_range_to_nonnegative): Likewise.
29740 (gimple_assign_nonzero_p): Likewise.
29741 (gimple_stmt_nonzero_p): Likewise.
29742 (vrp_int_const_binop): Likewise. Remove unreachable case.
29743 (adjust_range_with_scev): Adjust comments.
29744 (compare_range_with_value): Likewise.
29745 (extract_range_from_phi_node): Likewise.
29746 (test_for_singularity): Likewise.
29748 2017-05-09 Richard Biener <rguenther@suse.de>
29750 * tree-vrp.c (get_single_symbol): Add assert that we don't
29751 get overflowed constants as invariant part.
29752 (compare_values_warnv): Add comment before the TREE_NO_WARNING
29753 checks. Use wi::cmp instead of recursing for integer constants.
29754 (compare_values): Just ignore whether we assumed undefined
29755 overflow instead of failing the compare.
29756 (extract_range_for_var_from_comparison_expr): Add comment before the
29757 TREE_NO_WARNING sets.
29758 (test_for_singularity): Likewise.
29759 (extract_range_from_comparison): Do not disable optimization
29760 when we assumed undefined overflow.
29761 (extract_range_basic): Remove init of unused var.
29763 2017-05-09 Richard Biener <rguenther@suse.de>
29765 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
29766 (extract_range_from_multiplicative_op_1): Adjust.
29767 (extract_range_from_binary_expr_1): Use int_const_binop.
29769 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
29772 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
29773 rs6000_store_data_bypass_p in seven define_bypass directives and
29774 in several comments.
29775 * config/rs6000/rs6000-protos.h: Add prototype for
29776 rs6000_store_data_bypass_p function.
29777 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
29778 function implements slightly different (rs6000-specific) semantics
29779 than store_data_bypass_p, returning false rather than aborting
29780 with assertion error when arguments do not satisfy the
29781 requirements of store data bypass.
29782 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
29783 rs6000_store_data_bypass_p.
29785 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
29787 * config/xtensa/xtensa-protos.h
29788 (xtensa_initial_elimination_offset): New declaration.
29789 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29790 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
29791 macro definition, add case for FRAME_POINTER_REGNUM when
29792 FRAME_GROWS_DOWNWARD.
29793 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
29794 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
29795 xtensa_initial_elimination_offset.
29797 2017-05-08 Nathan Sidwell <nathan@acm.org>
29799 * doc/invoke.texi: Alphabetize -fdump options.
29801 2017-05-08 Martin Sebor <msebor@redhat.com>
29803 PR translation/80280
29804 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
29806 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
29808 * target.def (compute_frame_layout): New optional target hook.
29809 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
29810 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
29811 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
29813 * reload1.c (verify_initial_elim_offsets): Likewise.
29814 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
29815 (use_simple_return_p): Call arm_compute_frame_layout if needed.
29816 (arm_get_frame_offsets): Split up into this ...
29817 (arm_compute_frame_layout): ... and this function.
29819 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
29821 * config/aarch64/constraints.md (Usa): New constraint.
29822 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
29824 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
29826 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
29827 with_multilib_list after it has been checked.
29829 2017-05-08 Richard Biener <rguenther@suse.de>
29831 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
29832 (bitmap_set_subtract_values): Likewise.
29834 2017-05-08 Richard Biener <rguenther@suse.de>
29836 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
29837 (gimple_assign_nonzero): ... this and remove strict_overflow_p
29839 (gimple_stmt_nonzero_warnv_p): Rename to ...
29840 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
29842 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
29843 (extract_range_basic): Adjust, do not disable propagation on
29844 strict overflow sensitive simplification.
29845 (vrp_visit_cond_stmt): Likewise.
29847 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
29849 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
29850 body size unconditionally.
29852 2017-05-07 Jeff Law <law@redhat.com>
29855 2017-05-06 Jeff Law <law@redhat.com>
29856 PR tree-optimization/78496
29857 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29860 PR tree-optimization/78496
29861 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29862 (simplify_stmt_using_ranges): Call it.
29863 (vrp_dom_walker::before_dom_children): Extract equivalences
29864 from an ASSERT_EXPR with an equality comparison against a
29867 2017-05-06 Jeff Law <law@redhat.com>
29869 PR tree-optimization/78496
29870 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29873 PR tree-optimization/78496
29874 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29875 (simplify_stmt_using_ranges): Call it.
29876 (vrp_dom_walker::before_dom_children): Extract equivalences
29877 from an ASSERT_EXPR with an equality comparison against a
29880 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29882 * lra-constraints.c (lra_copy_reg_equiv): New function.
29883 (split_reg): Use it to copy equivalence information from the
29884 original register to the spill register.
29886 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29888 PR rtl-optimization/75964
29889 * simplify-rtx.c (simplify_const_relational_operation): Remove
29890 invalid handling of comparisons of integer ABS.
29892 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
29894 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
29895 initialize to zero.
29896 (init_regs): Remove declaration.
29897 (function_arg_advance_32): Initialize error_p as boolean variable.
29899 2017-05-05 Nathan Sidwell <nathan@acm.org>
29901 * store-motion.c (remove_reachable_equiv_notes): Reformat long
29902 lines. Use for (;;).
29904 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29906 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
29907 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
29908 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
29909 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
29910 VF=2 that require versioning.
29912 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29914 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
29917 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29919 * diagnostic.h (diagnostic_override_option_index): Convert from
29920 macro to inline function.
29922 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29924 * diagnostic.c (last_module_changed_p): New function.
29925 (set_last_module): New function.
29926 (diagnostic_report_current_module): Convert macro usage to
29927 the above functions.
29928 * diagnostic.h (diagnostic_context::last_module): Strengthen
29929 from const line_map * to const line_map_ordinary *.
29930 (diagnostic_last_module_changed): Delete macro.
29931 (diagnostic_set_last_module): Delete macro.
29933 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29935 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29936 with diagnostic_report_diagnostic.
29937 (diagnostic_n_impl_richloc): Likewise.
29938 * diagnostic.h (report_diagnostic): Delete macro.
29939 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29940 with diagnostic_report_diagnostic.
29941 * substring-locations.c (format_warning_va): Likewise.
29943 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29945 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29946 save/restor of format_spec. Move option-printing code to...
29947 (print_option_information): ...this new function, and
29948 reimplement by simply printing to the pretty_printer,
29949 rather than appending to the format string.
29951 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29953 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29954 handling logic into...
29955 (update_effective_level_from_pragmas): ...this new function.
29957 2017-05-04 Andrew Waterman <andrew@sifive.com>
29959 * config/riscv/riscv.opt (mstrict-align): New option.
29960 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29961 (SLOW_UNALIGNED_ACCESS): Define.
29962 (riscv_slow_unaligned_access): Declare.
29963 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29965 (riscv_slow_unaligned_access): New variable.
29966 (rocket_tune_info): Set slow_unaligned_access to true.
29967 (optimize_size_tune_info): Set slow_unaligned_access to false.
29968 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29969 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29970 (riscv_option_override): Set riscv_slow_unaligned_access.
29971 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29973 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29975 * config/riscv/riscv.md: Unify indentation.
29977 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29982 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29984 (extendsi<mode>2): Add support for doing sign extension via
29985 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29986 don't have ISA 3.0 instructions.
29987 (extendsi<mode>2 splitter): Likewise.
29988 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29989 generate the normal insns since SImode can now go in vector
29990 registers. Disallow the special UNSPECs needed for previous
29991 machines to hide SImode being used. Add new insns
29992 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29993 (fix_trunc<mode>si2_stfiwx): Likewise.
29994 (fix_trunc<mode>si2_internal): Likewise.
29995 (fixuns_trunc<mode>si2): Likewise.
29996 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29997 (fctiw<u>z_<mode>_smallint): Likewise.
29998 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29999 of floating point to 32-bit integer from doing a direct move to
30000 the GPR registers to do a store.
30001 (fctiwz_<mode>): Break long line.
30003 2017-05-05 Bin Cheng <bin.cheng@arm.com>
30005 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
30006 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
30007 (addr_list, addr_offset_valid_p): New.
30008 (split_address_groups): Check offset validity with above function.
30009 (gt-tree-ssa-loop-ivopts.h): Include header file.
30011 2017-05-05 Nathan Sidwell <nathan@acm.org>
30013 * config.gcc (arm*-*-*): Add missing 'fi'.
30015 2017-05-05 Steve Ellcey <sellcey@cavium.com>
30017 * doc/invoke.texi (-fopt-info): Explicitly say order of options
30018 included in -fopt-info does not matter.
30019 * doc/optinfo.texi (-fopt-info): Fix description of default
30020 behavour. Explicitly say order of options included in -fopt-info
30023 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
30025 * config.gcc: Allow combinations of aprofile and rmprofile values for
30026 --with-multilib-list.
30027 * config/arm/t-multilib: New file.
30028 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
30029 variables. Remove setting of ISA and floating-point ABI in
30030 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
30031 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
30032 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
30033 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
30035 * config/arm/t-rmprofile: Likewise except for the matches changes.
30036 * doc/install.texi (--with-multilib-list): Document the combination of
30037 aprofile and rmprofile values and warn about pitfalls in doing that.
30039 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30041 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
30042 (movdi_aarch64): Likewise.
30044 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30046 PR tree-optimization/80632
30047 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
30049 (build_arrays): Initialize it for virtual phis.
30050 (fix_phi_nodes): Use it for virtual phis.
30052 PR tree-optimization/80558
30053 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
30054 [x, y] op z into [x op, y op z] for op & or | if conditions
30057 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
30058 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30061 * config/arm/arm.md (use_literal_pool): Remove.
30062 (64-bit immediate split): No longer takes cost into consideration
30063 if arm_disable_literal_pool is enabled.
30064 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
30065 used when arm_disable_literal_pool is enabled.
30066 (arm_max_const_double_inline_cost): Remove use of
30067 arm_disable_literal_pool.
30068 (push_minipool_fix): Add assert.
30069 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
30070 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
30071 (no_literal_pool_sf_immediate): New.
30073 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30075 PR tree-optimization/80613
30076 * tree-ssa-dce.c (propagate_necessity): Remove cases for
30077 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
30079 2017-05-05 Richard Biener <rguenther@suse.de>
30081 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
30083 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
30085 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
30086 of this flag from insn conditions due to removal from r247495.
30088 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30090 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
30092 (arm_early_store_addr_dep_ptr): Likewise.
30093 * config/arm/aarch-common-protos.h
30094 (arm_early_load_addr_dep_ptr): Add prototype.
30095 (arm_early_store_addr_dep_ptr): Likewise.
30096 * config/arm/cortex-a53.md: Add new bypasses.
30098 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30100 * tree.c (next_type_uid): Change type to unsigned.
30101 (type_hash_canon): Decrement back next_type_uid if
30102 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
30103 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
30106 2017-05-04 Martin Sebor <msebor@redhat.com>
30108 * builtins.c: Fix a trivial typo in a comment.
30110 PR middle-end/79234
30111 * builtins.c (check_sizes): Adjust to handle reading past the end.
30112 Avoid printing excessive upper bound of ranges. Use %E to print
30113 tree nodes instead of converting them to %wu.
30114 (expand_builtin_memchr): New function.
30115 (compute_dest_size): Rename...
30116 (compute_objsize): ...to this.
30117 (expand_builtin_memcpy): Adjust.
30118 (expand_builtin_mempcpy): Adjust.
30119 (expand_builtin_strcat): Adjust.
30120 (expand_builtin_strcpy): Adjust.
30121 (check_strncat_sizes): Adjust.
30122 (expand_builtin_strncat): Adjust.
30123 (expand_builtin_strncpy): Adjust and simplify.
30124 (expand_builtin_memset): Adjust.
30125 (expand_builtin_bzero): Adjust.
30126 (expand_builtin_memcmp): Adjust.
30127 (expand_builtin): Handle memcmp.
30128 (maybe_emit_chk_warning): Check strncat just once.
30130 2017-05-04 Martin Sebor <msebor@redhat.com>
30132 PR preprocessor/79214
30133 PR middle-end/79222
30134 PR middle-end/79223
30135 * builtins.c (check_sizes): Add inlining context and issue
30136 warnings even when -Wno-system-headers is set.
30137 (check_strncat_sizes): Same.
30138 (expand_builtin_strncat): Same.
30139 (expand_builtin_memmove): New function.
30140 (expand_builtin_stpncpy): Same.
30141 (expand_builtin): Handle memmove and stpncpy.
30143 2017-05-04 Bin Cheng <bin.cheng@arm.com>
30145 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
30146 which is not used any more.
30148 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30150 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
30152 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30154 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
30155 (cortexa53_tunings): Likewise.
30156 (cortexa57_tunings): Likewise.
30157 (cortexa72_tunings): Likewise.
30158 (cortexa73_tunings): Likewise.
30160 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30162 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
30163 Set loop alignment to 8.
30165 2017-05-04 Martin Sebor <msebor@redhat.com>
30167 PR translation/80280
30168 * builtins.c (expand_builtin_object_size): Add missing quoting to
30169 %D and like directives.
30170 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
30171 (hsa_type_for_tree_type): Same.
30172 (verify_function_arguments): Same.
30173 * symtab.c (symbol_table::change_decl_assembler_name): Same.
30174 * varasm.c (get_section): Same.
30177 2017-05-04 Martin Sebor <msebor@redhat.com>
30179 PR translation/80280
30180 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
30182 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30184 * config/aarch64/aarch64.c (generic_addrcost_table):
30185 Change HI/TI mode setting.
30187 2017-05-04 Martin Jambor <mjambor@suse.cz>
30189 PR tree-optimization/80622
30190 * tree-sra.c (comes_initialized_p): New function.
30191 (build_accesses_from_assign): Only set write lazily when
30192 comes_initialized_p is false.
30193 (analyze_access_subtree): Use comes_initialized_p.
30194 (propagate_subaccesses_across_link): Assert !comes_initialized_p
30195 instead of testing for PARM_DECL.
30197 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30199 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
30200 constraint on operand 0 to allow more general addressing modes.
30201 Adjust output template.
30202 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
30204 * config/aarch64/aarch64-protos.h
30205 (aarch64_address_valid_for_prefetch_p): Declare prototype.
30206 * config/aarch64/constraints.md (Dp): New address constraint.
30207 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
30210 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
30212 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
30213 update use of estimate_ipcp_clone_size_and_time.
30214 (estimate_local_effects): Update use of
30215 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
30216 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
30217 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
30218 Return nonspecialized time.
30220 2017-05-04 Richard Biener <rguenther@suse.de>
30222 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
30223 for the last VUSE which def dominates the PHI. Directly call
30225 (get_continuation_for_phi_1): Remove.
30227 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
30229 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
30230 to explain the use of truncating division. Cap the number of
30231 iterations to the maximum given by nb_iterations_upper_bound,
30234 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30236 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
30237 * configure: Regenerate.
30238 * config.in: Regenerate.
30239 * config/i386/driver-mingw32.c: new file.
30240 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
30241 * config.host: Link driver-mingw32.o on MinGW host.
30242 * doc/install.texi: Document new --enable-mingw-wildcard configure
30245 2017-05-04 Marek Polacek <polacek@redhat.com>
30247 PR tree-optimization/80612
30248 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
30250 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30251 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
30253 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
30254 (movt splitter): Likewise.
30255 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
30256 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
30257 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
30258 block for Thumb-1 with MOVT.
30259 (thumb2_legitimate_address_p): Move code block ...
30260 (can_avoid_literal_pool_for_label_p): ... into this new function.
30261 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
30263 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
30264 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
30265 "M-profile targets with the MOVT instruction".
30267 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30269 * config/arm/arm-builtins.c (arm_init_builtins): Rename
30270 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
30271 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
30273 2017-05-04 Martin Liska <mliska@suse.cz>
30275 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
30276 variable cond_code.
30278 2017-05-04 Richard Biener <rguenther@suse.de>
30280 * tree.c (array_at_struct_end_p): Handle arrays at struct
30281 end with flexarrays more conservatively. Refactor and treat
30282 arrays of arrays or aggregates more strict. Fix
30283 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
30284 * tree.c (array_at_struct_end_p): Adjust prototype.
30285 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
30286 * gimple-fold.c (get_range_strlen): Likewise.
30287 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
30289 2017-05-04 Richard Biener <rguenther@suse.de>
30291 PR tree-optimization/31130
30292 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
30294 (supports_overflow_infinity): Likewise.
30295 (is_negative_overflow_infinity): Likewise.
30296 (is_positive_overflow_infinity): Likewise.
30297 (is_overflow_infinity): Likewise.
30298 (stmt_overflow_infinity): Likewise.
30299 (overflow_infinity_range_p): Likewise.
30300 (usable_range_p): Remove as always returning true.
30301 (make_overflow_infinity): Remove.
30302 (negative_overflow_infinity): Likewise.
30303 (positive_overflow_infinity): Likewise.
30304 (avoid_overflow_infinity): Likewise.
30305 (set_value_range): Adjust accordingly.
30306 (set_value_range_to_nonnegative): Likewise, remove now unused
30307 overflow_infinity arg.
30308 (vrp_operand_equal_p): Adjust.
30309 (update_value_range): Likewise.
30310 (range_int_cst_singleton_p): Likewise.
30311 (operand_less_p): Likewise.
30312 (compare_values_warnv): Likewise.
30313 (extract_range_for_var_from_comparison_expr): Likewise.
30314 (vrp_int_const_binop): Likewise.
30315 (zero_nonzero_bits_from_vr): Likewise.
30316 (extract_range_from_multiplicative_op_1): Likewise.
30317 (extract_range_from_binary_expr_1): Likewise.
30318 (extract_range_from_unary_expr): Likewise.
30319 (extract_range_from_comparison): Likewise.
30320 (extract_range_basic): Likewise.
30321 (adjust_range_with_scev): Likewise.
30322 (compare_ranges): Likewise.
30323 (compare_range_with_value): Likewise.
30324 (dump_value_range): Likewise.
30325 (test_for_singularity): Likewise, remove strict_overflow_p parameter
30327 (simplify_cond_using_ranges): Adjust.
30329 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30331 * brig-builtins.def: Added a builtin for class_f64.
30332 * builtin-types.def: Added a builtin type needed by class_f64.
30334 2017-05-03 Jason Merrill <jason@redhat.com>
30336 * timevar.def: Add TV_CONSTEXPR.
30338 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30340 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
30342 2017-05-03 Martin Jambor <mjambor@suse.cz>
30344 * ipa-prop.c (ipa_update_after_lto_read): Removed.
30345 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
30346 * ipa-cp.c (ipcp_propagate_stage): Do not call
30347 ipa_update_after_lto_read.
30348 * ipa-inline.c (ipa_inline): Likewise.
30350 2017-05-03 Martin Jambor <mjambor@suse.cz>
30352 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
30353 tag. Added a default constructor and a destructor.
30354 (ipa_edge_args_sum_t): New class;
30355 (ipa_edge_args_sum): Declare.
30356 (ipa_edge_args_vector): Remove declaration.
30357 (IPA_EDGE_REF): Use ipa_edge_args_sum.
30358 (ipa_free_edge_args_substructures): Remove declaration.
30359 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
30360 (ipa_edge_args_info_available_for_edge_p): Likewise.
30361 * ipa-prop.c (ipa_edge_args_vector): Removed.
30362 (edge_removal_hook_holder): Likewise.
30363 (edge_duplication_hook_holder): Likewise.
30364 (ipa_edge_args_sum): New variable.
30365 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
30366 ipa_edge_args_vector.
30367 (ipa_free_edge_args_substructures): Likewise.
30368 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
30369 ipa_edge_args_vector.
30370 (ipa_edge_removal_hook): Turned into method
30371 ipa_edge_args_sum_t::remove.
30372 (ipa_edge_duplication_hook): Turned into method
30373 ipa_edge_args_sum_t::duplicate.
30374 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
30375 registering edge hooks.
30376 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
30377 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
30378 ipa_edge_args_sum instead of ipa_edge_args_vector.
30379 * ipa-profile.c (ipa_profile): Likewise.
30381 2017-05-03 Martin Jambor <mjambor@suse.cz>
30383 * symbol-summary.h (function_summary): New method exists.
30384 (function_summary::symtab_removal): Deallocate through release.
30385 (call_summary): New class.
30386 (gt_ggc_mx): New overload.
30387 (gt_pch_nx): Likewise.
30388 (gt_pch_nx): Likewise.
30390 2017-05-03 Jeff Law <law@redhat.com>
30392 PR tree-optimization/78496
30393 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
30394 from simplify_cond_using_ranges. Split off code to walk
30395 backwards through casts into ...
30396 (simplify_cond_using_ranges_2): New function.
30397 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
30398 (execute_vrp): After identifying jump threads, call
30399 simplify_cond_using_ranges_2.
30401 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
30404 * ipa-inline.h (inline_summary): Add ctor.
30405 (create_ggc): Do not use ggc_cleared_alloc.
30407 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
30408 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30410 * gcc.c (handle_braces): Support escaping in switch matching
30412 * doc/invoke.texi (Spec Files): Document it.
30413 Remove superfluous @code markup in items.
30415 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30417 * diagnostic-show-locus.c (struct column_range): New struct.
30418 (get_affected_columns): New function.
30419 (get_printed_columns): New function.
30420 (struct correction): New struct.
30421 (correction::ensure_capacity): New function.
30422 (correction::ensure_terminated): New function.
30423 (struct line_corrections): New struct.
30424 (line_corrections::~line_corrections): New dtor.
30425 (line_corrections::add_hint): New function.
30426 (layout::print_trailing_fixits): Reimplement in terms of the new
30428 (selftest::test_overlapped_fixit_printing): New function.
30429 (selftest::diagnostic_show_locus_c_tests): Call it.
30431 2017-05-03 Nathan Sidwell <nathan@acm.org>
30433 Canonicalize canonical type hashing
30434 * tree.h (type_hash_canon_hash): Declare.
30435 * tree.c (type_hash_list, attribute_hash_list): Move into
30436 type_hash_canon_hash.
30437 (build_type_attribute_qual_variant): Break out hash code calc into
30438 type_hash_canon_hash.
30439 (type_hash_canon_hash): New. Generic type hash computation.
30440 (build_range_type_1, build_array_type_1, build_function_type,
30441 build_method_type_directly, build_offset_type, build_complex_type,
30442 make_vector_type): Call it.
30444 2017-05-03 Richard Biener <rguenther@suse.de>
30446 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
30447 When all DRs have unknown misaligned do not always peel
30448 when there is a store but apply the same costing model as if
30449 there were only loads.
30451 2017-05-03 Richard Biener <rguenther@suse.de>
30454 PR tree-optimization/80492
30455 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30456 compare_base_decls returning dont-know properly.
30458 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
30460 * config/arm/iterators.md (CCSI): New mode iterator.
30461 (arch): New mode attribute.
30462 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
30463 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
30464 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
30465 code iterator for success result mode.
30466 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
30467 the corresponding new insn generators.
30469 2017-05-03 Bin Cheng <bin.cheng@arm.com>
30472 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30473 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30475 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
30477 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
30478 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
30479 (DDR_A): Wrap DDR argument in brackets.
30480 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
30481 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
30482 (DDR_REVERSED_P): Likewise.
30484 2017-05-03 Jakub Jelinek <jakub@redhat.com>
30486 PR tree-optimization/79472
30487 * tree-switch-conversion.c (struct switch_conv_info): Add
30488 contiguous_range and default_case_nonstandard fields.
30489 (collect_switch_conv_info): Compute contiguous_range and
30490 default_case_nonstandard fields, don't clear final_bb if
30491 contiguous_range and only the default case doesn't have the required
30493 (check_all_empty_except_final): Set default_case_nonstandard instead
30494 of failing if contiguous_range and the default case doesn't have empty
30496 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
30497 and only the default case doesn't have the required constants. Skip
30499 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
30500 if default_case_nonstandard.
30501 (build_constructors): Build constant 1 just once. Assert that default
30502 values aren't inserted in between cases if contiguous_range. Skip
30504 (build_arrays): Skip virtual phis.
30505 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
30506 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
30507 Handle virtual phis.
30508 (gen_inbound_check): Handle default_case_nonstandard case.
30509 (process_switch): Adjust check_final_bb caller. Call
30510 gather_default_values with the first non-default case instead of
30511 default case if default_case_nonstandard.
30513 2017-05-02 Nathan Sidwell <nathan@acm.org>
30515 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
30516 check. Fix formatting.
30518 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
30520 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
30521 errors when comparing specialized and unspecialized times.
30523 2017-05-02 David Malcolm <dmalcolm@redhat.com>
30525 * diagnostic-show-locus.c
30526 (layout::should_print_annotation_line_p): Make private.
30527 (layout::print_annotation_line): Make private.
30528 (layout::annotation_line_showed_range_p): Make private.
30529 (layout::show_ruler): Make private.
30530 (layout::print_source_line): Make private. Pass in line and
30531 line_width, rather than calling location_get_source_line. Drop
30533 (layout::print_leading_fixits): New method.
30534 (layout::print_any_fixits): Rename to...
30535 (layout::print_trailing_fixits): ...this, and make private.
30536 Don't print newline fixits.
30537 (diagnostic_show_locus): Move logic for printing one row into...
30538 (layout::print_line): ...this new function. Move the
30539 location_get_source_line call and error-handling from
30540 print_source_line to here. Call print_leading_fixits, and rename
30541 print_any_fixits to print_trailing_fixits.
30542 (selftest::test_fixit_insert_containing_newline): Update now that
30543 newlines are partially supported.
30544 (selftest::test_fixit_insert_containing_newline_2): New test.
30545 (selftest::test_fixit_replace_containing_newline): Update comments.
30546 (selftest::diagnostic_show_locus_c_tests): Call the new test.
30547 * edit-context.c (class added_line): New class.
30548 (class edited_line): Describe newline handling in comment.
30549 (edited_line::actually_edited_p): New method.
30550 (edited_line::print_content): Delete redundant decl.
30551 (edited_line::m_predecessors): New field.
30552 (edited_file::print_content): Call edited_line::print_content.
30553 (edited_file::print_diff): Update to support newlines.
30554 (edited_file::print_diff_hunk): Likewise.
30555 (edited_file::print_run_of_changed_lines): New function.
30556 (edited_file::print_diff_line): Convert to...
30557 (print_diff_line): ...this.
30558 (edited_file::get_effective_line_count): New function.
30559 (edited_line::edited_line): Initialize new field m_predecessors.
30560 (edited_line::~edited_line): Clean up m_predecessors.
30561 (edited_line::apply_fixit): Handle newlines.
30562 (edited_line::get_effective_line_count): New function.
30563 (edited_line::print_content): New function.
30564 (edited_line::print_diff_lines): New function.
30565 (selftest::test_applying_fixits_insert_containing_newline): New
30567 (selftest::test_applying_fixits_replace_containing_newline): New
30569 (selftest::insert_line): New function.
30570 (selftest::test_applying_fixits_multiple_lines): Add example of
30572 (selftest::edit_context_c_tests): Call the new tests.
30574 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30576 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
30577 parameter cand. Update dump information.
30578 (get_computation_cost): Update uses.
30580 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30582 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
30583 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
30584 (get_computation_at, rewrite_use_address): Update use of
30585 get_computation_aff.
30587 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30589 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
30590 (get_computation): Delete.
30591 (get_computation_cost): Implement like get_computation_cost_at.
30592 Use get_computation_at.
30593 (get_computation_cost_at): Delete.
30594 (rewrite_use_nonlinear_expr): Use get_computation_at.
30595 (rewrite_use_compare, remove_unused_ivs): Ditto.
30597 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30599 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
30601 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30603 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
30604 (ivopts_global_cost_for_size): Rename parameter and update uses.
30605 (iv_ca_recount_cost): Update uses.
30606 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
30607 candidates seperately in n_invs and n_cands.
30608 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
30610 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30612 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
30613 (find_inv_vars_cb): New.
30614 (find_depends): Renamed to ...
30615 (find_inv_vars): ... this.
30616 (add_candidate_1, force_var_cost): Call find_inv_vars.
30617 (split_address_cost, determine_group_iv_cost_cond): Ditto.
30619 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30621 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
30622 inv_vars. Add inv_exprs.
30623 (struct iv_cand): Rename depends_on to inv_vars.
30624 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
30625 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
30626 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
30627 (dump_cand): Dump inv_vars.
30628 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
30629 (record_invariant, find_depends, add_candidate_1): Ditto.
30630 (set_group_iv_cost, force_var_cost): Ditto.
30631 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
30632 (get_computation_cost_at, get_computation_cost): Ditto.
30633 (determine_group_iv_cost_generic): Ditto.
30634 (determine_group_iv_cost_address): Ditto.
30635 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
30636 (determine_group_iv_costs): Ditto.
30637 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
30638 (iv_ca_set_remove_invariants): Renamed to ...
30639 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
30640 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
30641 (iv_ca_set_add_invariants): Renamed to ...
30642 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
30643 (iv_ca_set_cp): Use iv_ca_set_add_invs.
30644 (iv_ca_has_deps): Support inv_vars and inv_exprs.
30645 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
30646 (create_new_ivs): Remove useless dump.
30648 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30650 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
30652 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
30653 (iv_ca_set_no_cp, create_new_iv): Ditto.
30655 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30657 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30659 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30661 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
30662 function tree_check2.
30664 2017-05-02 Martin Liska <mliska@suse.cz>
30666 * doc/gcov.texi: Add missing preposition.
30667 * gcov.c (function_info::function_info): Properly fill up
30668 all member variables.
30670 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30672 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
30674 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30676 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
30678 2017-05-02 Martin Liska <mliska@suse.cz>
30681 * lto-streamer-in.c (lto_read_tree_1): Remove
30682 LTO_STREAMER_DEBUG.
30683 * lto-streamer.c (struct tree_hash_entry): Likewise.
30684 (struct tree_entry_hasher): Likewise.
30685 (tree_entry_hasher::hash): Likewise.
30686 (tree_entry_hasher::equal): Likewise.
30687 (lto_streamer_init): Likewise.
30688 (lto_orig_address_map): Likewise.
30689 (lto_orig_address_get): Likewise.
30690 (lto_orig_address_remove): Likewise.
30691 * lto-streamer.h: Likewise.
30692 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
30693 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
30695 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
30697 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
30698 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
30699 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
30700 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
30701 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
30702 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
30703 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
30704 (mm_maskz_sub_ss): New intrinsics.
30705 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30706 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30707 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
30708 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
30709 (__builtin_ia32_subss_mask_round): New builtins.
30710 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30711 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30712 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
30714 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
30715 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30717 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30720 2017-05-02 Martin Jambor <mjambor@suse.cz>
30722 PR tree-optimization/78687
30723 * tree-sra.c (access): New field parent.
30724 (process_subtree_disqualification): New function.
30725 (disqualify_candidate): Call it.
30726 (build_accesses_from_assign): Reset write flag if creating an
30728 (build_access_subtree): Fill in parent field and also prpagate
30729 down grp_write flag.
30730 (create_artificial_child_access): New parameter set_grp_write, set
30731 grp_write to its value.
30732 (propagate_subaccesses_across_link): Also propagate grp_write flag
30734 (propagate_all_subaccesses): Push the closest parent back to work
30735 queue if add_access_to_work_queue returned true.
30737 2017-05-02 Richard Biener <rguenther@suse.de>
30739 * common.opt (fstrict-overflow): Alias negative to fwrapv.
30740 * doc/invoke.texi (fstrict-overflow): Remove all traces of
30741 -fstrict-overflow documentation.
30742 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
30743 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
30744 flag_strict_overflow.
30745 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
30746 * lto-opts.c (lto_write_options): Do not stream it.
30747 * lto-wrapper.c (merge_and_complain): Do not handle it.
30748 * opts.c (default_options_table): Do not set -fstrict-overflow.
30749 (finish_options): Likewise do not clear it when sanitizing.
30750 * simplify-rtx.c (simplify_const_relational_operation): Do not
30751 test flag_strict_overflow.
30753 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30755 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
30756 using enabled attribute.
30757 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
30758 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
30759 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
30760 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
30761 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
30762 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
30763 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
30764 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
30765 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
30766 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
30768 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30770 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
30772 2017-05-02 Richard Biener <rguenther@suse.de>
30774 PR tree-optimization/80591
30776 2017-04-10 Richard Biener <rguenther@suse.de>
30778 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30781 2017-05-02 Richard Biener <rguenther@suse.de>
30783 PR tree-optimization/80549
30784 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
30785 (cleanup_tree_cfg_noloop): Create forwarders to known loop
30786 headers if they do not have a preheader.
30788 2017-05-02 Martin Liska <mliska@suse.cz>
30791 * common.opt: Fix typo.
30792 * doc/invoke.texi: Likewise.
30794 2017-05-01 Jan Beulich <jbeulich@suse.com>
30796 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
30797 swapping, add (x,x,m,x,n) alternative.
30799 2017-05-01 Nathan Sidwell <nathan@acm.org>
30801 * calls.c (combine_pending_stack_adjustment_and_call): Remove
30802 unnecessary unadjusted_alignment check.
30804 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30807 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
30809 * gimplify.c (gimplify_cilk_detach): New function.
30810 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
30811 * tree-core.h: Document EXPR_CILK_SPAWN.
30812 * tree.h (EXPR_CILK_SPAWN): Define.
30814 2017-05-01 David Malcolm <dmalcolm@redhat.com>
30816 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
30817 to use new fixit_hint representation, using the "replace" logic.
30818 (get_line_span_for_fixit_hint): Likewise.
30819 (layout::print_any_fixits): Likewise.
30820 (selftest::test_one_liner_many_fixits): Rename to...
30821 (selftest::test_one_liner_many_fixits_1): ...this, and update
30822 comment and expected output to reflect that the multiple fix-it
30823 hints are now consolidated into one insertion.
30824 (selftest::test_one_liner_many_fixits_2): New test.
30825 (selftest::test_diagnostic_show_locus_one_liner): Update for
30827 (selftest::test_fixit_consolidation): Update for fix-it API
30829 * diagnostic.c (print_parseable_fixits): Likewise.
30830 * edit-context.c (edited_line::m_line_events): Convert from
30831 auto_vec <line_event *> to auto_vec <line_event>.
30832 (class line_event): Convert from abstract base class to a concrete
30833 class, taking over the role of replace_event.
30834 (class insert_event): Delete.
30835 (class replace_event): Rename to class line_event. Convert to
30837 (edit_context::add_fixits): Reimplement.
30838 (edit_context::apply_insert): Delete.
30839 (edit_context::apply_replace): Rename to...
30840 (edit_context::apply_fixit): ...this. Convert to half-open range.
30841 (edited_file::apply_insert): Delete.
30842 (edited_file::apply_replace): Rename to...
30843 (edited_file::apply_fixit): ...this.
30844 (edited_line::~edited_line): Drop deletion of events.
30845 (edited_line::apply_insert): Delete.
30846 (edited_line::apply_replace): Rename to...
30847 (edited_line::apply_fixit): ...this. Convert to half-open range.
30848 Update for change to type of m_line_events.
30849 * edit-context.h (edit_context::apply_insert): Delete.
30850 (edit_context::apply_replace): Rename to...
30851 (edit_context::apply_fixit): ...this.
30853 2017-05-01 Martin Sebor <msebor@redhat.com>
30855 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
30858 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
30861 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
30862 __get_cpuid_max returns 0.
30863 (__get_cpuid_count): Ditto.
30865 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
30867 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
30868 replacement expression is another instance of one of its arguments.
30870 2017-05-01 Jakub Jelinek <jakub@redhat.com>
30873 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
30874 check for stack push/pop autoinc.
30875 * config/i386/i386.c (ix86_agi_dependent): Return false
30876 if the only reason why modified_in_p returned true is that
30877 addr is SP based and set_insn is a push or pop.
30879 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30881 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
30884 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30887 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
30888 (account_size_time): Use two predicates - exec_pred and
30890 (evaluate_conditions_for_known_args): Compute both clause and
30892 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
30893 (inline_summary_t::duplicate): Update.
30894 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
30896 (compute_inline_parameters): Likewise.
30897 (estimate_edge_size_and_time): Update caluclation of time.
30898 (estimate_node_size_and_time): Compute both time and nonspecialized
30900 (estimate_ipcp_clone_size_and_time): Update.
30901 (inline_merge_summary): Update.
30902 (do_estimate_edge_time): Update.
30903 (do_estimate_edge_size): Update.
30904 (do_estimate_edge_hints): Update.
30905 (inline_read_section, inline_write_summary): Stream both new predicates.
30906 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
30908 (compute_inlined_call_time): Cleanup.
30909 (big_speedup_p): Update.
30910 (edge_badness): Update.
30911 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
30912 (size_time_entry): Replace predicate by exec_predicate and
30913 nonconst_predicate.
30914 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
30915 (estimate_edge_time): Return also nonspec_time.
30916 (reset_edge_growth_cache): Update.
30918 2017-04-29 Jakub Jelinek <jakub@redhat.com>
30920 PR rtl-optimization/80491
30921 * ifcvt.c (noce_process_if_block): When looking for x setter
30922 with missing else_bb, don't check only the insn right before
30923 cond_earliest, but look for the last insn that x is modified in
30924 within the same bb.
30926 PR rtl-optimization/80491
30927 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30929 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30931 PR tree-optimization/80487
30932 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30934 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30936 PR tree-optimization/79697
30937 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30938 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30939 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30941 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30942 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30944 2017-04-28 Martin Sebor <msebor@redhat.com>
30946 PR tree-optimization/80523
30947 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30948 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30950 (maybe_warn, format_directive, parse_directive): Use new functions.
30951 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30953 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30955 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30957 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30959 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30960 target_header_dir): Set correctly.
30961 * configure: Regenerated.
30962 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30963 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30964 instead of SYSTEM_HEADER_DIR.
30966 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30968 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30969 (estimate_local_effects): Likewise.
30970 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30971 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30972 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30973 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30974 do_estimate_edge_time, estimate_edge_time): Likewise.
30975 * ipa-inline-analysis.c (estimate_node_size_and_time,
30976 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30977 (estimate_time_after_inlining): Remove.
30979 2017-04-28 Martin Liska <mliska@suse.cz>
30981 * doc/gcov.texi: Enhance documentation of gcov.
30983 2017-04-28 Martin Liska <mliska@suse.cz>
30985 * doc/gcov.texi: Sort options in alphabetic order.
30986 * doc/gcov-dump.texi: Likewise.
30987 * doc/gcov-tool.texi: Likewise.
30988 * gcov.c (print_usage): Likewise.
30989 * gcov-dump.c (print_usage): Likewise.
30990 * gcov-tool.c (print_merge_usage_message): Likewise.
30991 (print_rewrite_usage_message): Likewise.
30992 (print_overlap_usage_message): Likewise.
30994 2017-04-28 Martin Liska <mliska@suse.cz>
30996 PR gcov-profile/53915
30997 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30999 2017-04-28 Martin Liska <mliska@suse.cz>
31001 PR gcov-profile/79891
31002 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
31003 is marked by compiler as living on a line.
31004 (get_cycles_count): Remove usage of the union.
31005 (output_intermediate_file): Likewise.
31006 (find_source): Fix GNU coding style.
31007 (accumulate_line_counts): Remove old non-all block mode.
31008 (output_lines): Remove usage of the union.
31009 * profile.c (output_location): Include all BBs, even if
31010 belonging to a same line (and file) as a previous BB.
31012 2017-04-28 Martin Liska <mliska@suse.cz>
31014 * gcov.c (process_args): Handle new argument 'w'.
31015 (read_graph_file): Assign ID to BBs.
31016 (output_branch_count): Display BB # if verbose flag is set.
31017 (output_lines): Likewise for arcs.
31018 (print_usage): Add '--verbose' option help.
31019 * doc/gcov.texi: Document --verbose (-w) option.
31021 2017-04-28 Martin Liska <mliska@suse.cz>
31023 * gcov.c (struct block_location_info): New struct.
31024 (process_file): Fill up the new structure.
31025 (read_graph_file): Replace usage of encoding by the newly added
31027 (add_line_counts): Likewise.
31028 (accumulate_line_counts): Remove usage of the union.
31029 (function_info::function_info): New function.
31030 (function_info::~function_info): Likewise.
31031 (process_file): Call delete instead of release_function.
31032 (release_function): Release the function.
31033 (release_structures): Call delete instead of release_function.
31034 (solve_flow_graph): Replace usage of num_blocks.
31035 (find_exception_blocks): Likewise.
31036 (output_lines): Fix GNU coding style.
31038 2017-04-28 Martin Liska <mliska@suse.cz>
31041 * coverage.c (coverage_remove_note_file): New function.
31042 * coverage.h: Declare the function.
31043 * toplev.c (finalize): Clean if an error has been seen.
31045 2017-04-28 Martin Liska <mliska@suse.cz>
31047 PR gcov-profile/80031
31048 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
31049 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
31050 * gcov.c (read_graph_file): Read just number of blocks.
31051 * profile.c (branch_prob): Do not stream 0 flags per a basic
31054 2017-04-28 Martin Liska <mliska@suse.cz>
31056 * gcov-dump.c (tag_*): Add new argument to declarations.
31057 (dump_gcov_file): Likewise.
31058 (tag_blocks): Add and use new argument depth.
31059 (tag_arcs): Likewise.
31060 (tag_lines): Likewise.
31061 (tag_counters): Likewise.
31062 (tag_summary): Likewise.
31063 (dump_working_sets): Use depth to do a proper indentation.
31065 2017-04-28 Jakub Jelinek <jakub@redhat.com>
31068 * cgraph.h (symtab_node::debug_symtab): No longer inline.
31069 * symtab.c (symtab_node::debug_symtab): Move definition here.
31071 2017-04-28 Richard Biener <rguenther@suse.de>
31073 * lto-streamer.h (LTO_major_version): Bump to 7.
31075 2017-04-28 Richard Biener <rguenther@suse.de>
31077 * tree-vrp.c (assert_info): New struct.
31078 (add_assert_info): New helper.
31079 (register_edge_assert_for_2): Refactor to add asserts to a vector
31081 (register_edge_assert_for_1): Likewise.
31082 (register_edge_assert_for): Likewise.
31083 (finish_register_edge_assert_for): New helper actually registering
31084 asserts where live on edge.
31085 (find_conditional_asserts): Adjust.
31086 (find_switch_asserts): Likewise.
31087 (evrp_dom_walker::try_find_new_range): Generalize.
31088 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
31090 2017-04-27 Marek Polacek <polacek@redhat.com>
31093 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
31094 arg10 and arg11 to itype.
31096 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
31098 * doc/extend.texi (Object Size Checking): Improve grammar.
31100 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
31103 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
31104 that the logic for permitting reciprocal estimates matches that
31107 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31110 * tree.c (type_cache_hasher::equal): Only compare
31111 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
31112 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
31113 non-aggregate element types.
31114 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
31115 about the flag on ARRAY_TYPEs in the comment, formatting fix.
31117 2017-04-27 Richard Biener <rguenther@suse.de>
31119 PR middle-end/80533
31120 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
31121 stripping ARRAY_REFs from MEM_EXPR make sure we're not
31122 keeping a reference to a trailing array.
31124 2017-04-27 Richard Biener <rguenther@suse.de>
31126 PR middle-end/80539
31127 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
31128 being in loop-closed SSA form conservatively.
31129 (chrec_fold_multiply_poly_poly): Likewise.
31131 2017-04-27 Tamar Christina <tamar.christina@arm.com>
31133 PR middle-end/79665
31134 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
31135 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
31137 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31140 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
31141 (aarch64_function_arg_alignment): Return unsigned int again, but still
31142 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
31143 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31144 Don't emit -Wpsabi note.
31145 (aarch64_function_arg_boundary): Likewise.
31146 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31149 2017-04-26 Nathan Sidwell <nathan@acm.org>
31151 * tree.h (crc32_unsigned_n): Declare.
31152 (crc32_unsigned, crc32_unsigned): Make inline.
31153 * tree.c (crc32_unsigned_bits): Replace with ...
31154 (crc32_unsigned_n): ... this.
31155 (crc32_unsigned, crc32_byte): Remove.
31156 (crc32_string): Remove unnecessary braces.
31158 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31160 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
31161 * ipa-inline-analysis.c (MAX_TIME): Remove.
31162 (account_size_time): Use sreal for time.
31163 (dump_inline_summary): Update.
31164 (estimate_function_body_sizes): Update.
31165 (estimate_edge_size_and_time): Update.
31166 (estimate_calls_size_and_time): Update.
31167 (estimate_node_size_and_time): Update.
31168 (inline_merge_summary): Update.
31169 (inline_update_overall_summary): Update.
31170 (estimate_time_after_inlining): Update.
31171 (inline_read_section): Update.
31172 (inline_write_summary): Update.
31173 * ipa-inline.c (compute_uninlined_call_time): Update.
31174 (compute_inlined_call_time): Update.
31175 (recursive_inlining): Update.
31176 (inline_small_functions): Update.
31177 (dump_overall_stats): Update.
31178 * ipa-inline.h: Include sreal.h.
31179 (size_time_entry): Turn time to sreal.
31180 (inline_summary): Turn self_time nad time to sreal.
31182 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31184 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
31186 (sreal::stream_out, sreal::stream_in): New.
31187 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
31189 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31191 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
31194 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
31197 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
31199 (dimode_scalar_chain::compute_convert_gain): Ditto.
31200 (dimode_scalar_chain::make_vector_copies): Ditto.
31201 (dimode_scalar_chain::convert_reg): Ditto.
31202 (dimode_scalar_chain::convert_insn): Ditto.
31203 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
31204 (VI248_AVX512BW_1): New mode iterator.
31205 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
31206 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
31209 2017-04-25 Martin Sebor <msebor@redhat.com>
31211 PR tree-optimization/80497
31212 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
31213 constants are representable in HOST_WIDE_INT.
31214 (parse_directive): Ditto.
31216 2017-04-25 Martin Sebor <msebor@redhat.com>
31219 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
31220 (new_zero_array): Adjust signature.
31221 (dom_info::dom_init): Used unsigned rather that size_t.
31222 (dom_info::dom_info): Same.
31224 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31225 Jakub Jelinek <jakub@redhat.com>
31228 * config/arm/arm.c: Include gimple.h.
31229 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31230 returns negative, increment ncrn only if it returned positive.
31231 (arm_needs_doubleword_align): Return int instead of bool,
31232 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
31233 members, but if there is any such non-FIELD_DECL
31234 > PARM_BOUNDARY aligned decl, return -1 instead of false.
31235 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31236 returns negative, increment nregs only if it returned positive.
31237 (arm_setup_incoming_varargs): Likewise.
31238 (arm_function_arg_boundary): Emit -Wpsabi note if
31239 arm_needs_doubleword_align returns negative, return
31240 DOUBLEWORD_ALIGNMENT only if it returned positive.
31242 2017-04-25 Marek Polacek <polacek@redhat.com>
31245 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31246 first argument to type.
31248 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
31251 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
31252 type checks to test for compatibility instead of equality.
31254 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31255 Jakub Jelinek <jakub@redhat.com>
31258 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
31260 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
31261 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
31262 the alignment computation, but return their maximum in warn_alignment.
31263 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31264 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
31266 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
31267 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31270 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31272 * config/arc/simdext.md (dmpyh): Fix typo.
31274 2017-04-25 Richard Biener <rguenther@suse.de>
31276 PR tree-optimization/80492
31277 * alias.c (compare_base_decls): Handle registers with asm
31278 specification conservatively.
31279 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31280 compare_base_decls returning dont-know properly.
31282 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31284 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
31285 (legitimate_offset_address_p): New function.
31286 (arc_legitimate_address_p): Use above function.
31288 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31290 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
31292 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31294 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
31295 ACCH registers whenever they are available.
31297 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31299 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
31300 double regs fix when not used.
31302 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31304 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
31306 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
31307 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
31309 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31311 * config/arc/arc.c (arc_output_addsi): Check for h-register class
31312 when emitting short ADD instructions.
31314 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31316 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
31318 (cmpsi_cc_c_insn): Likewise.
31319 (cbranchsi4_scratch): Compute proper instruction length using
31320 compact_hreg_operand.
31321 * config/arc/predicates.md (compact_hreg_operand): New predicate.
31323 2017-04-25 Richard Biener <rguenther@suse.de>
31325 PR middle-end/80509
31326 * passes.c (pass_manager::pass_manager): Initialize
31327 m_name_to_pass_map.
31329 2017-04-25 Richard Biener <rguenther@suse.de>
31331 PR tree-optimization/79201
31332 * tree-ssa-sink.c (statement_sink_location): Handle calls.
31334 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31337 * config/s390/vector.md: Split MEM->GPR vector moves for
31338 non-s_operand addresses.
31340 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31343 * config/s390/predicates.md (reload_const_wide_int_operand): New
31345 * config/s390/s390.md ("movti"): Remove d/P alternative.
31346 ("movti_bigconst"): New pattern definition.
31348 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31351 * s390-protos.h (s390_expand_cs_hqi): Removed.
31352 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
31353 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
31354 modes as well as CCZ1mode and CCZmode.
31355 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
31356 signature of s390_emit_compare_and_swap.
31357 (s390_expand_cs_hqi): Likewise, make static.
31358 (s390_expand_cs_tdsi): Generate an explicit compare before trying
31359 compare-and-swap, in some cases.
31360 (s390_expand_cs): Wrapper function.
31361 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
31363 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
31364 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
31365 patterns for small and large integers. Forbid symref memory operands.
31366 Move expander to s390.c. Require cc register.
31367 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
31368 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
31369 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
31370 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
31371 symref memory operands. Remove CC mode and call s390_match_ccmode
31373 ("atomic_exchange<mode>"): Allow and implement all integer modes.
31375 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31377 * config/s390/s390.md (define_peephole2): New peephole to help
31378 combining the load-and-test pattern with volatile memory.
31380 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31382 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
31383 with CCZmode for TARGET_Z196.
31385 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31387 PR rtl-optimization/80501
31388 * combine.c (make_compound_operation_int): Set subreg_code to SET
31389 even for AND with mask of the sign bit of mode.
31391 PR rtl-optimization/80500
31392 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
31393 sum's initial value.
31395 2017-04-25 Julian Brown <julian@codesourcery.com>
31396 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31398 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
31400 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
31402 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
31404 2017-04-25 Julian Brown <julian@codesourcery.com>
31405 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31407 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
31408 (thunderx2t99_sha): New Reservation.
31410 2017-04-25 Julian Brown <julian@codesourcery.com>
31411 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31413 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
31414 type for 1-element load.
31416 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
31418 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
31420 2017-04-24 Martin Jambor <mjambor@suse.cz>
31422 PR tree-optimization/80293
31423 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
31424 char arrays not totally scalarizable if it is false.
31425 (analyze_all_variable_accesses): Pass correct value in the new
31426 parameter. Add a statistics counter.
31428 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
31430 PR middle-end/79931
31431 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
31433 2017-04-24 Richard Biener <rguenther@suse.de>
31435 PR tree-optimization/80494
31436 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
31437 out for complex types.
31439 2017-04-24 Richard Biener <rguenther@suse.de>
31441 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
31442 * tree-ssa-sccvn.c (print_scc): Print SCC size.
31443 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
31444 (DFS): Adjust and never fail.
31445 (sccvn_dom_walker::fail): Remove.
31446 (sccvn_dom_walker::before_dom_children): Adjust.
31447 (run_scc_vn): Likewise and never fail.
31448 * tree-ssa-pre.c (pass_pre::execute): Adjust.
31449 (pass_fre::execute): Likewise.
31451 2017-04-24 Richard Biener <rguenther@suse.de>
31453 PR tree-optimization/79725
31454 * tree-ssa-sink.c (statement_sink_location): Return whether
31455 failure reason was zero uses. Move that check later.
31456 (sink_code_in_bb): Deal with zero uses by removing the stmt
31459 2017-04-24 Richard Biener <rguenther@suse.de>
31462 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
31463 pointer-based references.
31465 2017-04-24 Richard Biener <rguenther@suse.de>
31468 * pass_manager.h (pass_manager::operator new): Remove.
31469 (pass_manager::operator delete): Likewise.
31470 * passes.c (pass_manager::operator new): Remove.
31471 (pass_manager::operator delete): Likewise.
31472 (pass_manager::pass_manager): Zero individual pass members.
31474 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
31477 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
31478 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
31479 Check "XEXP (src, 1)" operand here.
31480 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
31481 Check "XEXP (src, 1)" operand here.
31482 (dimode_scalar_chain::make_vector_copies): Detect count register
31483 of a shift instruction. Zero extend count register from QImode
31484 to DImode to satisfy vector shift pattern count operand predicate.
31485 Substitute vector shift count operand with a DImode copy.
31486 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
31489 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
31491 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
31492 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31493 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
31494 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31495 (UNSPEC_NOREX_MEM): Remove definition.
31497 2017-04-21 Richard Biener <rguenther@suse.de>
31499 PR tree-optimization/79547
31500 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
31501 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
31502 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
31503 without any constraints.
31505 2017-04-21 Richard Biener <rguenther@suse.de>
31507 PR tree-optimization/78847
31508 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
31510 2017-04-21 Richard Biener <rguenther@suse.de>
31512 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
31513 (build_distinct_type_copy): Likewise.
31514 (build_variant_type_copy): Likewise.
31515 * tree.c (build_qualified_type): Pass down mem-stat info.
31516 (build_distinct_type_copy): Likewise.
31517 (build_variant_type_copy): Likewise.
31519 2017-04-21 Richard Biener <rguenther@suse.de>
31521 PR tree-optimization/80237
31522 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
31524 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
31525 for a simplified result.
31527 2016-04-21 Richard Biener <rguenther@suse.de>
31529 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
31530 sth as strict as a simple_iv but a chrec without symbols and an
31531 operand defined in the loop we are peeling (and not some subloop).
31532 (propagate_constants_for_unrolling): Propagate all constants.
31534 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31537 * config/i386/i386.c (print_reg): Remove assert for disalowed
31538 regno values, call output_operand_lossage instead.
31540 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31543 * config/i386/constraints.md (Yc): New register constraint.
31544 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
31545 Use Yc constraint for alternative 2 of operand 0. Remove
31546 preferred_for_speed attribute.
31548 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31550 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
31551 lastprivate clauses in SIMT case.
31553 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
31555 * doc/invoke.texi (-Wextra-semi): Document new warning option.
31557 2017-04-20 Richard Biener <rguenther@suse.de>
31559 PR tree-optimization/57796
31560 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
31561 as N scalar stores.
31562 (vect_model_load_cost): Cost gathers as N scalar loads.
31564 2017-04-20 Richard Biener <rguenther@suse.de>
31566 * ggc-page.c (ggc_allocated_p): Rename to ...
31567 (safe_lookup_page_table_entry): ... this and return the lookup
31569 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
31571 2017-04-20 Richard Biener <rguenther@suse.de>
31573 PR tree-optimization/80453
31574 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
31575 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
31576 from the conditions.
31577 (vn_phi_eq): Pass them down.
31578 (vn_phi_lookup): Record them.
31579 (vn_phi_insert): Likewise.
31581 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
31583 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
31584 uninitialized variable warning to avoid buffer overrun.
31586 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31589 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
31590 is suppressed for '{ 0 }' in C.
31592 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31594 * BASE-VER: Set to 8.0.0.
31596 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
31598 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
31599 priority .init_array and .fini_array section with SECTION_NOTYPE
31602 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31604 PR middle-end/80423
31605 * tree.h (build_array_type): Add typeless_storage default argument.
31606 * tree.c (type_cache_hasher::equal): Also compare
31607 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
31608 (build_array_type): Add typeless_storage argument, set
31609 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
31611 (build_nonshared_array_type): Adjust build_array_type_1 caller.
31612 (build_array_type): Likewise. Add typeless_storage argument.
31614 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31615 Jakub Jelinek <jakub@redhat.com>
31617 PR tree-optimization/80426
31618 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
31619 operation on symbolic operands, also compute the overflow for the
31620 invariant part when the operation degenerates into a negation.
31622 2017-04-19 Jakub Jelinek <jakub@redhat.com>
31625 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
31626 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
31629 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
31631 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
31634 * config/avr/avr.c (tree.h): Include it.
31635 (cgraph.h): Include it.
31636 (avr_encode_section_info): Don't warn for uninitialized progmem
31637 variable if it's just an alias.
31639 2017-04-19 Richard Biener <rguenther@suse.de>
31642 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
31643 when needed by AutoPGO.
31645 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
31648 * doc/lto.texi: Remove an extra 'that'.
31650 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
31652 PR rtl-optimization/80429
31653 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
31654 are only used in debug insns.
31656 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31657 Vladimir Makarov <vmakarov@redhat.com>
31659 * config/sparc/predicates.md (input_operand): Add comment. Return
31660 true for any memory operand when LRA is in progress.
31661 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
31663 2017-04-18 Jeff Law <law@redhat.com>
31666 * mips.md ({return,simple_return}_internal): Do not overwrite
31669 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31671 PR tree-optimization/80443
31672 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
31673 instead of adding 1, subtract -1 and similarly instead of subtracting
31676 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
31678 PR rtl-optimization/80357
31679 * haifa-sched.c (tmp_bitmap): New variable.
31680 (model_recompute): Handle duplicate use records.
31681 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
31682 (free_global_sched_pressure_data): Free it.
31684 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
31687 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31688 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31689 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31690 instead of SYSTEM_HEADER_DIR.
31692 2017-04-18 Jeff Law <law@redhat.com>
31694 PR middle-end/80422
31695 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
31696 predecessors after walking up the insn chain.
31698 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31701 * dwarf2out.c (modified_type_die): Try harder not to emit internal
31702 sizetype type into debug info.
31704 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31707 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
31708 unneeded test for TARGET_UPPER_REGS_SF.
31709 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
31711 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31714 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
31715 instead of gsi_after_labels.
31717 2017-04-18 Jeff Law <law@redhat.com>
31719 * regcprop.c (maybe_mode_change): Avoid creating copies of the
31723 2017-04-13 Jeff Law <law@redhat.com>
31724 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31725 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31727 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
31730 * config/avr/avr.c (intl.h): Include it.
31731 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
31733 2017-04-18 Martin Liska <mliska@suse.cz>
31735 PR gcov-profile/78783
31736 * gcov-tool.c (gcov_output_files): Validate that destination
31737 file is either removed by the tool or by a user.
31739 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31740 Guy Benyei <guybe@mellanox.com>
31742 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
31743 block, and do not negate it, the stored id is already negative.
31745 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31747 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
31749 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31752 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
31753 masks of options that should be turned off if the VSX vector
31754 options are turned off.
31755 (OTHER_P8_VECTOR_MASKS): Likewise.
31756 (OTHER_VSX_VECTOR_MASKS): Likewise.
31757 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
31758 rs6000_disable_incompatible_switches to validate no type switches
31760 (rs6000_incompatible_switch): New function to disallow turning on
31761 other vector options if -mno-vsx, -mno-power8-vector, or
31762 -mno-power9-vector are specified.
31764 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31766 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
31768 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31770 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
31771 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
31772 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
31773 (ARG_POINTER_CFA_OFFSET): Likewise.
31775 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31777 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
31778 conditions to take advantage of various optimizations.
31780 2017-04-13 Jeff Law <law@redhat.com>
31782 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31783 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31784 (zero_extendsidi2_dext): Likewise.
31786 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31789 * fold-const.c (fold_ternary_loc): Revert
31790 use op0 instead of fold_convert_loc (loc, type, arg0) part of
31793 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
31795 PR rtl-optimization/80343
31796 * lra-remat.c (update_scratch_ops): Assign original hard reg to
31797 new scratch pseudo.
31799 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
31802 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
31803 to ubsan_encode_value.
31805 2017-04-13 Jeff Law <law@redhat.com>
31807 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
31808 appearing in DEBUG_INSNs.
31810 2017-04-13 Martin Liska <mliska@suse.cz>
31812 PR gcov-profile/80413
31813 * gcov-io.c (gcov_write_string): Copy to buffer just when
31814 allocated size is greater than zero.
31816 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31819 * dwarf2out.c (decls_for_scope): Ignore declarations of
31820 current_function_decl in BLOCK_NONLOCALIZED_VARS.
31822 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
31825 * ipa-visibility.c (non_local_p): Fix typos.
31826 (localize_node): When localizing symbol in same comdat group,
31827 dissolve the group only when we know external symbols are going
31829 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
31831 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31833 PR tree-optimization/79390
31834 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
31835 order does not result in usable sequence, retry with reversed operand
31841 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
31842 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
31843 op0 instead of fold_convert_loc (loc, type, arg0).
31845 2017-04-12 Jeff Law <law@redhat.com>
31847 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
31848 has a delay slot in the generated code.
31850 * config/cris/cris.md (cris_preferred_reload_class): Return
31851 GENNONACR_REGS rather than GENERAL_REGS.
31853 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31856 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
31857 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
31858 signedness of the result type.
31860 2017-04-12 Richard Biener <rguenther@suse.de>
31861 Jeff Law <law@redhat.com>
31863 PR tree-optimization/80359
31864 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
31865 trim stores to TARGET_MEM_REFs.
31867 2017-04-12 Richard Biener <rguenther@suse.de>
31869 PR tree-optimization/79390
31870 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
31871 threading case even more.
31873 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
31876 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
31877 for quad_address_p for TImode, instead of just not indexed_address.
31879 2017-04-12 Richard Biener <rguenther@suse.de>
31880 Bernd Edlinger <bernd.edlinger@hotmail.de>
31882 PR middle-end/79671
31883 * alias.c (component_uses_parent_alias_set_from): Handle
31884 TYPE_TYPELESS_STORAGE.
31885 (get_alias_set): Likewise.
31886 * tree-core.h (tree_type_common): Add typeless_storage flag.
31887 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
31888 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
31889 for types containing members with TYPE_TYPELESS_STORAGE.
31890 (place_field): Likewise.
31891 (layout_type): Likewise for ARRAY_TYPE.
31892 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
31893 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
31894 TYPE_TYPELESS_STORAGE.
31895 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
31897 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31900 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31901 first argument to type.
31903 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31907 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
31908 CONST0_RTX (mode) rather than const0_rtx where appropriate.
31909 (rs6000_expand_binop_builtin): Likewise.
31910 (rs6000_expand_ternop_builtin): Likewise; also add missing
31911 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
31912 vshasigma built-ins.
31913 * doc/extend.texi: Document that vec_xxpermdi's third argument
31914 must be a constant.
31916 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
31918 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
31919 Use shift_const cost parameter when calculating gain of STV shifts.
31921 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
31923 PR rtl-optimization/70478
31924 * lra-constraints.c (process_alt_operands): Check memory for
31925 disfavoring memory insn operand.
31927 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31929 PR middle-end/80100
31930 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31931 left shift in unsigned HOST_WIDE_INT type.
31933 PR rtl-optimization/80385
31934 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31935 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31938 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31939 if they have any depend clauses.
31941 2017-04-11 Martin Liska <mliska@suse.cz>
31944 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31945 * ipa-split.c (split_function): Create a local comdat symbol
31946 if caller is in a comdat group.
31948 2017-04-11 Martin Liska <mliska@suse.cz>
31951 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31954 2017-04-11 Martin Sebor <msebor@redhat.com>
31956 PR middle-end/80364
31957 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31958 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31960 (directive::set_width, directive::set_precision, format_character):
31962 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31965 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31968 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31969 conflict, set target->arch_name instead of target->cpu_name.
31971 2017-04-11 Richard Biener <rguenther@suse.de>
31973 PR tree-optimization/80374
31974 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31975 build_zero_cst, remove fold_convertible_p check again.
31977 2017-04-11 Martin Liska <mliska@suse.cz>
31980 * ubsan.c (instrument_object_size): Do not instrument register
31983 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31986 * config/i386/i386-builtin-types.def
31987 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31988 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31989 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31990 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31991 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31992 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31993 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31994 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31995 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31996 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31997 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31998 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31999 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
32000 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
32001 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
32002 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
32003 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
32004 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
32005 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
32006 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
32007 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
32008 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
32009 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
32010 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
32011 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
32012 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
32013 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
32014 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
32015 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
32016 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
32017 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
32018 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
32019 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
32020 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
32021 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
32022 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
32023 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
32024 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
32025 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
32026 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
32027 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
32028 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
32029 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
32030 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
32031 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
32032 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
32034 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
32035 flag to second_arg_count, handle 4 argument function type _COUNT
32036 aliases, handle second_arg_count on second argument rather than last.
32038 2017-04-10 Jeff Law <law@redhat.com>
32040 PR tree-optimization/80374
32041 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
32042 record anything if we can not convert integer_zero_node to the
32045 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
32048 * config/rs6000/rs6000.c (rs6000_option_override_internal):
32049 Enhance special handling given to the TARGET_P9_MINMAX option in
32050 relation to certain other options.
32052 2017-04-10 Bin Cheng <bin.cheng@arm.com>
32054 PR tree-optimization/80153
32055 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
32056 remove POINTER_PLUS_EXPR's base part directly, rather than through
32059 2017-04-10 Richard Biener <rguenther@suse.de>
32060 Bin Cheng <bin.cheng@arm.com>
32062 PR tree-optimization/80153
32063 * tree-affine.c (aff_combination_to_tree): Get base pointer from
32064 the first element of pointer type aff_tree. Build result expr in
32066 (add_elt_to_tree): Convert to type unconditionally. Remove other
32067 fold_convert calls.
32068 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
32069 (rewrite_use_nonlinear_expr): Check invariant using iv information.
32071 2017-04-10 Richard Biener <rguenther@suse.de>
32073 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32076 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
32078 PR rtl-optimization/70478
32079 * lra-constraints.c (curr_small_class_check): New.
32080 (update_and_check_small_class_inputs): New.
32081 (process_alt_operands): Update curr_small_class_check. Disfavor
32082 alternative insn memory operands. Check available regs for small
32085 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32088 * config/mips/mips.opt (-mvirt): Update description.
32089 * doc/invoke.texi (-mvirt): Likewise.
32091 2017-04-10 Richard Biener <rguenther@suse.de>
32093 PR middle-end/80362
32094 * fold-const.c (fold_binary_loc): Look at unstripped ops when
32095 looking for NEGATE_EXPR in -A / -B to A / B folding.
32097 2017-04-10 Martin Liska <mliska@suse.cz>
32099 PR gcov-profile/80224
32100 * gcov.c (print_usage): Fix usage string.
32101 (get_gcov_intermediate_filename): Remove.
32102 (output_gcov_file): Use both for normal and intermediate format.
32103 (generate_results): Do not initialize special file for
32104 intermediate format.
32106 2017-04-10 Richard Biener <rguenther@suse.de>
32108 PR tree-optimization/80304
32109 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
32112 2017-04-10 Nathan Sidwell <nathan@acm.org>
32115 * config/rs6000/rs6000.c (rs6000_vector_type): New.
32116 (rs6000_init_builtins): Use it.
32118 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32120 * config/arm/arm.md (<mrc>): Add mode to SET source.
32121 (<mrrc>): Likewise.
32123 2017-04-10 Richard Biener <rguenther@suse.de>
32125 PR middle-end/80344
32126 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
32128 2017-04-10 Jakub Jelinek <jakub@redhat.com>
32131 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
32132 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
32133 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
32134 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
32135 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
32136 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
32137 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
32138 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
32139 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
32140 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
32141 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
32142 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
32143 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
32144 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
32145 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
32146 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
32147 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
32148 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
32149 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
32150 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
32151 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
32152 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
32153 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
32155 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
32157 PR rtl-optimization/70478
32158 * lra-constraints.c: Reverse the last patch.
32160 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
32162 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
32163 Add comment for WCHAR_T.
32165 2017-04-08 Martin Liska <mliska@suse.cz>
32168 2017-04-07 Martin Liska <mliska@suse.cz>
32171 * ipa-split.c (split_function): Add function part to a same comdat
32174 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32177 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
32179 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
32181 * rs6000/rs6000.c (vec_load_pendulum): Rename...
32182 (vec_pairing): ...to this.
32183 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
32184 (rs6000_sched_init): Adjust for name change.
32185 (struct rs6000_sched_context): Likewise.
32186 (rs6000_init_sched_context): Likewise.
32187 (rs6000_set_sched_context): Likewise.
32189 2017-04-07 Jakub Jelinek <jakub@redhat.com>
32195 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
32197 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
32198 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
32199 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
32201 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
32203 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
32205 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32207 PR rtl-optimization/70703
32208 * ira-color.c (update_conflict_hard_regno_costs): Use
32209 int64_t instead of HOST_WIDE_INT.
32211 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32213 PR rtl-optimization/70478
32214 * lra-constraints.c (process_alt_operands): Disfavor alternative
32215 insn memory operands.
32217 2017-04-07 Jeff Law <law@redhat.com>
32219 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
32220 CALL and NOTE_INSN_CALL_ARG_LOCATION.
32222 2017-04-07 Martin Liska <mliska@suse.cz>
32225 * config/aarch64/aarch64.c (aarch64_process_target_attr):
32226 Show error message instead of an ICE.
32228 2017-04-07 Martin Liska <mliska@suse.cz>
32231 * ipa-split.c (split_function): Add function part to a same comdat
32234 2017-04-07 Richard Biener <rguenther@suse.de>
32236 PR middle-end/80341
32237 * tree.c (get_unwidened): Also handle ! for_type case for
32239 * convert.c (do_narrow): Split out from ...
32240 (convert_to_integer_1): ... here. Do not pass final truncation
32241 type to get_unwidened for TRUNC_DIV_EXPR.
32243 2017-04-07 Richard Biener <rguenther@suse.de>
32245 * tree-affine.c (wide_int_ext_for_comb): Take type rather
32247 (aff_combination_const): Adjust.
32248 (aff_combination_scale): Likewise.
32249 (aff_combination_add_elt): Likewise.
32250 (aff_combination_add_cst): Likewise.
32251 (aff_combination_convert): Likewise.
32252 (add_elt_to_tree): Likewise. Remove unused argument.
32253 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
32255 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
32257 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
32259 * config/arm/arm.c (arm_default_short_enums): Use
32260 ARM_DEFAULT_SHORT_ENUMS.
32261 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
32263 2017-04-06 Jakub Jelinek <jakub@redhat.com>
32266 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
32267 members with redundant out-of-class redeclaration.
32269 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32272 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
32273 * config/i386/i386.md (*zero_extendsidi2):
32274 Add (?*x,*x) and (?*v,*v) alternatives.
32276 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32279 * config/i386/i386.c (ix86_expand_builtin)
32280 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
32281 mode from insn data. Convert operands to insn operand mode.
32282 Copy operands that don't satisfy insn predicate to a register.
32284 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
32286 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
32289 2017-04-06 Richard Biener <rguenther@suse.de>
32291 PR tree-optimization/80334
32292 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
32293 preserve alignment of accesses.
32295 2017-04-06 Richard Biener <rguenther@suse.de>
32297 PR tree-optimization/80262
32298 * tree-sra.c (build_ref_for_offset): Preserve address-space
32300 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
32301 Drop useless address-space information on MEM_REF offsets.
32303 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
32305 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
32307 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32309 PR rtl-optimization/70703
32310 * ira-color.c (update_conflict_hard_regno_costs): Use
32311 HOST_WIDE_INT instead of long.
32313 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
32316 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
32317 not defined for x86_64 target. Add -mmmx target option when __SSE2__
32319 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
32320 for x86_64 target. Handle -m3dnowa option.
32322 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32324 PR rtl-optimization/70703
32325 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
32326 (update_conflict_hard_regno_costs): Use long instead of unsigned
32327 arithmetic for cost calculation.
32329 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32330 Bernd Edlinger <bernd.edlinger@hotmail.de>
32333 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
32336 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
32339 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
32340 ptr_mode with Pmode throughout.
32341 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
32342 into probe_stack_range and use DImode.
32344 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
32347 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
32348 call_eh_return is true.
32350 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32352 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
32353 Initialize last_match_fntype_index.
32355 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32358 * tree-nvr.c: Include internal-fn.h.
32359 (pass_return_slot::execute): Ignore internal calls without
32362 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32363 Richard Biener <rguenther@suse.de>
32366 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
32367 captures used multiple times, except for the last use.
32368 * generic-match-head.c: Include gimplify.h.
32370 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32372 PR tree-optimization/79390
32373 * target.h (struct noce_if_info): Declare.
32374 * targhooks.h (default_noce_conversion_profitable_p): Declare.
32375 * target.def (noce_conversion_profitable_p): New target hook.
32376 * ifcvt.h (struct noce_if_info): New type, moved from ...
32377 * ifcvt.c (struct noce_if_info): ... here.
32378 (noce_conversion_profitable_p): Renamed to ...
32379 (default_noce_conversion_profitable_p): ... this. No longer
32381 (noce_try_store_flag_constants, noce_try_addcc,
32382 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
32383 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
32384 instead of noce_conversion_profitable_p.
32385 * config/i386/i386.c: Include ifcvt.h.
32386 (ix86_option_override_internal): Don't override
32387 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
32388 (ix86_noce_conversion_profitable_p): New function.
32389 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
32390 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
32391 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
32392 * doc/tm.texi: Regenerated.
32394 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32396 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
32399 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32402 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
32403 instructions for small multiply cores.
32405 2017-04-04 Jeff Law <law@redhat.com>
32407 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
32409 (mips_expand_vec_perm_const): Initialize elements in orig_perm
32410 that are not set by the loop over the elements.
32412 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32415 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
32416 int mode, convert_modes it to mode as unsigned, otherwise use
32417 lowpart_subreg to mode rather than SImode.
32418 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
32419 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
32420 Use DImode instead of SImode for the shift count operand.
32421 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
32424 2017-04-04 Richard Biener <rguenther@suse.de>
32426 PR middle-end/80281
32427 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
32428 arithmetic done for the negate or the plus. Simplify.
32429 (A - (-B) -> A + B): Likewise.
32430 * fold-const.c (split_tree): Make sure to not negate pointers.
32432 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
32434 PR rtl-optimization/60818
32435 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
32436 a compare of comparisons with the thing compared if this results
32437 in a different machine mode.
32439 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
32441 * alias.c (base_alias_check): Fix typo in comment.
32442 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
32443 * cgraphunit.c (symbol_table::compile): Likewise.
32444 * collect2.c (maybe_run_lto_and_relink): Likewise.
32445 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
32446 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
32447 * config/avr/avr.c (avr_map_op_t): Likewise.
32448 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
32449 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
32450 * config/epiphany/epiphany.md (movcc): Likewise.
32451 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
32452 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
32454 * config/mips/mips.c (mips_save_restore_reg): Likewise.
32455 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
32456 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
32457 * config/sh/sh.c (sh_rtx_costs): Likewise.
32458 * fold-const.c (fold_truth_andor): Likewise.
32459 * genautomata.c (collapse_flag): Likewise.
32460 * gengtype.h (struct type::u::s): Likewise.
32461 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
32462 * input.c (FORMAT_AMOUNT): Likewise.
32463 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
32464 (known_aggs_to_agg_replacement_list): Likewise.
32465 * ipa-inline-analysis.c: Likewise.
32466 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
32467 * ipa-polymorphic-call.c
32468 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
32469 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
32470 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
32472 * modulo-sched.c (apply_reg_moves): Likewise.
32473 * omp-expand.c (build_omp_regions_1): Likewise.
32474 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
32475 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
32476 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
32477 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32478 * value-prof.c: Likewise.
32479 * var-tracking.c (val_reset): Likewise.
32481 2017-04-03 Richard Biener <rguenther@suse.de>
32483 PR tree-optimization/80275
32484 * fold-const.c (split_address_to_core_and_offset): Handle
32487 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
32489 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
32490 descriptors is at least equal to that of functions.
32492 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32494 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
32496 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32499 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
32500 (mov<IMOD4:mode>): New expander.
32501 (*mov<IMOD4:mode>_internal): New insn and split pattern.
32503 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
32505 PR rtl-optimization/79405
32506 * fwprop.c (propagations_left): New variable.
32507 (forward_propagate_into): Decrement it.
32508 (fwprop_init): Initialize it.
32509 (fw_prop): If the variable has reached zero, stop propagating.
32510 (fwprop_addr): Ditto.
32512 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32515 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
32516 a FUNCTION_DECL, pass it as decl instead of origin to
32519 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
32521 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
32524 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
32527 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
32528 TARGET_VSX_SMALL_INTEGER.
32530 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32532 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32533 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
32535 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32537 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
32538 extraction from odd-numbered MSA register.
32540 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32542 PR middle-end/80173
32543 * expmed.c (store_bit_field_1): Don't attempt to create
32544 a word subreg out of hard registers wider than word if they
32545 have HARD_REGNO_NREGS of 1 for their mode.
32547 PR middle-end/80163
32548 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
32549 conversions to integer types wider than word and pointer.
32552 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
32553 (rtx_equal_for_cselib_p): Pass 0 to it.
32554 * cselib.c (cselib_hasher::equal): Likewise.
32555 (rtx_equal_for_cselib_1): Add depth argument. If depth
32556 is 128, don't look up VALUE locs and punt. Increment
32557 depth in recursive calls when walking VALUE locs.
32559 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
32561 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
32562 (make_gcov_file_name): Use the canonical path name for generating
32564 (read_line): Fix handling of files with ascii null bytes.
32566 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
32568 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
32569 to initialise a vector register instead
32570 of using a const_int.
32572 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32574 PR translation/80189
32575 * gimplify.c (omp_default_clause): Use %qs instead of %s in
32576 diagnostic messages.
32578 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
32581 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
32582 (dfp_diex_<mode>): Update mode of operand 1.
32583 * doc/extend.texi (dxex, dxexq): Document change to return type.
32584 (diex, diexq): Document change to argument type.
32586 2017-03-30 Martin Jambor <mjambor@suse.cz>
32589 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
32590 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
32591 it reflects the signature changes performed at the callee side.
32592 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
32593 to cgraph_build_function_type_skip_args.
32594 (build_function_decl_skip_args): Adjust call to the above function.
32596 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32599 * config/i386/sse.md
32600 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
32601 register as dest whenever it is a MEM not rtx_equal_p to the
32602 corresponding dup operand, and when forcing into reg move the
32603 reg into the memory afterwards.
32604 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
32605 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
32606 for the force_reg mode.
32607 (avx512vl_vextractf128<mode>): Use register as dest either
32608 always when a MEM, or when it is a MEM not rtx_equal_p to the
32609 corresponding dup operand, or even not when it is a CONST_VECTOR
32610 depending on the mode and lo vs. hi.
32611 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
32613 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
32614 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
32615 Likewise. Require that operands[2] is even.
32616 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
32617 Remove extraneous parens. Require that operands[2] is a multiple
32619 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
32620 operands[0] is a MEM if <mask_applied>, the predicates/constraints
32621 disallow memory then.
32623 2017-03-30 Richard Biener <rguenther@suse.de>
32625 PR tree-optimization/77498
32626 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
32627 to non-constants over backedges.
32629 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
32631 PR rtl-optimization/80233
32632 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
32633 as last_combined_insn. Do not test for BARRIER_P separately.
32635 2017-03-29 Andreas Schwab <schwab@suse.de>
32638 * calls.c (prepare_call_address): Convert funexp to Pmode before
32639 copying to temp reg.
32641 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32643 PR tree-optimization/80158
32644 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
32645 Handle possible future case of more than one alternate
32647 (replace_rhs_if_not_dup): Likewise.
32648 (replace_one_candidate): Likewise.
32650 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
32652 PR rtl-optimization/80193
32653 * ira.c (ira): Do not check allocation for LRA.
32655 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
32657 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
32658 (nvptx_output_simt_exit): Declare.
32659 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
32660 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
32661 (init_softstack_frame): Move initialization of crtl->is_leaf to...
32662 (nvptx_declare_function_name): ...here. Emit declaration of local
32663 memory space buffer for omp_simt_enter insn.
32664 (nvptx_output_unisimt_switch): New.
32665 (nvptx_output_softstack_switch): New.
32666 (nvptx_output_simt_enter): New.
32667 (nvptx_output_simt_exit): New.
32668 * config/nvptx/nvptx.h (struct machine_function): New fields
32669 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
32670 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
32671 (UNSPECV_SIMT_EXIT): Ditto.
32672 (omp_simt_enter_insn): New insn.
32673 (omp_simt_enter): New expansion.
32674 (omp_simt_exit): New insn.
32675 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
32677 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
32678 (expand_GOMP_SIMT_ENTER_ALLOC): New.
32679 (expand_GOMP_SIMT_EXIT): New.
32680 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
32681 (GOMP_SIMT_ENTER_ALLOC): Ditto.
32682 (GOMP_SIMT_EXIT): Ditto.
32683 * target-insns.def (omp_simt_enter): New insn.
32684 (omp_simt_exit): Ditto.
32685 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
32687 (lower_rec_simd_input_clauses): Implement SIMT privatization.
32688 (lower_rec_input_clauses): Likewise.
32689 (lower_lastprivate_clauses): Handle SIMT privatization.
32691 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
32692 (ompdevlow_adjust_simt_enter): New.
32693 (find_simtpriv_var_op): New.
32694 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
32695 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
32697 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
32698 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
32699 (copy_decl_for_dup_finish): Ditto.
32701 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
32703 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
32706 * config/i386/i386.c (ix86_option_override_internal): Always
32707 allow -mpreferred-stack-boundary=3 for 64-bit targets.
32709 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32711 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
32713 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32715 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
32716 mark new edge's irreducible flag accordign to it.
32717 (vect_do_peeling): Check loop preheader edge's irreducible flag
32718 and pass it to function slpeel_add_loop_guard.
32720 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
32722 PR tree-optimization/80218
32723 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32724 Update block frequencies and counts.
32726 2017-03-28 Richard Biener <rguenther@suse.de>
32728 PR tree-optimization/78644
32729 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
32730 of a simplification result we may not use it at all.
32732 2017-03-28 Richard Biener <rguenther@suse.de>
32735 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
32736 without arguments, generate default definition of a SSA name.
32738 2017-03-28 Richard Biener <rguenther@suse.de>
32740 PR middle-end/80222
32741 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
32742 TYPE_REF_CAN_ALIAS_ALL references.
32743 * fold-const.c (fold_indirect_ref_1): Likewise.
32745 2017-03-28 Martin Liska <mliska@suse.cz>
32748 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
32749 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
32751 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32752 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
32754 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
32755 (EXTRA_SPECS): Define.
32756 (SUBTARGET_EXTRA_SPECS): Likewise.
32757 (SUBTARGET_CPP_SPEC): Likewise.
32758 * config/arc/elf.h (EXTRA_SPECS): Renamed to
32759 SUBTARGET_EXTRA_SPECS.
32760 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
32762 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32764 * config/arc/simdext.md (vst64_insn): Update pattern.
32765 (vld32wh_insn): Likewise.
32766 (vld32wl_insn): Likewise.
32767 (vld64_insn): Likewise.
32768 (vld32_insn): Likewise.
32770 2017-03-28 Marek Polacek <polacek@redhat.com>
32773 * fold-const.c (fold_comparison): Use protected_set_expr_location
32774 instead of SET_EXPR_LOCATION.
32776 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
32778 * tree.c (add_expr): Avoid name lookup warning.
32780 2017-03-27 Jeff Law <law@redhat.com>
32782 PR tree-optimization/80216
32783 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
32784 function name. Limit recursion depth.
32785 (record_temporary_equivalences): Corresponding changes.
32787 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
32789 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
32792 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32795 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
32797 * cfgcleanup.c (reg_note_cfa_p): New array.
32798 (insns_have_identical_cfa_notes): New function.
32799 (old_insns_match_p): Don't cross-jump in between /f
32800 and non-/f instructions. If both i1 and i2 are frame related,
32801 verify all CFA notes, their order and content.
32803 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
32806 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
32807 HImode and SImode with zero extend to DImode to one insn.
32808 (bswap<mode>2_extenddi): Likewise.
32809 (bswapsi2_extenddi): Likewise.
32810 (bswaphi2_extendsi): Likewise.
32811 (bswaphi2): Combine bswap HImode and SImode into one insn.
32812 Separate memory insns from swapping register.
32813 (bswapsi2): Likewise.
32814 (bswap<mode>2): Likewise.
32815 (bswaphi2_internal): Delete, no longer used.
32816 (bswapsi2_internal): Likewise.
32817 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
32818 store, and gpr<-gpr swap insns.
32819 (bswap<mode>2_store): Likewise.
32820 (bswaphi2_reg): Register only splitter, combine with the splitter.
32821 (bswaphi2 splitter): Likewise.
32822 (bswapsi2_reg): Likewise.
32823 (bswapsi2 splitter): Likewise.
32824 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
32825 the insns into load, store, and register/register insns.
32826 (bswapdi2_ldbrx): Likewise.
32827 (bswapdi2_load): Likewise.
32828 (bswapdi2_store): Likewise.
32829 (bswapdi2_reg): Likewise.
32831 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
32833 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
32834 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
32836 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
32839 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
32841 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32842 special handling for target option conflicts between dform
32843 options (-mpower9-dform, -mpower9-dform-vector,
32844 -mpower9-dform-scalar) and -mno-direct-move.
32846 2017-03-27 Richard Biener <rguenther@suse.de>
32848 PR tree-optimization/80181
32849 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
32851 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32853 * config/arc/predicates.md (move_double_src_operand): Replace the
32854 call to move_double_src_operand with a call to address_operand.
32856 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32858 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
32859 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
32860 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
32862 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32864 * config/arc/predicates.md (long_immediate_loadstore_operand):
32865 Consider scaled addresses cases.
32867 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32869 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
32870 restored when in interrupt.
32871 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
32872 doesn't have delay slot.
32874 2017-03-27 Richard Biener <rguenther@suse.de>
32877 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
32878 inlined thunk clones.
32880 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32883 * asan.c (instrument_derefs): Copy over last operand from
32884 original COMPONENT_REF to the new COMPONENT_REF with
32885 DECL_BIT_FIELD_REPRESENTATIVE.
32886 * ubsan.c (instrument_object_size): Likewise.
32888 2017-03-27 Richard Biener <rguenther@suse.de>
32890 PR tree-optimization/80170
32891 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
32892 sure DR/SCEV didnt fold in constants we do not see when looking
32893 at the reference base alignment.
32895 2017-03-27 Richard Biener <rguenther@suse.de>
32897 PR middle-end/80171
32898 * gimple-fold.c (fold_ctor_reference): Properly guard against
32899 NULL return value from canonicalize_constructor_val.
32901 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
32904 * config/i386/i386.c (ix86_expand_builtin)
32905 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
32906 flags reg setting and flags reg using instructions.
32907 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
32908 clobbering instructions to zero extend op2.
32910 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
32912 * doc/install.texi (Configuration) <--with-aix-soname>:
32913 Update link to AIX ld.
32915 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
32917 PR rtl-optimization/80160
32918 PR rtl-optimization/80159
32919 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
32920 reg_alternate_class into account.
32922 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
32925 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
32926 to consider in curr_insn_transform.
32928 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32930 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32931 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32932 and emit_mode_inner.
32934 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32936 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32937 argument to the overloaded builtin variants. Use the new flag to
32938 deprecate certain builtin variants.
32939 * config/s390/s390-builtin-types.def: Add new builtin types.
32940 * config/s390/s390-builtins.h: Support new flags field for
32941 overloaded builtins.
32942 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32943 (s390_macro_to_expand): Enable vector float data type.
32944 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32945 builtins by incrementing the __VEC__ version number.
32946 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32948 (s390_resolve_overloaded_builtin): Emit error messages depending
32949 on the builtin flags.
32950 * config/s390/s390.c (s390_expand_builtin): Support additional
32951 flags argument. Change error message to match the messages
32952 emitted in s390-c.c.
32953 * config/s390/s390.md: New UNSPEC_* constants.
32954 (op_type): Add new instruction types.
32955 * config/s390/vecintrin.h: Add new builtins and test data class
32957 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32958 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32959 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32960 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32961 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32962 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32964 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32965 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32966 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32967 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32969 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32970 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32971 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32972 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32974 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32975 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32976 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32977 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32978 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32979 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32980 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32982 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32983 ("vec_scatter_element<V_HW_4:mode>_DI")
32984 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32985 ("vec_fpint<mode>", "vflls")
32986 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32987 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32988 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32989 ("*vec_cmphe<mode>_cc"): ... these.
32991 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32992 mode constant instead of magic value.
32994 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32996 * config/s390/s390.c (s390_expand_vec_compare): Support other
32997 vector floating point modes than just V2DF.
32998 (s390_expand_vcond): Likewise.
32999 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
33000 (s390_cannot_change_mode_class): Prevent mode changes between TF
33001 and V1TF in vector registers.
33002 * config/s390/s390.md (DF, SF): New mode attributes.
33003 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
33004 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
33005 SFmode support for VRs.
33006 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
33008 (VFT, VF_HW): New mode iterators.
33009 (vw, sdx): New mode attributes.
33010 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
33011 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
33012 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
33013 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
33014 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
33015 also the new vector floating point modes. Renaming to ...
33017 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
33018 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
33019 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
33020 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
33021 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
33022 ("vec_unordered<mode>"): ... these.
33024 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
33025 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
33026 ("*vec_extendv2df"): New insn definitions.
33028 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33030 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
33031 ("mulditi3_2", "*muldi3_sign"): New patterns.
33032 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
33033 rename the pattern definition.
33035 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33037 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
33039 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
33041 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33043 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
33044 instruction if possible.
33045 * config/s390/vector.md (vec_halfnumelts): New mode
33047 ("*vec_vllezlf<mode>"): New pattern.
33049 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33051 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
33052 ("popcountv4si2", "popcountv2di2"): Rename to ...
33053 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
33054 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
33056 ("popcount<mode>2_vxe"): New pattern.
33058 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33060 * common/config/s390/s390-common.c (processor_flags_table): Add
33062 * config.gcc: Add arch12.
33063 * config/s390/driver-native.c (s390_host_detect_local_cpu):
33064 Default to arch12 for unknown CPU model numbers.
33065 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
33066 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
33067 PROCESSOR_max sanity check.
33068 * config/s390/s390-opts.h (enum processor_type): Add
33070 * config/s390/s390.c (processor_table): Add arch12.
33071 (s390_expand_builtin): Add check for B_VXE flag.
33072 (s390_issue_rate): Add PROCESSOR_ARCH12.
33073 (s390_get_sched_attrmask): Likewise.
33074 (s390_get_unit_mask): Likewise.
33075 (s390_sched_score): Enable z13 scheduling for arch12.
33076 (s390_sched_reorder): Likewise.
33077 (s390_sched_variable_issue): Likewise.
33078 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
33080 (s390_tune_attr): Use z13 scheduling also for arch12.
33081 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
33082 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
33083 (TARGET_VXE_P): New macros.
33084 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
33085 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
33086 * config/s390/s390.opt: Add arch12 as processor_type.
33088 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33090 * config/s390/s390.md
33091 ("fixuns_truncdddi2", "fixuns_trunctddi2")
33092 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
33093 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
33095 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
33096 Rename expanders to ...
33098 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
33099 ("fixuns_truncdddi2_emu"): ... these.
33101 ("fixuns_trunc<mode>si2_emu"): New expander.
33103 ("*fixuns_truncdfdi2_z13"): Rename to ...
33104 ("*fixuns_truncdfdi2_vx"): ... this.
33106 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33108 * config/s390/2964.md: Remove the single element vector compare
33109 instructions which are no longer used.
33110 * config/s390/s390.c (s390_select_ccmode): Remove handling of
33112 (s390_canonicalize_comparison): Remove handling of DFmode
33114 (s390_expand_vec_compare_scalar): Remove function.
33115 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
33116 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
33118 ("*cmp<mode>_ccs"): Add wfcdb instruction.
33120 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33122 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
33124 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
33125 will anyway by matched by mov<mode>_64dfp.
33127 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33129 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
33130 vlef/vstef. Add missing operand to vleif.
33132 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33134 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
33135 pair for all vector types with 64 bit elements.
33136 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
33137 * config/s390/vector.md (V_HW_64): ... here.
33138 (V_128_NOSINGLE): New mode iterator.
33139 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
33140 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
33141 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
33142 ("*vec_load_pairv2di"): Change to ...
33143 ("*vec_load_pair<mode>"): ... this one.
33145 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33147 * config/s390/constraints.md: Add comments.
33148 (jKK): Reject element sizes > 8 bytes.
33149 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
33151 * config/s390/s390.md: Add the s_operand checks formerly in
33152 s390_split_ok_p to various splitters where they are still
33154 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
33155 for 128 bit vectors. Plus two splitters.
33157 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33159 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
33162 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33165 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
33166 error if the boundary argument is not constant.
33168 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33170 PR rtl-optimization/80112
33171 * loop-doloop.c (doloop_condition_get): Don't check condition
33172 if cmp isn't SET with IF_THEN_ELSE src.
33174 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33176 PR tree-optimization/80158
33177 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
33178 replacing a candidate statement, also replace it for the
33179 candidate's alternate interpretation.
33180 (replace_rhs_if_not_dup): Likewise.
33181 (replace_one_candidate): Likewise.
33183 2017-03-24 Richard Biener <rguenther@suse.de>
33185 PR tree-optimization/80167
33186 * graphite-isl-ast-to-gimple.c
33187 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
33189 (translate_isl_ast_to_gimple::get_rename): Likewise.
33191 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33193 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
33194 handling of certain combinations of target options, including the
33195 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
33196 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
33198 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33201 * config/arm/arm.md (*load_multiple): Add reload_completed to
33202 matching condition.
33204 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33205 Richard Biener <rguenth@suse.de>
33207 PR tree-optimization/79908
33208 PR tree-optimization/80136
33209 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33210 been cast away, gimplify_and_add suffices.
33212 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
33214 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
33216 2017-03-23 Richard Biener <rguenther@suse.de>
33218 PR tree-optimization/80032
33219 * gimplify.c (gimple_push_cleanup): Forced unconditional
33220 cleanups still have to go to the conditional_cleanups
33223 2017-03-22 Jakub Jelinek <jakub@redhat.com>
33225 PR tree-optimization/80072
33226 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
33228 (next_operand_entry_id): Change type to unsigned int.
33229 (sort_by_operand_rank): Make sure to return the right return value
33230 even if unsigned fields are bigger than INT_MAX.
33231 (struct oecount): Change cnt and id type to unsigned int.
33232 (oecount_hasher::equal): Formatting fix.
33233 (oecount_cmp): Make sure to return the right return value
33234 even if unsigned fields are bigger than INT_MAX.
33235 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
33238 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
33239 TREE_READONLY on result if writing it more than once.
33242 * doc/invoke.texi (-fsanitize=thread): Document that with
33243 -fnon-call-exceptions atomics are not able to throw
33247 * tsan.c: Include tree-eh.h.
33248 (instrument_builtin_call): Call maybe_clean_eh_stmt or
33249 maybe_clean_or_replace_eh_stmt where needed.
33250 (instrument_memory_accesses): Add cfg_changed argument.
33251 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
33252 if it returned true.
33253 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
33255 PR rtl-optimization/63191
33256 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
33257 wrapper function, moved the whole old content into ...
33258 (ix86_delegitimize_address_1): ... this. New inline function.
33259 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
33260 true as last argument instead of ix86_delegitimize_address.
33262 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33264 * config/aarch64/aarch64.c (generic_branch_cost): Copy
33265 cortexa57_branch_cost.
33267 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33269 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
33271 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33274 * doc/md.texi (Constraints): Document wA constraint.
33275 * config/rs6000/constraints.md (wA): New.
33276 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
33277 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
33278 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
33279 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
33281 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
33284 * gimplify.c (is_oacc_declared): New function.
33285 (oacc_default_clause): Use it to set default flags for acc declared
33286 variables inside parallel regions.
33287 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
33288 declared variables.
33289 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
33290 declare attribute to any decl as necessary.
33292 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
33295 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
33296 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
33297 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
33298 (arm_arch_lpae): This.
33299 * config/arm/arm.c (arm_arch7ve): Rename into ...
33300 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
33301 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
33304 2017-03-22 Martin Liska <mliska@suse.cz>
33307 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
33308 error message instead of an ICE.
33310 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33312 * doc/extend.texi (6.11 Additional Floating Types): Revise.
33314 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
33316 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33318 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33321 2017-03-21 Martin Sebor <msebor@redhat.com>
33323 * doc/extend.texi: Use "cannot" instead of "can't."
33324 * doc/hostconfig.texi: Same.
33325 * doc/install.texi: Same.
33326 * doc/invoke.texi: Same.
33327 * doc/loop.texi: Same.
33328 * doc/md.texi: Same.
33329 * doc/objc.texi: Same.
33330 * doc/rtl.texi: Same.
33331 * doc/tm.texi: Same.
33332 * doc/tm.texi.in: Same.
33333 * doc/trouble.texi: Same.
33335 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
33338 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
33339 (collect_checksum_attributes): Set it.
33340 (die_checksum_ordered): Use it.
33342 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33344 PR tree-optimization/79908
33345 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
33346 change: For a VA_ARG whose LHS has been cast away, use
33347 force_gimple_operand to construct the side effects.
33349 2017-03-21 David Malcolm <dmalcolm@redhat.com>
33351 PR translation/80001
33352 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
33353 more amenable to translation.
33354 (oacc_loop_auto_partitions): Likewise.
33356 2017-03-21 Marek Polacek <polacek@redhat.com>
33357 Martin Sebor <msebor@redhat.com>
33359 PR tree-optimization/80109
33360 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
33361 on INTEGRAL_TYPE_P.
33363 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33364 Segher Boessenkool <segher@kernel.crashing.org>
33367 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
33368 check reg_used_between_p between insn and one of succ or succ2
33369 depending on if succ is artificial insn not inserted into insn
33372 2017-03-21 Martin Liska <mliska@suse.cz>
33374 PR gcov-profile/80081
33375 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
33376 * doc/gcc.texi: Include gcov-dump stuff.
33377 * doc/gcov-dump.texi: New file.
33379 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
33381 PR rtl-optimization/79150
33382 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
33383 conditional jump, if the jump is the last insn of the loop.
33385 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33386 Richard Biener <rguenth@suse.de>
33388 PR tree-optimization/79908
33389 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33390 been cast away, use force_gimple_operand to construct the side
33393 2017-03-21 Martin Liska <mliska@suse.cz>
33395 PR libfortran/79956
33396 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
33399 2017-03-21 Brad Spengler <spender@grsecurity.net>
33402 * plugin.c (htab_hash_plugin): New function.
33403 (add_new_plugin): Use it and adjust.
33404 (parse_plugin_arg_opt): Adjust.
33405 (init_one_plugin): Likewise.
33407 2017-03-21 Richard Biener <rguenther@suse.de>
33409 PR tree-optimization/80032
33410 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
33411 if set force the cleanup to happen unconditionally.
33412 (gimplify_target_expr): Push inserted clobbers with force_uncond
33413 to avoid them being removed by control-dependent DCE.
33415 2017-03-21 Richard Biener <rguenther@suse.de>
33417 PR tree-optimization/80122
33418 * tree-inline.c (copy_bb): Do not expans va-arg packs or
33419 va_arg_pack_len when the inlined call stmt requires pack
33421 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
33423 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33426 * tsan.c (instrument_builtin_call): If the memory model argument
33427 is not a constant, assume it is valid.
33430 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
33433 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33435 PR rtl-optimization/79910
33436 * combine.c (can_combine_p): Do not allow combining an I0 or I1
33437 if its dest is used by an insn before I2 (other than the combined
33438 insns themselves, which are properly handled already).
33440 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33443 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33445 * combine.c (record_used_regs): New static function.
33446 (try_combine): Handle situations where there is an additional
33447 instruction between I2 and I3 which needs to have a LOG_LINK
33451 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33453 * combine.c (try_combine): Delete redundant i1 test. Call
33454 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33456 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33459 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
33460 alternatives 13/14.
33462 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33464 PR tree-optimization/80054
33465 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
33466 the optimization if a PHI or any of its arguments is not dominated
33467 by the candidate's basis. Use gphi* rather than gimple* as
33469 (replace_profitable_candidates): Clean up a gimple* variable that
33470 should be a gphi* variable.
33472 2017-03-20 Martin Sebor <msebor@redhat.com>
33475 * doc/extend.texi (attribute constructor): Document present limitation.
33477 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
33480 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
33481 __POWER9_VECTOR__ #ifdef control, change template definition to
33482 use Power9-specific built-in function.
33483 (vec_any_eq): Likewise.
33484 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
33485 to control outcomes from this test.
33486 (vector_ae_<mode>p): For VEC_F modes, likewise.
33488 2017-03-20 Ian Lance Taylor <iant@google.com>
33490 * config/i386/i386.c (ix86_function_regparm): Save an extra
33491 register for -fsplit-stack with DECL_STATIC_CHAIN.
33493 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33496 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
33497 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
33499 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33501 * config/riscv/riscv.c (riscv_print_operand): Use "fence
33503 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
33506 2017-03-20 Marek Polacek <polacek@redhat.com>
33509 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
33511 2017-03-20 Richard Biener <rguenther@suse.de>
33513 PR tree-optimization/80113
33514 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
33515 allocate extra SSA name for PHI def.
33516 (add_close_phis_to_outer_loops): Likewise.
33517 (add_close_phis_to_merge_points): Likewise.
33518 (copy_loop_close_phi_args): Likewise.
33519 (copy_cond_phi_nodes): Likewise.
33521 2017-03-20 Martin Liska <mliska@suse.cz>
33523 PR middle-end/79753
33524 * tree-chkp.c (chkp_build_returned_bound): Do not build
33525 returned bounds for a LHS that's not a BOUNDED_P type.
33527 2017-03-20 Martin Liska <mliska@suse.cz>
33531 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
33532 COMPLEX_CST and VECTOR_CST.
33534 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33537 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
33538 target operand. A new splitter adds the clobber statement in case
33539 the target operand is dead anyway.
33541 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
33543 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
33544 to age-old versions of binutils and glibc.
33546 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
33548 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
33550 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33552 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
33554 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33556 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
33557 requirement for binutils 2.13.
33559 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33561 * combine.c (try_combine): Delete redundant i1 test. Call
33562 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33564 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
33566 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
33567 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
33569 <riscv64-*-elf>: Re-arrange section
33570 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
33571 <riscv32-*-linux>: Likewise.
33572 <riscv64-*-elf>: Likewise
33573 <riscv64-*-linux>: Likewise.
33575 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
33578 * aarch64.opt(verbose-cost-dump): Fix typo.
33580 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
33583 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
33584 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
33586 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33588 * reload.c (find_reloads): When reloading a nonoffsettable address,
33589 use RELOAD_OTHER for it and its address reloads.
33591 PR rtl-optimization/79910
33592 * combine.c (record_used_regs): New static function.
33593 (try_combine): Handle situations where there is an additional
33594 instruction between I2 and I3 which needs to have a LOG_LINK
33597 2017-03-17 Jeff Law <law@redhat.com>
33599 PR tree-optimization/71437
33600 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
33601 conditional in the hash table first.
33602 (vrp_dom_walker::before_dom_children): Extract condition from
33603 ASSERT_EXPR. Record condition, its inverion and any implied
33604 conditions as well.
33606 2017-03-17 Marek Polacek <polacek@redhat.com>
33607 Markus Trippelsdorf <markus@trippelsdorf.de>
33609 PR tree-optimization/80079
33610 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
33613 2017-03-17 Richard Biener <rguenther@suse.de>
33615 PR middle-end/80075
33616 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
33617 Properly verify the LHS before the RHS possibly claims to be
33619 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
33622 2017-03-17 Martin Jambor <mjambor@suse.cz>
33624 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
33625 (List of -O2 options): Likewise.
33626 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
33629 2017-03-17 Tom de Vries <tom@codesourcery.com>
33631 * gcov-dump.c (print_usage): Print bug_report_url.
33633 2017-03-17 Richard Biener <rguenther@suse.de>
33635 PR middle-end/80050
33636 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
33637 (parser::peek): Likewise.
33639 2017-03-17 Richard Biener <rguenther@suse.de>
33641 PR tree-optimization/80048
33642 * sese.c (free_sese_info): Properly release rename_map and
33643 copied_bb_map elements.
33645 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
33647 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
33648 Add linked-list forward and backlinks. Insert on
33649 construction, remove on destruction.
33650 (class pass_store_merging): Add m_stores_head field.
33651 (pass_store_merging::terminate_and_process_all_chains):
33652 Iterate over m_stores_head list.
33653 (pass_store_merging::terminate_all_aliasing_chains):
33655 (pass_store_merging::execute): Check for debug stmts first.
33656 Push new chains onto the m_stores_head stack.
33658 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
33661 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
33662 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
33663 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
33665 2017-03-16 Jeff Law <law@redhat.com>
33667 PR tree-optimization/71437
33668 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
33669 member function. Implementation moved into after_dom_children
33670 member function and into the threader's thread_outgoing_edges
33672 (dom_opt_dom_walker::after_dom_children): Simplify by moving
33673 some code into new thread_outgoing_edges.
33674 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
33675 definition. Simplify marker handling (do it here). Assume we always
33676 have the available expression and the const/copies tables.
33677 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
33679 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
33680 * tree-vrp.c (equiv_stack): No longer file scoped.
33681 (vrp_dom_walker): New class.
33682 (vrp_dom_walker::before_dom_children): New member function.
33683 (vrp_dom_walker::after_dom_children): Likewise.
33684 (identify_jump_threads): Setup domwalker. Use it rather than
33685 walking edges in a random order by hand. Simplify setup/finalization.
33686 (finalize_jump_threads): Remove.
33687 (vrp_finalize): Do not call identify_jump_threads here.
33688 (execute_vrp): Do it here instead and call thread_through_all_blocks
33691 PR tree-optimization/71437
33692 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
33694 (simplify_stmt_for_jump_threading): Add basic_block argument. All
33696 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
33697 (dom_opt_dom_walker::thread_across_edge): Remove
33698 handle_dominating_asserts argument. All callers changed.
33699 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
33700 changes. Remove calls to lhs_of_dominating_assert. Other
33701 uses of handle_dominating_asserts turn into unconditional code
33702 (simplify_control_stmt_condition_1): Likewise.
33703 (simplify_control_stmt_condition): Likewise.
33704 (thread_through_normal_block, thread_across_edge): Likewise.
33705 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
33706 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
33707 object if it is not an SSA_NAME.
33708 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
33709 before calling into the VRP specific simplifiers.
33710 (identify_jump_threads): Remove handle_dominating_asserts
33713 2017-03-16 Jakub Jelinek <jakub@redhat.com>
33716 * tree-diagnostic.c (default_tree_printer): No longer static.
33717 * tree-diagnostic.h (default_tree_printer): New prototype.
33719 2017-03-16 Tamar Christina <tamar.christina@arm.com>
33721 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
33722 Change ins into fmov.
33724 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33726 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
33727 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
33728 Use h_con constraint for operand 1.
33729 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
33730 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
33732 2017-03-15 Jeff Law <law@redhat.com>
33734 PR tree-optimization/71437
33735 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
33736 (record_temporary_equivalences): Use it.
33738 PR tree-optimization/71437
33739 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
33740 tree-ssa-scopedtables.
33741 (lookup_avail_expr, build_and_record_new_cond): Likewise.
33742 (record_conditions, record_cond, vuse_eq): Likewise.
33743 (record_edge_info): Adjust to API tweak of record_conditions.
33744 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
33745 (record_temporary_equivalences, optimize_stmt): Likewise.
33746 (eliminate_redundant_computations): Likewise.
33747 (record_equivalences_from_stmt): Likewise.
33748 * tree-ssa-scopedtables.c: Include options.h and params.h.
33749 (vuse_eq): New function, moved from tree-ssa-dom.c
33750 (build_and_record_new_cond): Likewise.
33751 (record_conditions): Likewise. Accept vector of conditions rather
33752 than edge_equivalence structure for first argument.
33753 for the first argument.
33754 (avail_exprs_stack::lookup_avail_expr): New member function, moved
33755 from tree-ssa-dom.c.
33756 (avail_exprs_stack::record_cond): Likewise.
33757 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
33758 from tree-ssa-dom.c.
33759 (avail_exprs_stack): Add new member functions lookup_avail_expr
33761 (record_conditions): Declare.
33763 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
33766 * lra-constraints.c (process_alt_operands): Increase reject for
33767 reloading an input/output operand.
33769 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
33772 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
33773 insns to convert from signed/unsigned char/short to IEEE 128-bit
33775 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
33777 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
33780 * config/i386/i386.c (ix86_vector_duplicate_value): Create
33781 subreg of inner mode for values already in registers.
33783 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
33785 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
33786 iteration reg is used after the loop.
33788 2017-03-14 Martin Sebor <msebor@redhat.com>
33790 PR tree-optimization/79800
33791 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
33792 precision in negative-positive range.
33793 (format_floating): Call non-const overload with adjusted precision.
33795 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33798 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
33801 2017-03-14 Martin Sebor <msebor@redhat.com>
33803 PR middle-end/80020
33804 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
33805 * builtins.def (aligned_alloc): Use it.
33808 * Makefile.in (GTFILES): Add calls.c.
33809 * calls.c: Include "gt-calls.h".
33811 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
33813 PR rtl-optimization/79728
33814 * regs.h (struct target_regs): New field
33815 x_contains_allocatable_regs_of_mode.
33816 (contains_allocatable_regs_of_mode): New macro.
33817 * reginfo.c (init_reg_sets_1): Initialize it, and change
33818 contains_reg_of_mode so it includes global regs as well.
33819 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
33820 rather than contains_regs_of_mode.
33822 2017-03-14 Martin Liska <mliska@suse.cz>
33824 * doc/invoke.texi: Document options that can't be combined with
33825 -fcheck-pointer-bounds.
33827 2017-03-14 Martin Liska <mliska@suse.cz>
33829 PR middle-end/79831
33830 * doc/invoke.texi (-Wchkp): Document the option.
33832 2017-03-14 Martin Liska <mliska@suse.cz>
33834 * Makefile.in: Install gcov-dump.
33836 2017-03-14 Martin Liska <mliska@suse.cz>
33838 * multiple_target.c (expand_target_clones): Bail out for
33839 an invalid attribute.
33841 2017-03-14 Richard Biener <rguenther@suse.de>
33843 * alias.c (struct alias_set_entry): Pack properly.
33844 * cfgloop.h (struct loop): Likewise.
33845 * cse.c (struct set): Likewise.
33846 * ipa-utils.c (struct searchc_env): Likewise.
33847 * loop-invariant.c (struct invariant): Likewise.
33848 * lra-remat.c (struct cand): Likewise.
33849 * recog.c (struct change_t): Likewise.
33850 * rtl.h (struct address_info): Likewise.
33851 * symbol-summary.h (function_summary): Likewise.
33852 * tree-loop-distribution.c (struct partition): Likewise.
33853 * tree-object-size.c (struct object_size_info): Likewise.
33854 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
33855 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
33856 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
33857 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
33858 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
33859 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
33860 (struct _stmt_vec_info): Likewise.
33862 2017-03-14 Martin Liska <mliska@suse.cz>
33865 * multiple_target.c (create_dispatcher_calls): Check that
33866 a target can create a function dispatcher.
33868 2017-03-14 Martin Liska <mliska@suse.cz>
33871 * multiple_target.c (expand_target_clones): Drop local.local
33872 flag for default implementation.
33874 2017-03-14 Richard Biener <rguenther@suse.de>
33876 PR tree-optimization/80030
33877 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
33879 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
33881 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
33882 gcc_fallthrough() instead of __attribute__((fallthrough));
33884 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33886 * doc/gcc.texi: Remove "up" link to (DIR).
33887 * doc/gccint.texi: Ditto.
33889 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33891 * doc/install.texi (Specific) <avr>: Remove reference to
33894 2017-03-13 Jeff Law <law@redhat.com>
33896 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
33897 attribute rather than comments.
33899 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
33900 match_scratch operand is highest.
33902 2017-03-13 Martin Liska <mliska@suse.cz>
33904 PR middle-end/78339
33905 * ipa-pure-const.c (warn_function_noreturn): If the declarations
33906 is a CHKP clone, use original declaration.
33908 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33910 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
33911 (arc_conditional_register_usage): Use a different allocation order
33912 when optimizing for size.
33913 * common/config/arc/arc-common.c (arc_option_optimization_table):
33914 Section anchors default on when optimizing for size.
33916 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33918 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
33920 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33922 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
33923 * config/arc/arc.md (cpu_facility): Add cd variant.
33924 (*movqi_insn): Add code density variant.
33925 (*movhi_insn): Likewise.
33926 (*movqi_insn): Likewise.
33927 (*addsi3_mixed): Likewise.
33928 (subsi3_insn): Likewise.
33930 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33932 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33934 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33936 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33937 expressions with MINUS and UNARY ops.
33939 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33942 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33944 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33945 between vec_select and vector argument.
33946 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33947 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33948 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33949 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33950 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33951 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33953 2017-03-13 Richard Biener <rguenther@suse.de>
33956 * params.def (vect-max-peeling-for-alignment): Fix typo.
33958 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33960 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33961 issue that only occurred with binutils below 2.18.
33963 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33965 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33966 refer to binutils 2.11/2.12 minimum.
33968 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33970 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33971 ftp.kernel.org and simplify binutils requirement.
33973 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33975 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33977 (Optimize Options): Ditto. Also remove redundancy.
33979 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33981 PR translation/79848
33982 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33984 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33985 to G_ to avoid double translation.
33987 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33989 PR translation/79923
33990 * auto-profile.c (get_combined_location): Convert leading
33991 character of diagnostics to lower case and remove trailing period.
33992 (read_profile): Likewise for various diagnostics.
33993 * config/arm/arm.c (arm_option_override): Remove trailing period
33994 from various diagnostics.
33995 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33996 (msp430_expand_delay_cycles): Likewise.
33998 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34001 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
34002 full command-line argument, rather than just "str".
34003 (aarch64_validate_march): Likewise.
34004 (aarch64_validate_mtune): Likewise.
34006 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
34008 PR rtl-optimization/78911
34009 * lra-assigns.c (must_not_spill_p): New function.
34010 (spill_for): Use it.
34012 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34014 PR tree-optimization/79981
34015 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
34016 ATOMIC_COMPARE_EXCHANGE ifn result.
34017 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
34018 IFN_ATOMIC_COMPARE_EXCHANGE.
34020 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34023 * opts.c (parse_sanitizer_options): Add missing question mark to
34024 "did you mean" message.
34026 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34028 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
34030 (VMULEUH_UNS): Likewise.
34031 (VMULOUB_UNS): Likewise.
34032 (VMULOUH_UNS): Likewise.
34033 * config/rs6000/rs6000.c (builtin_function_type): Remove
34034 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
34036 2017-03-10 David Malcolm <dmalcolm@redhat.com>
34039 * read-rtl-function.c (function_reader::read_rtx_operand): Update
34040 x with result of extra_parsing_for_operand_code_0.
34041 (function_reader::extra_parsing_for_operand_code_0): Convert
34042 return type from void to rtx, returning x. When reading
34043 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
34044 larger size containing struct block_symbol.
34046 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
34048 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
34049 -mfloat128-hardware without -m64.
34051 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
34054 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
34055 entries to the case statement that marks unsigned arguments to
34056 overloaded functions.
34058 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34060 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
34061 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
34063 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
34066 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
34067 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
34069 2017-03-10 Martin Liska <mliska@suse.cz>
34073 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
34074 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
34077 2017-03-10 Olivier Hainque <hainque@adacore.com>
34079 * tree-switch-conversion (array_value_type): Start by resetting
34080 candidate type to it's main variant.
34082 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34084 PR rtl-optimization/79909
34085 * combine.c (try_combine): Use simplify_replace_rtx on individual
34086 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
34087 of the whole CALL_INSN_FUNCTION_USAGE.
34089 PR tree-optimization/79972
34090 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
34091 get_range_info on SSA_NAMEs. Formatting fixes.
34093 2017-03-10 Richard Biener <rguenther@suse.de>
34094 Jakub Jelinek <jakub@redhat.com>
34096 PR tree-optimization/77975
34097 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
34098 edge to be constant.
34099 (get_val_for): For constant x return it. Formatting fix.
34100 (loop_niter_by_eval): Avoid pointless looping if the next iteration
34101 would use the same bases as the current one.
34103 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34105 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
34106 instead of vec_select for V1TImode.
34107 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
34109 (VSX_LE_128): Add V1TI to this mode iterator.
34110 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
34111 (*vsx_le_perm_store_<mode>): Likewise.
34112 (pre-reload splitter for VSX stores): Likewise.
34113 (post-reload splitter for VSX stores): Likewise.
34114 (*vsx_xxpermdi2_le_<mode>): Likewise.
34115 (*vsx_lxvd2x2_le_<mode>): Likewise.
34116 (*vsx_stxvd2x2_le_<mode>): Likewise.
34118 2017-03-09 Michael Eager <eager@eagercon.com>
34120 Correct failures with --enable-checking=yes,rtl.
34122 * config/microblaze/microblaze.c (microblaze_expand_shift):
34123 Replace GET_CODE test with CONST_INT_P and INTVAL test with
34124 test for const0_rtx.
34125 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
34126 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
34128 2017-03-09 Richard Biener <rguenther@suse.de>
34130 PR tree-optimization/79977
34131 * graphite-scop-detection.c (scop_detection::merge_sese):
34132 Handle the case of extra exits to blocks dominating the entry.
34134 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
34136 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
34139 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
34141 PR rtl-optimization/79949
34142 * lra-constraints.c (process_alt_operands): Check memory when
34143 trying to predict a cycle. Print about the overall increase.
34145 2017-03-09 Richard Biener <rguenther@suse.de>
34147 PR middle-end/79971
34148 * gimple-expr.c (useless_type_conversion_p): Preserve
34149 TYPE_SATURATING for fixed-point types.
34151 2017-03-09 Richard Biener <rguenther@suse.de>
34154 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
34155 alignment of BLKmode params.
34157 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34160 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
34161 (VALL_NO_V2Q): Likewise.
34163 * config/aarch64/aarch64-simd.md
34164 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
34166 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
34167 VALL_NO_V2Q mode iterator.
34168 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
34170 2017-03-09 Martin Liska <mliska@suse.cz>
34172 PR tree-optimization/79631
34173 * tree-chkp-opt.c (chkp_is_constant_addr): Call
34174 tree_int_cst_sign_bit just for INTEGER constants.
34176 2017-03-09 Martin Liska <mliska@suse.cz>
34180 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
34183 2017-03-09 Marek Polacek <polacek@redhat.com>
34186 * tree.c (inchash::add_expr): Handle TREE_VEC.
34188 2017-03-09 Martin Liska <mliska@suse.cz>
34191 (chkp_narrow_size_and_offset): New function.
34192 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
34193 (void chkp_parse_bit_field_ref): New function.
34194 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
34195 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
34197 2017-03-09 Martin Liska <mliska@suse.cz>
34200 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
34201 (chkp_find_bounds_1): Remove gcc_unreachable.
34203 2017-03-09 Jakub Jelinek <jakub@redhat.com>
34206 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
34207 BUILT_IN_SYNC*, determine the access type from the size suffix and
34208 always build a MEM_REF with that type. Handle forgotten
34209 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
34212 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
34213 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
34214 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
34215 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
34216 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
34217 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
34218 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
34219 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
34220 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
34221 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
34222 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
34223 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
34224 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
34225 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
34226 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
34227 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
34228 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
34229 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
34230 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
34231 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
34232 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
34233 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
34234 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
34235 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
34236 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
34237 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
34238 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
34239 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
34240 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
34241 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
34242 definitions outside of __OPTIMIZE__ guarded section.
34245 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
34246 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
34247 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
34248 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
34251 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34253 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
34254 ("vfenez<mode>"): Add missing constraints.
34256 2017-03-08 Martin Sebor <msebor@redhat.com>
34259 * config/nds32/nds32.c (nds32_option_override):
34260 Fix misspelled diagnostic.
34262 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34265 * gimplify.c (gimplify_omp_for): Replace index var in outer
34266 taskloop statement with an artificial variable and add
34267 OMP_CLAUSE_PRIVATE clause for it.
34269 2017-03-08 Richard Biener <rguenther@suse.de>
34271 PR tree-optimization/79955
34272 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
34273 for accesses that are completely outside of the variable.
34275 2017-03-08 Andrew Haley <aph@redhat.com>
34277 PR tree-optimization/79943
34278 * tree-ssa-loop-split.c (compute_new_first_bound): When
34279 calculating the new upper bound, (END-BEG) should be added, not
34282 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34284 * config/avr/avr.md (setmemhi): Make sure match_dup
34285 operand number comes before match_scratch.
34287 2017-03-08 Richard Biener <rguenther@suse.de>
34289 PR tree-optimization/79920
34290 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
34291 with ncopies == 1 to ...
34292 (vect_transform_slp_perm_load): ... here. Properly compute
34293 all element loads by iterating VF times over the group. Do
34294 not handle ncopies (computed in a broken way) in
34295 vect_create_mask_and_perm.
34297 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34300 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
34301 is a uniform vector, use uniform_vector_p return value instead of
34302 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
34304 2017-03-07 Marek Polacek <polacek@redhat.com>
34306 PR middle-end/79809
34307 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
34308 (alloca_call_type): Likewise.
34310 2017-03-07 Martin Liska <mliska@suse.cz>
34312 * gcov.c (process_args): Put comment to correct location.
34314 2017-03-07 Martin Liska <mliska@suse.cz>
34316 PR middle-end/68270
34317 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
34318 Use array_at_struct_end_p instead of DECL_CHAIN (field).
34319 (chkp_narrow_bounds_for_field): Likewise.
34320 (chkp_parse_array_and_component_ref): Pass one more argument to
34323 2017-03-07 Richard Biener <rguenther@suse.de>
34325 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
34328 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
34330 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
34331 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
34333 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34336 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
34337 to end of description.
34338 (PARAM_MAX_STORES_TO_MERGE): Likewise.
34340 2017-03-07 Jakub Jelinek <jakub@redhat.com>
34342 PR rtl-optimization/79901
34343 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
34345 (*avx512f_<code><mode>3<mask_name>): ... this.
34346 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
34347 iterator instead of VI8_AVX2_AVX512BW.
34349 PR rtl-optimization/79901
34350 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
34351 min/max expander, expand it using expand_vec_cond_expr.
34354 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
34357 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34360 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
34361 to void * for PCH reasons.
34362 * dwarf2out.c (output_loc_operands, output_die): Cast
34363 v.val_vec.array to unsigned char *.
34365 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
34368 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
34371 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
34373 PR rtl-optimization/79571
34374 * lra-constraints.c (process_alt_operands): Calculate static
34375 reject and subtract it from overall when only addresses will be
34378 2017-03-06 Julia Koval <julia.koval@intel.com>
34381 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
34382 incoming stack boundary to 128 for 64-bit targets.
34384 2017-03-06 Richard Biener <rguenther@suse.de>
34386 PR tree-optimization/79894
34387 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
34388 to NULL after folding it.
34390 2017-03-06 Richard Biener <rguenther@suse.de>
34392 PR tree-optimization/79824
34393 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
34394 check disabling peeling for gaps.
34396 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
34398 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
34399 attributes): Document gettimeofday.
34401 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34403 * config/s390/s390.c (s390_option_override_internal): Set
34404 PARAM_MIN_VECT_LOOP_BOUND
34406 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34408 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
34409 * config/s390/s390.md: Likewise.
34411 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34414 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
34415 (<avx2_avx512>_perm<mode>): Rename to ...
34416 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
34418 (<avx512>_perm<mode>_mask): Rename to ...
34419 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
34421 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
34422 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
34423 instead of VI8F_256_512.
34424 (avx512f_perm<mode>): New define_expand.
34425 (avx512f_perm<mode>_mask): Likewise.
34426 (avx512f_perm<mode>_1<mask_name>): New define_insn.
34427 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
34429 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34431 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
34432 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
34434 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
34436 2017-03-06 Martin Liska <mliska@suse.cz>
34439 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
34440 when having a SSA NAME w/o VAR_DECL assigned to it.
34442 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34444 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
34445 msa_dpsub_<su>_d): Fix MODE for vec_select.
34447 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34449 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
34451 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
34453 2017-03-06 Richard Biener <rguenther@suse.de>
34455 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
34456 * plugin.c (register_plugin_info): Likewise.
34457 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
34459 2017-03-05 Jakub Jelinek <jakub@redhat.com>
34461 * config/i386/sse.md (sse_storehps, sse_storelps,
34462 avx_<castmode><avxsizesuffix>_<castmode>,
34463 avx512f_<castmode><avxsizesuffix>_<castmode>,
34464 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
34465 in condition that at least one operand is not a MEM.
34467 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34469 PR middle-end/79805
34470 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
34471 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
34473 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
34474 gimple_call_nothrow_p flag based on whether original builtin can throw.
34475 If it can, emit following stmts on the fallthrough edge.
34476 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
34477 don't create new bb if inserting just debug stmts on the edge, try to
34478 insert them on the fallthru bb or just reset debug stmts.
34480 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
34483 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
34484 restore recog_data (including the operand rtxes inside it) around
34485 the call to get_insn_template.
34487 2017-03-03 Martin Sebor <msebor@redhat.com>
34489 PR tree-optimization/79699
34490 * context.c (context::~context): Free MPFR caches to avoid
34491 a memory leak on program exit.
34493 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34495 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
34496 Use wide_int::ulow () instead of .elt (0).
34498 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34500 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
34501 (*pushxf): Limit oF constraint to 32bit targets and add oC
34502 constraint for 64bit targets.
34503 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
34504 (*pushdf): Change rmF constraint to rmC.
34506 2017-03-03 Martin Liska <mliska@suse.cz>
34508 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
34509 Remove unused variable.
34511 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34514 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
34515 is a memory operand, increase num_memory.
34516 (ix86_expand_args_builtin): Likewise.
34518 2017-03-03 Jan Hubicka <jh@suse.cz>
34521 * ipa-devirt.c (maybe_record_node): Properly handle
34522 __cxa_pure_virtual visibility.
34524 2017-03-03 Martin Liska <mliska@suse.cz>
34526 PR tree-optimization/79803
34527 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
34529 (pass_loop_prefetch::execute): Disabled optimization if an
34530 assumption about L1 cache size is not met.
34532 2017-03-03 Martin Liska <mliska@suse.cz>
34534 PR rtl-optimization/79574
34535 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
34536 (hash_scan_set): Likewise.
34537 (dump_hash_table): Likewise.
34538 (hoist_code): Likewise.
34540 2017-03-03 Richard Biener <rguenther@suse.de>
34542 * fixed-value.c (fixed_from_string): Restore use of elt (1)
34543 in place of uhigh ().
34544 (fixed_convert_from_real): Likewise.
34546 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34549 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
34551 2017-03-03 Richard Biener <rguenther@suse.de>
34553 PR middle-end/79818
34554 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
34555 TYPE_OVERFLOW_UNDEFINED check.
34557 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34559 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
34561 (vector_ae_<mode>_p): Likewise.
34562 (vector_nez_<mode>_p): Likewise.
34563 (vector_ne_v2di_p): Likewise.
34564 (vector_ae_v2di_p): Likewise.
34565 (vector_ne_<mode>_p): Likewise.
34566 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
34568 (vsx_tsqrt<mode>2_fe): Likewise.
34570 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
34573 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
34575 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34577 PR rtl-optimization/79780
34578 * cprop.c (one_cprop_pass): When second and further conditional trap
34579 in a single basic block is turned into an unconditional trap, turn it
34580 into a deleted note to avoid RTL verification failures.
34582 2017-03-02 Richard Biener <rguenther@suse.de>
34584 * fold-const.c (const_binop): Use ulow () instead of elt (0).
34586 2017-03-02 Richard Biener <rguenther@suse.de>
34588 PR tree-optimization/79345
34590 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
34591 param and abort the walk, returning -1 if it is hit.
34592 (walk_aliased_vdefs): Take a limit param and pass it on.
34593 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
34594 defaulting to 0 and return a signed int.
34595 * tree-ssa-uninit.c (struct check_defs_data): New struct.
34596 (check_defs): New helper.
34597 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
34598 about uninitialized memory.
34599 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
34600 bogus uninitialized warning.
34601 (fixed_convert_from_real): Likewise.
34603 2017-03-02 Bin Cheng <bin.cheng@arm.com>
34605 PR tree-optimization/66768
34606 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
34607 iv_use if base object can't be determined.
34609 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34611 PR tree-optimization/79345
34612 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
34613 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
34614 (get_pattern_stats): Initialize it.
34615 * genemit.c (gen_expand): Verify match_scratch numbers come after
34616 match_operand/match_dup numbers.
34617 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
34618 match_scratch numbers.
34619 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
34621 * config/s390/s390.md (trunctdsd2): Likewise.
34623 2017-03-02 Richard Biener <rguenther@suse.de>
34625 * wide-int.h (wide_int_storage::operator=): Implement in terms
34628 2017-03-02 Richard Biener <rguenther@suse.de>
34630 PR tree-optimization/79777
34631 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
34632 the to insert expression to sth existing.
34634 2017-03-01 Martin Sebor <msebor@redhat.com>
34636 PR middle-end/79692
34637 * gimple-ssa-sprintf.c
34638 (directive::known_width_and_precision): New function.
34639 (format_integer): Use it.
34640 (get_mpfr_format_length): Consider the full range of precision
34641 when computing %g output with the # flag. Set the likely byte
34642 count to 3 rather than 1 when precision is indeterminate.
34643 (format_floating): Correct the lower bound of precision.
34645 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34647 * doc/invoke.texi: Document default code model for 64-bit Linux.
34649 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34652 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
34653 udiv rather than div since input pattern is unsigned.
34655 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
34657 * config/i386/i386.c (print_reg): Warn for values of
34658 unsupported size in integer register.
34660 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
34663 * config/rs6000/predicates.md (current_file_function_operand): Do
34664 not allow self calls to be local if the function is replaceable.
34666 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
34669 * config/rs6000/altivec.h (vec_ctz and others): Change the
34670 preprocessor macro that controls conditional compilation from
34671 _ARCH_PWR9 to __POWER9_VECTOR__.
34672 (vec_all_ne): Change parameterization of __altivec_scalar_pred
34673 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
34674 control (instead of _ARCH_PWR9 control) so that template
34675 definition uses power9-specific function.
34676 (vec_any_eq): Likewise.
34677 (vec_all_ne): Change macro definition to use a power9-specific
34678 expansion under #ifdef __POWER9_VECTOR__ control (instead of
34679 _ARCH_PWR9 control).
34680 (vec_any_eq) Likewise.
34681 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
34682 expansion for CMPNEF to remove support for xvcmpnesp instruction.
34683 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
34684 support for xvcmpnedp instruction.
34685 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
34686 macro expansion so that Power9 implementation of vec_all_ne does
34687 not use the AltiVec predicate framework.
34688 (VCMPNEH_P): Likewise.
34689 (VCMPNEW_P): Likewise.
34690 (VCMPNED_P): Likewise.
34691 (VCMPNEFP_P): Likewise.
34692 (VCMPNEDP_P): Likewise.
34693 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
34694 implementation of vec_any_eq to not use AltiVec predicate
34696 (VCMPAEH_P): Likewise.
34697 (VCMPAEW_P): Likewise.
34698 (VCMPAED_P): Likewise.
34699 (VCMPAEFP_P): Likewise.
34700 (VCMPAEDP_P): Likewise.
34701 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
34702 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
34703 not use the AltiVec predicate framework.
34704 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
34705 of vec_any_eq to not use AltiVec predicate framework.
34706 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34707 support for predefined __POWER9_VECTOR__ macro to indicate that
34708 Power9 instruction selection is enabled.
34709 (altivec_overloaded_builtins): Remove extraneous
34710 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
34711 function argument types RS6000_BTI_bool_V16QI and
34712 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
34713 entry for overloaded function argument types RS6000_BTI_bool_V4SI
34714 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
34715 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
34716 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
34717 Power9 for implementations of vec_cmpne. Change the signature for
34718 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
34719 (representing vec_all_ne) to remove the previously described first
34720 argument of type RS6000_BTI_INTSI, as this was an artifact of
34721 reliance on the AltiVec predicate framework, which is no longer
34722 used in the implementation of these functions. Add
34723 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
34724 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
34725 since, unlike the AltiVec predicate framework implementation, we
34726 do not share function descriptors between vec_alle and vec_anyeq.
34727 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
34728 set of modes that receive special treatment even when
34729 TARGET_P9_VECTOR is true. The special treatment emits code that
34730 does not depend on Power9 instructions.
34731 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
34732 define_expand to not rely on AltiVec predicate framework.
34733 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34735 (vector_ne_v2di_p): Change this define_expand to not rely on
34736 AltiVec predicate framework.
34737 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
34739 (vector_ne_<mode>_p): Change this define_expand to not rely on
34740 AltiVec predicate framework.
34741 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34743 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
34744 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
34745 define_insn pattern.
34746 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
34747 define_insn pattern because the xvcmpne<VSs>. instruction is not
34749 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
34750 instruction is not supported.
34752 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34754 * config/nvptx/nvptx.c: Include intl.h.
34756 2017-03-01 Martin Jambor <mjambor@suse.cz>
34759 * ipa-prop.h (ipa_bits): Removed field known.
34760 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
34761 to pointers. Adjusted their comments to warn about their sharing.
34762 (ipcp_transformation_summary): Change bits to a vector of pointers.
34763 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
34764 (ipa_get_ipa_bits_for_value): Declare.
34765 * tree-vrp.h (value_range): Mark as GTY((for_user)).
34766 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
34767 (ipa_bits_hash_table): Likewise.
34768 (ipa_vr_ggc_hash_traits): Likewise.
34769 (ipa_vr_hash_table): Likewise.
34770 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
34771 being pointers and vr_known being removed.
34772 (ipa_set_jf_unknown): Likewise.
34773 (ipa_get_ipa_bits_for_value): New function.
34774 (ipa_set_jfunc_bits): Likewise.
34775 (ipa_get_value_range): New overloaded functions.
34776 (ipa_set_jfunc_vr): Likewise.
34777 (ipa_compute_jump_functions_for_edge): Use the above functions to
34778 construct bits and vr parts of jump functions.
34779 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
34780 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34782 (ipcp_grow_transformations_if_necessary): Also allocate
34783 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34785 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
34786 them. Fix too long lines.
34787 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
34788 vr_known being removed.
34789 (ipa_read_jump_function): Use new setter functions to construct bits
34790 and vr parts of jump functions or set them to NULL.
34791 (write_ipcp_transformation_info): Adjust for bits being pointers.
34792 (read_ipcp_transformation_info): Likewise.
34793 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
34795 Include gt-ipa-prop.h.
34796 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
34798 (ipcp_store_bits_results): Likewise.
34799 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
34800 Do not write to existing jump functions but use a temporary instead.
34802 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34805 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
34806 attempt to use its first operand as BIT_FIELD_REF base.
34808 2017-03-01 Richard Biener <rguenther@suse.de>
34810 PR middle-end/79721
34811 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
34812 interpolating formula in wrapping arithmetic.
34813 (chrec_apply): Convert chrec_evaluate return value to wanted type.
34815 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34817 PR tree-optimization/79734
34818 * tree-vect-generic.c (expand_vector_condition): Optimize
34819 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
34820 Handle VEC_COND_EXPR where comparison has different inner width from
34821 type's inner width.
34823 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
34825 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
34826 markup, and similar issues. Remove @opindex entries for things
34827 that aren't options. Add missing -mmpy-option entries.
34829 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34831 PR tree-optimization/79737
34832 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
34833 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
34834 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
34835 instead of byte_size. Formatting fix.
34836 (shift_bytes_in_array_right): Formatting fix.
34838 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
34841 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
34842 condition on optimize for the leaf function test.
34844 2017-02-28 Martin Liska <mliska@suse.cz>
34847 * read-rtl-function.c (function_reader::handle_unknown_directive):
34848 Bail out when one uses -flto.
34850 2017-02-28 Martin Liska <mliska@suse.cz>
34852 * common.opt: Replace space with tabular for options of <number>
34854 * config/i386/i386.opt: Show <number> value for
34855 -mlarge-data-threshold.
34856 * opts.c (print_filtered_help): Do not display number in hexadecimal
34859 2017-02-28 Martin Liska <mliska@suse.cz>
34861 * common.opt: Fix --help=option -Q for options which are of
34864 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
34866 * config/i386/i386.c (print_reg): Error out for values
34867 of 8-bit size in invalid integer register.
34869 2017-02-28 Martin Sebor <msebor@redhat.com>
34871 PR tree-optimization/79691
34872 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
34874 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34877 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
34878 gcc_unreachable with output_operand_lossage.
34880 2017-02-28 Richard Biener <rguenther@suse.de>
34882 PR tree-optimization/79740
34883 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
34885 (visit_nary_op): Insert the nary into the hashtable if we
34886 pattern-matched sth.
34887 * tree-ssa-pre.c (eliminate_insert): Robustify.
34889 2017-02-28 Richard Biener <rguenther@suse.de>
34891 PR middle-end/79731
34892 * fold-const.c (decode_field_reference): Reject out-of-bound
34895 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34897 * config/i386/i386.c: Include intl.h.
34898 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
34899 instead of just cond ? "..." : "...".
34900 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
34901 * coverage.c (read_counts_file): Likewise.
34902 * omp-offload.c: Include intl.h.
34903 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
34904 of just cond ? "..." : "...".
34905 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
34906 of just cond ? "..." : "...".
34908 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
34911 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
34913 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
34914 'tune for' CPU name.
34915 * config/arm/arm-cpu-data.h: Regenerated.
34917 2017-02-28 Richard Biener <rguenther@suse.de>
34919 PR tree-optimization/79732
34920 * tree-inline.c (expand_call_inline): Do not shadow var.
34922 2017-02-28 Richard Biener <rguenther@suse.de>
34924 PR tree-optimization/79723
34925 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
34926 address-space properly.
34928 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34930 * doc/optinfo.texi (Optimization groups): Fix option used for
34932 * doc/invoke.texi (-fopt-info): Document "omp".
34933 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34934 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34935 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34936 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34937 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34939 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34941 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34942 OPTGROUP_OMP with "omp".
34944 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34947 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34948 for arithmetic shift of unsigned V2DI.
34950 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34952 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34953 arc/linux.h headers.
34954 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34955 (LINK_SPEC): Likewise.
34956 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34957 (EXTRA_SPECS): Likewise.
34958 (STARTFILE_SPEC): Likewise.
34959 (ENDFILE_SPEC): Likewise.
34960 (LIB_SPEC): Likewise.
34961 (TARGET_SDATA_DEFAULT): Likewise.
34962 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34963 (MULTILIB_DEFAULTS): Likewise.
34964 (DWARF2_UNWIND_INFO): Likewise.
34965 * config/arc/big.h: New file.
34966 * config/arc/elf.h: Likewise.
34967 * config/arc/linux.h: Likewise.
34968 * config/arc/t-uClibc: Remove.
34970 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34972 PR tree-optimization/77536
34973 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34974 (tree_transform_and_unroll_loop): Use above function to compute the
34975 estimated niter of unrolled loop and use it when scaling profile.
34976 Also use count info rather than frequency if it's non-zero.
34977 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34978 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34979 (vect_transform_loop): Call above function.
34981 2017-02-27 Richard Biener <rguenther@suse.de>
34983 PR tree-optimization/45397
34984 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34985 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34986 (visit_nary_op): Add pattern matching for CSEing sign-changed
34987 or truncated operations with wider ones.
34989 2017-02-27 Richard Biener <rguenther@suse.de>
34991 PR tree-optimization/79690
34992 * tree-vect-stmts.c (vectorizable_store): Use vector type
34993 built from the DR with address-space.
34995 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34997 * doc/invoke.texi (Optimize Options): Refine the description
34998 of asan-use-after-return.
35000 2017-02-25 Alan Modra <amodra@gmail.com>
35002 PR rtl-optimization/79584
35003 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
35004 base, not ad->base_term, the reg within base. Remove assertion
35005 that ad->base == ad->base_term. Replace gen_int_mode using
35006 bogus mode with const0_rtx.
35008 2017-02-25 Jakub Jelinek <jakub@redhat.com>
35010 PR middle-end/79396
35011 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
35012 FMA_EXPR like tcc_binary or tcc_unary.
35014 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
35017 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
35019 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
35020 (output_loc_operands): Handle DW_OP_call_ref and
35021 DW_OP_GNU_variable_value.
35022 (struct variable_value_struct): New type.
35023 (struct variable_value_hasher): Likewise.
35024 (variable_value_hash): New variable.
35025 (string_types): Remove.
35026 (copy_loc_descr): New function.
35027 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
35028 (prepend_loc_descr_to_each): New function.
35029 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
35030 instead of add_loc_descr_to_each if the first argument is single
35031 location list and the second has multiple.
35032 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
35033 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
35034 when looking for variable value which doesn't have other location info.
35035 (loc_list_from_tree): Formatting fix.
35036 (gen_array_type_die): Simplify DW_AT_string_length handling.
35037 (adjust_string_types): Remove.
35038 (gen_subprogram_die): Don't call adjust_string_types nor test/set
35039 string_types. Call resolve_variable_values.
35040 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
35041 (resolve_addr_in_expr): Likewise. Add A argument.
35042 (copy_deref_exprloc): Remove deref argument. Adjust for the
35043 original expression being DW_OP_GNU_variable_value with optionally
35044 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
35045 optionally after it.
35046 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
35047 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
35048 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
35049 (variable_value_hasher::hash, variable_value_hasher::equal): New
35051 (resolve_variable_value_in_expr, resolve_variable_value,
35052 resolve_variable_values, note_variable_value_in_expr,
35053 note_variable_value): New functions.
35054 (dwarf2out_early_finish): Call note_variable_value on all toplevel
35057 2017-02-24 Jakub Jelinek <jakub@redhat.com>
35060 * opts.h (handle_generated_option): Add GENERATED_P argument.
35061 * opts-common.c (handle_option): Adjust function comment.
35062 (handle_generated_option): Add GENERATED_P argument, pass it to
35064 (control_warning_option): Pass false to handle_generated_option
35066 * opts.c (maybe_default_option): Pass true to handle_generated_option
35068 * optc-gen.awk: Likewise.
35070 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35072 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
35073 a REG, look at the REG it is a SUBREG of.
35074 (splitter for cmpeqsi_t): Ditto.
35076 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35078 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
35079 the special USEs with the pattern of the insn, not the insn itself.
35081 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
35084 * doc/invoke.texi: Document -mload-store-pairs.
35086 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35087 Sandra Loosemore <sandra@codesourcery.com>
35089 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
35090 argument isn't a CONST_INT.
35091 (nios2_alternate_compare_const): Assert op is a CONST_INT.
35092 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
35093 (nios2_validate_compare): Bypass alternate compare logic if *op2
35094 is not a CONST_INT.
35095 (ldstwm_operation_p): Return false if first_base is not a REG or
35096 if first_offset is not a CONST_INT.
35098 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35100 * config/cris/cris.md: Use correct operand in a define_peephole2.
35102 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35104 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
35106 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35108 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
35109 this_insn if it is an INSN or JUMP_INSN.
35110 (force_offsettable): Look at base, not at addr.
35111 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
35112 on things that aren't necessarily CONST_INTs.
35114 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
35116 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
35117 -mfpmath=sse is the default also for x86-32 targets with SSE2
35118 instruction set when @option{-ffast-math} is enabled
35120 2017-02-24 Jeff Law <law@redhat.com>
35122 PR rtl-optimizatoin/79286
35123 * ira.c (update_equiv_regs): Drop may_trap_p exception to
35126 2017-02-24 Richard Biener <rguenther@suse.de>
35128 PR tree-optimization/79389
35129 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
35132 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
35134 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
35135 function comment to reflect reality.
35136 (loop_exits_before_overflow): Fix typo in function description.
35138 2017-02-24 Richard Biener <rguenther@suse.de>
35140 PR tree-optimization/79389
35141 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
35142 properly that a threading opportunity exists. Detect conditional
35143 copy/constant propagation opportunities.
35145 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
35147 * config/visium/visium.md (type): Add trap.
35148 (b): New mode attribute.
35149 (*btst): Rename into...
35150 (*btst<mode>): ...this and adjust.
35151 (*cbranchsi4_btst_insn): Rename into...
35152 (*cbranch<mode>4_btst_insn): ...this and adjust.
35153 (trap): New define_insn.
35155 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35157 PR tree-optimization/79389
35158 * ifcvt.c (struct noce_if_info): Add rev_cond field.
35159 (noce_reversed_cond_code): New function.
35160 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
35161 reversed_comparison_code. Formatting fix.
35162 (noce_try_store_flag): Test rev_cond != NULL in addition to
35163 reversed_comparison_code.
35164 (noce_try_store_flag_constants): Likewise.
35165 (noce_try_store_flag_mask): Likewise.
35166 (noce_try_addcc): Use rev_cond if non-NULL instead of
35167 reversed_comparison_code.
35168 (noce_try_cmove_arith): Likewise. Formatting fixes.
35169 (noce_try_minmax, noce_try_abs): Clear rev_cond.
35170 (noce_find_if_block): Initialize rev_cond.
35171 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
35172 instead of false as last argument never attempt to reverse it
35175 2017-02-23 Bin Cheng <bin.cheng@arm.com>
35177 PR tree-optimization/79663
35178 * tree-predcom.c (combine_chains): Process refs in reverse order
35179 only for ZERO length chains, and add explaining comment.
35181 2017-02-23 Jeff Law <law@redhat.com>
35183 PR tree-optimization/79578
35184 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
35185 in call to operand_equal_p.
35187 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
35190 * config/i386/cpuid.h: Fix another undefined behavior.
35192 2017-02-23 Richard Biener <rguenther@suse.de>
35194 PR tree-optimization/79683
35195 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
35196 vector types for data-refs.
35198 2017-02-23 Martin Liska <mliska@suse.cz>
35200 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
35202 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35204 PR middle-end/79665
35205 * internal-fn.c (get_range_pos_neg): Moved to ...
35206 * tree.c (get_range_pos_neg): ... here. No longer static.
35207 * tree.h (get_range_pos_neg): New prototype.
35208 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
35209 are known to be in between 0 and signed maximum inclusive, try to
35210 expand both unsigned and signed divmod and use the cheaper one from
35213 2017-02-22 Jeff Law <law@redhat.com>
35215 PR tree-optimization/79578
35216 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
35217 to compare base operands.
35219 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
35222 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
35223 gpc_reg_operand instead of fpr_reg_operand.
35225 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
35227 * config/mips/mips.c (mips_return_in_memory): Force FP
35228 vector types to be returned in memory for o32 ABI.
35230 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35232 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
35233 instead of DW_TAG_member for static data member declarations and don't
35234 set no_linkage_name for static inline data members.
35235 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
35238 2017-02-22 Martin Liska <mliska@suse.cz>
35240 * doc/invoke.texi: Replace inequality signs with square brackets
35243 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
35246 * lra-constraints.c (simplify_operand_subreg): Handle
35247 WORD_REGISTER_OPERATIONS targets.
35249 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35252 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
35253 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
35254 elimination by swapping fld*.
35256 2017-02-22 Richard Biener <rguenther@suse.de>
35258 PR tree-optimization/79673
35259 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
35260 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
35261 irrelevant address-space qualifiers and avoiding a
35262 ADDR_SPACE_CONVERT_EXPR from fold_convert.
35264 2017-02-22 Richard Biener <rguenther@suse.de>
35266 PR tree-optimization/79666
35267 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
35268 to not symbolically negate if that may introduce undefined
35271 2017-02-22 Martin Liska <mliska@suse.cz>
35274 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
35275 * data-streamer-out.c (streamer_write_gcov_count_stream):
35277 * value-prof.c (stream_out_histogram_value): Make assert more
35278 precise based on type of counter.
35280 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
35283 * config/i386/i386.md (standard_x87sse_constant_load splitter):
35284 Use nonimmediate_operand instead of memory_operand for operand 1.
35285 (float-extend standard_x87sse_constant_load splitter): Ditto.
35287 2017-02-21 Jeff Law <law@redhat.com>
35289 PR tree-optimization/79621
35290 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
35291 blocks with edges to themselves.
35293 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35296 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
35297 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
35298 Use gimple_call_builtin_p.
35301 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
35302 on temporarily removed DEBUG_INSNs.
35304 PR tree-optimization/79649
35305 * tree-loop-distribution.c (classify_partition): Give up on
35306 non-generic address space loads/stores.
35308 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
35310 * doc/loop.texi (Loop manipulation): Remove nonexistent
35311 tree_ssa_loop_version from the documentation.
35312 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
35314 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35317 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
35318 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
35319 * config/rs6000/rs6000.c: Include except.h.
35320 (rs6000_expand_split_stack_prologue): Call
35321 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
35323 2017-02-21 Martin Jambor <mjambor@suse.cz>
35326 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
35327 have been analyzed.
35329 2017-02-21 Martin Jambor <mjambor@suse.cz>
35331 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
35332 for backward compatibility only.
35333 * doc/invoke.texi (Option Summary): Remove all references to
35334 -fipa-cp-alignment.
35336 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
35340 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35342 * lra-constraints.c (curr_insn_transform): Handle
35343 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35345 2017-02-21 Martin Liska <mliska@suse.cz>
35347 * config/i386/i386.opt: Replace -masm-dialect with -masm.
35349 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
35351 PR translation/79638
35352 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
35354 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
35357 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
35358 (arm_function_ok_for_sibcall): Return false for an indirect call by
35359 descriptor if all the argument registers are used.
35360 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
35361 alignment of the function.
35363 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35365 PR tree-optimization/61441
35366 * simplify-rtx.c (simplify_const_unary_operation): For
35367 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
35368 the sNaN unmodified.
35370 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
35372 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
35373 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
35374 instead of SYSTEM_HEADER_DIR.
35376 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
35377 Martin Liška <mliska@suse.cz>
35379 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
35380 Fix typos and grammar, use active voice, and clarify.
35382 2017-02-20 Marek Polacek <polacek@redhat.com>
35384 PR middle-end/79537
35385 * gimplify.c (gimplify_expr): Handle unused *&&L;.
35388 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
35390 2017-02-20 Jakub Jelinek <jakub@redhat.com>
35393 * config/i386/i386.c (ix86_expand_builtin): Handle
35394 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
35395 ix86_builtins_isa[fcode].isa as a requirement of those
35396 flags and any other flag in the bitmask.
35397 (ix86_init_mmx_sse_builtins): Use 0 instead of
35398 ~OPTION_MASK_ISA_64BIT as mask.
35399 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
35400 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
35401 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
35402 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
35404 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35407 * lra-constraints.c (split_reg): Check requested split mode
35408 is supported by the register.
35410 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35412 * lra-constraints.c (simplify_operand_subreg): Remove early
35415 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35418 * lra-constraints.c (curr_insn_transform): Tighten condition
35419 for converting SUBREG reloads from OP_OUT to OP_INOUT.
35421 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35424 * lra-constraints.c (curr_insn_transform): Handle
35425 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35427 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
35430 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
35432 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
35434 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
35437 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
35440 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35442 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
35445 2017-02-18 Jakub Jelinek <jakub@redhat.com>
35447 * final.c (last_columnnum, override_columnnum): New variables.
35448 (final_start_function): Set last_columnnum, pass it to begin_prologue
35449 hook and pass 0 to dwarf2out_begin_prologue.
35450 (final_scan_insn): Update override_columnnum. Pass last_columnnum
35451 to source_line debug hook.
35452 (notice_source_line): Compute last_columnnum and for debug_column_info
35453 return true on column changes.
35454 * debug.h (struct gcc_debug_hooks): Add column argument to
35455 source_line and begin_prologue hooks.
35456 (debug_nothing_int_charstar_int_bool): Remove prototype.
35457 (debug_nothing_int_int_charstar,
35458 debug_nothing_int_int_charstar_int_bool): New prototypes.
35459 (dwarf2out_begin_prologue): Add column argument.
35460 * debug.c (do_nothing_debug_hooks): Adjust source_line and
35461 begin_prologue hooks.
35462 (debug_nothing_int_charstar_int_bool): Remove.
35463 (debug_nothing_int_int_charstar,
35464 debug_nothing_int_int_charstar_int_bool): New functions.
35465 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
35466 through to dwarf2out_source_line.
35467 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
35468 (dwarf2out_source_line): Add column argument, emit it if requested.
35469 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
35471 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35472 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35473 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
35474 through to dwarf2out_begin_prologue.
35475 (vmsdbgout_source_line): Add column argument, pass it through to
35476 dwarf2out_source_line.
35477 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
35478 dbxout_source_line caller.
35479 (dbxout_source_line): Add column argument.
35481 * common.opt (gno-column-info, gcolumn-info): New options.
35482 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
35483 (check_die): Also test for multiple DW_AT_decl_column attributes.
35484 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
35485 DW_AT_decl_column if requested.
35486 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
35488 (gen_variable_die): Likewise.
35489 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
35490 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
35493 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
35494 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
35495 (ix86_handle_option): Handle OPT_m3dnowa.
35496 * doc/invoke.texi (-m3dnowa): Document.
35497 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
35498 -m3dnowa instead of -m3dnow -march=athlon.
35501 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
35502 instead of gcc_assert for K, r and R code checks. Formatting fixes.
35504 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35507 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
35508 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
35509 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
35510 generator for vsx_xxpermdi_<mode>_be.
35511 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
35512 force big-endian semantics.
35513 (vsx_xxpermdi_<mode>_be): New define_expand with same
35514 implementation as previous version of vsx_xxpermdi_<mode>.
35516 2017-02-17 Jakub Jelinek <jakub@redhat.com>
35518 PR tree-optimization/79327
35519 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
35520 variable, its initialization and use.
35522 2017-02-17 Julia Koval <julia.koval@intel.com>
35524 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
35525 (OPTION_MASK_ISA_PKU_UNSET): New.
35526 (ix86_handle_option): Handle -mrdpid.
35527 * config/i386/cpuid.h (bit_RDPID): New.
35528 * config/i386/driver-i386.c (host_detect_local_cpu):
35529 Detect RDPID feature.
35530 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
35531 * config/i386/i386-c.c (ix86_target_macros_internal):
35533 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
35534 (ix86_valid_target_attribute_inner_p): Add "rdpid".
35535 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
35536 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
35537 * config/i386/i386.md (define_insn "rdpid"): New.
35538 * config/i386/i386.opt Add -mrdpid.
35539 * config/i386/immintrin.h (_rdpid_u32): New.
35541 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
35543 PR rtl-optimization/79541
35544 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
35545 instead of transforming it into USE.
35547 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
35549 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
35550 If HONOR_SNANS (SFmode) force the input to a register.
35551 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
35552 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
35553 an frsp or similar insn.
35555 2017-02-17 Martin Liska <mliska@suse.cz>
35557 PR rtl-optimization/79577
35558 * params.def (selsched-max-sched-times): Increase minimum to 1.
35560 2017-02-17 Martin Liska <mliska@suse.cz>
35562 PR rtl-optimization/79574
35563 * gcse.c (want_to_gcse_p): Prevent integer overflow.
35565 2017-02-17 Martin Liska <mliska@suse.cz>
35567 PR tree-optimization/79529
35568 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
35569 ssa_defined_default_def_p to handle cases which are implicitly
35571 * tree-ssa.c (ssa_defined_default_def_p): New function.
35572 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
35573 which are implicitly defined.
35574 * tree-ssa.h (ssa_defined_default_def_p): Declare.
35576 2017-02-17 Richard Biener <rguenther@suse.de>
35578 PR middle-end/79576
35579 * params.def (max-ssa-name-query-depth): Limit to 10.
35581 2017-02-17 Richard Biener <rguenther@suse.de>
35583 PR tree-optimization/79552
35584 * tree-ssa-structalias.c (visit_loadstore): Properly verify
35587 2017-02-17 Richard Biener <rguenther@suse.de>
35590 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
35592 2017-02-17 Marek Polacek <polacek@redhat.com>
35594 PR middle-end/79536
35595 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
35596 (fold_negate_expr): New wrapper.
35598 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
35600 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
35601 Correct terminology and de-emphasize pre-standard behavior.
35603 2017-02-16 Alan Modra <amodra@gmail.com>
35605 PR rtl-optimization/79286
35606 * ira.c (def_dominates_uses): New function.
35607 (update_equiv_regs): Don't create an equivalence for insns that
35608 may trap where the register def does not dominate the use.
35610 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
35612 PR rtl-optimization/78127
35613 * lra.c (lra): Call lra_eliminate before finish the loop after
35616 2017-02-16 Richard Biener <rguenther@suse.de>
35618 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
35620 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
35621 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
35622 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
35623 (isl_val_int_from_wi): New function.
35624 (extract_affine_gmp): Rename to ...
35625 (extract_affine_wi): ... this, take a widest_int.
35626 (extract_affine_int): Just wrap extract_affine_wi.
35627 (add_param_constraints): Use isl_val_int_from_wi.
35628 (add_loop_constraints): Likewise, and extract_affine_wi.
35630 2017-02-15 Jeff Law <law@redhat.com>
35632 PR middle-end/79521
35633 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
35634 ira_init_register_move_cost_if_necessary.
35636 2017-02-15 Martin Sebor <msebor@redhat.com>
35638 PR middle-end/32003
35639 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
35640 removed in a prior commit.
35642 2017-02-15 Bin Cheng <bin.cheng@arm.com>
35644 PR tree-optimization/79347
35645 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
35646 counters during peeling.
35648 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
35650 * Makefile.in (site.exp): Remove "set ISLVER".
35652 2017-02-15 Jakub Jelinek <jakub@redhat.com>
35655 * real.c (real_from_integer): Call real_convert even for decimal.
35657 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
35660 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
35662 2017-02-14 Andrew Pinski <apinski@cavium.com>
35664 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
35665 cores and change the partno/implementer to be correct.
35666 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
35667 the 'B" as the implementer.
35668 * config/aarch64/aarch64-tune.md: Regenerate.
35670 2017-02-14 Carl Love <cel@us.ibm.com>
35672 * config/rs6000/rs6000.c: Add case statement entry to make the
35673 xvcvuxdsp built-in argument unsigned.
35674 * config/rs6000/vsx.md: Fix the source and return operand types so they
35675 match the instruction definitions from the ISA document. Fix typo
35676 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
35679 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
35682 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
35683 member early_clobber_alts.
35684 * lra-lives.c (reg_early_clobber_p): New.
35685 (process_bb_lives): Use it.
35686 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
35687 (debug_operand_data): Initialize early_clobber_alts.
35688 (setup_operand_alternative): Set up early_clobber_alts.
35689 (collect_non_operand_hard_regs): Ditto. Pass early clobber
35690 alternatives to new_insn_reg.
35691 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
35693 (lra_update_insn_regno_info): Pass the new arg.
35695 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35697 PR middle-end/79505
35698 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
35699 (new_oacc_loop_raw): Don't clear already cleared fields.
35702 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
35703 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
35704 _mm512_prefetch_i64gather_ps): New inline functions and macros.
35706 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
35709 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
35711 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
35714 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
35715 the extra instruction to the right place to store 128-bit constant
35718 2017-02-14 Martin Sebor <msebor@redhat.com>
35720 PR middle-end/79448
35721 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
35722 warning for strings of unknown length.
35724 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
35726 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
35728 2017-02-14 Jeff Law <law@redhat.com>
35731 * ira-costs.c (scan_one_insn): Initialize register move costs
35732 for pseudos seen in USE/CLOBBER insns.
35734 PR tree-optimization/79095
35735 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
35736 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
35737 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
35738 if the operands are known to be not equal, then the resulting range
35740 (intersect_ranges): If the new range is ~[0,0] and the old range is
35741 wide, then prefer ~[0,0].
35742 * tree-vrp.c (overflow_comparison_p_1): New function.
35743 (overflow_comparison_p): New function.
35744 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
35745 if NAME is used in an overflow test.
35746 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
35747 overflow check that can be expressed as an equality test, then adjust
35748 ops to be that equality test.
35750 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35752 * config/s390/s390-builtin-types.def: Remove flags argument.
35753 * config/s390/s390.c (s390_init_builtins): Likewise.
35755 2017-02-14 Martin Liska <mliska@suse.cz>
35757 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
35758 vector. Fix trailing white spaces.
35760 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
35762 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
35765 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35767 PR rtl-optimization/68664
35768 * config/arm/arm.c (arm_sched_can_speculate_insn):
35769 New function. Declare prototype.
35770 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35772 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35774 PR rtl-optimization/68664
35775 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
35777 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35779 2017-02-14 Amit Pawar <amit.pawar@amd.com>
35781 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
35782 max skip bytes for function, loop and jump.
35784 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35786 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
35787 ABS_EXPR for gimple dump.
35789 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35792 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
35794 PR tree-optimization/79408
35795 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
35796 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
35797 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
35798 also if rhs1 is INTEGER_CST.
35800 2017-02-14 Richard Biener <rguenther@suse.de>
35802 PR middle-end/79432
35803 * tree-into-ssa.c (insert_phi_nodes): When the function can
35804 have abnormal edges rewrite SSA names with broken use-def
35805 dominance out of SSA and register them for PHI insertion.
35807 2017-02-13 Martin Sebor <msebor@redhat.com>
35809 PR middle-end/79496
35810 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
35811 clearing info.nowrite flag when snprintf size argument is a range.
35813 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35815 * cprop.c (cprop_jump): Add missing space in string literal.
35816 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
35817 (get_constraint_for_component_ref): Likewise.
35818 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
35819 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
35820 * lra-constraints.c (process_alt_operands): Likewise.
35821 * ipa-inline.c (inline_small_functions): Likewise.
35822 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
35823 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
35824 * trans-mem.c (diagnose_tm_1_op): Likewise.
35825 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
35826 (grid_parallel_clauses_gridifiable): Likewise.
35828 * config/nvptx/mkoffload.c (process): Add space in between
35831 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
35832 "MOD4_SSE_REGS" and "ALL_REGS".
35834 * spellcheck.c (test_data): Add , in between "foo" and "food".
35836 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35839 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
35840 boundary crossing check and subsequent code generation agree.
35842 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35844 * config/aarch64/aarch64.c (has_memory_op): Delete.
35845 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
35848 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35850 PR rtl-optimization/79388
35851 PR rtl-optimization/79450
35852 * combine.c (distribute_notes): When removing TEM_INSN for which
35853 corresponding dest has last value recorded, invalidate that last
35856 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35858 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
35859 of explicit '@'. Add missing assembly comment marker on branch costs
35862 2017-02-13 Nathan Sidwell <nathan@acm.org>
35864 * gengtype-lex.l (<in_struct>): Add '/'.
35866 2017-02-13 Martin Liska <mliska@suse.cz>
35869 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
35871 2017-02-13 Richard Biener <rguenther@suse.de>
35873 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
35875 * configure: Re-generate.
35876 * config.in: Likewise.
35877 * graphite-dependences.c: Simplify as if
35878 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
35879 * graphite-isl-ast-to-gimple.c: Likewise.
35880 * graphite-optimize-isl.c: Likewise.
35881 * graphite-poly.c: Likewise.
35882 * graphite-sese-to-poly.c: Likewise.
35883 * graphite.h: Likewise.
35884 * toplev.c: Include isl/version.h and use isl_version () for
35885 printing the ISL version.
35886 * doc/install.texi: Update ISL requirement.
35888 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35890 * doc/standards.texi (Standards): Update reference to
35893 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35895 * doc/extend.texi (Named Address Spaces): sourceware.org now
35897 * doc/install.texi (Binaries): Ditto.
35900 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35902 * doc/cpp.texi: Replace "stringify"/"stringification" with C
35903 standard terminology "stringize"/"stringizing" throughout.
35904 * doc/cppinternals.texi: Likewise.
35906 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35908 * doc/extend.texi: Fix some spelling mistakes and typos.
35909 * doc/invoke.texi: Likewise.
35911 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35914 * params.def (inline-min-speedup) Change from 10 to 8.
35916 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35918 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
35921 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35924 * ipa-inline-analysis.c (get_minimal_bb): New function.
35925 (record_modified): Use it.
35926 (remap_edge_change_prob): Handle also ancestor functions.
35928 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35930 * doc/contrib.texi (Contributors): Remove broken link into
35931 the Mauve CVS repository.
35933 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35935 PR middle-end/79454
35936 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35937 result computation whenever lhs doesn't have vector mode, not
35938 just when it has BLKmode.
35940 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35942 * doc/makefile.texi (profiledbootstrap): Refer to the
35943 installation instructions only in textual form.
35945 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35948 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35950 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35952 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35953 (Specific): Update mingw-w64 reference.
35955 (Specific): Remove broken link to Renesas RX processor.
35957 2017-02-10 Richard Biener <rguenther@suse.de>
35959 * toplev.c (process_options): Do not mention obsolete graphite
35960 options when printing sorry message about missing graphite support.
35961 Mention -floop-nest-optimize.
35963 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35965 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35966 (vtst_p16): Likewise.
35967 (vtstq_p8): Likewise.
35968 (vtstq_p16): Likewise.
35970 (vtstq_p64): Likewise.
35971 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35972 (vset_lane_p64): New.
35973 (vsetq_lane_p64): New.
35975 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35977 PR tree-optimization/79411
35978 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35979 stmt operands are SSA_NAMEs used in abnormal phis.
35980 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35983 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35986 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35989 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35991 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35993 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35995 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35999 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
36000 not arbitrary TREE_CONSTANT.
36003 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
36004 "omp declare target link" attribute unless is_global_var.
36005 * omp-offload.c (find_link_var_op): Likewise.
36007 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
36008 Chung-Lin Tang <cltang@codesourcery.com>
36010 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
36012 (gimplify_adjust_omp_clauses): Don't delete TILE.
36013 (gimplify_omp_for): Deal with TILE.
36014 * internal-fn.c (expand_GOACC_TILE): New function.
36015 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
36017 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
36018 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
36020 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
36021 avoid DIV for outermost collapse var.
36022 (expand_oacc_for): Insert tile element loop as needed. Adjust.
36023 Remove out of date comments, fix whitespace.
36024 * omp-general.c (omp_extract_for_data): Deal with tiling.
36025 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
36026 adjust OLF_DIM_BASE value.
36027 (struct omp_for_data): Add tiling field.
36028 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
36029 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
36030 for auto loops. Remove default auto determining, moved to
36031 oacc_loop_fixed_partitions.
36032 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
36033 stmts, add e_mask field.
36034 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
36035 (oacc_thread_numbers): Use oacc_dim_call.
36036 (oacc_xform_tile): New.
36037 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
36038 (finish_oacc_loop): Adjust for ifns vector.
36039 (oacc_loop_discover_walk): Append loop abstraction sites to list,
36040 add case for GOACC_TILE fns.
36041 (oacc_loop_xform_loop): Delete.
36042 (oacc_loop_process): Iterate over call list directly, and add
36043 handling for GOACC_TILE fns.
36044 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
36046 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
36047 vector partitioning to outer loops. Assign 2 partitions to loops
36048 when available. Add TILE handling.
36049 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
36050 (execite_oacc_device_lower): Process GOACC_TILE fns,
36051 ignore unknown specs.
36052 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
36053 * tree.c (omp_clause_num_ops): Adjust TILE ops.
36054 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
36056 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
36058 * configure.ac (ACX_BUGURL): Update.
36059 * configure: Regenerate.
36061 2017-02-09 Richard Biener <rguenther@suse.de>
36063 PR tree-optimization/69823
36064 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
36065 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
36067 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36069 * config/arc/arc-c.def: Add __NPS400__ definition.
36070 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
36071 (TARGET_NPS400): Define.
36073 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36075 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
36077 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
36078 pointer, arch_info.
36079 (arc_cpu_types): Fill the arch_info field with a pointer into the
36080 arc_arch_types table.
36081 (arc_selected_cpu): Declare.
36082 * config/arc/arc.c (arc_selected_cpu): Make global.
36083 (arc_selected_arch): Delete.
36084 (arc_base_cpu): Delete.
36085 (arc_override_options): Remove references to deleted variables,
36086 update access to arch information.
36087 (ARC_OPT): Update access to arch information.
36088 (ARC_OPTX): Likewise.
36089 * config/arc/arc.h (arc_base_cpu): Remove declaration.
36090 (TARGET_ARC600): Update access to arch information.
36091 (TARGET_ARC601): Likewise.
36092 (TARGET_ARC700): Likewise.
36093 (TARGET_EM): Likewise.
36094 (TARGET_HS): Likewise.
36095 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
36098 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
36101 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
36102 condition/operands for integer GE/LE/GEU/LEU operations.
36104 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
36106 PR translation/79397
36107 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
36110 2017-02-08 Martin Jambor <mjambor@suse.cz>
36113 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
36114 whether allocation happened.
36115 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
36116 nothing was allocated.
36118 2017-02-08 Jakub Jelinek <jakub@redhat.com>
36120 PR tree-optimization/79408
36121 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
36122 constant, but SSA_NAME with a known integer range, use the minimum
36123 of that range instead of op1 to determine if modulo can be replaced
36124 with its first operand.
36126 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36128 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
36130 2017-02-08 Richard Biener <rguenther@suse.de>
36132 PR tree-optimization/71824
36133 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36134 Check all loops contained in the merged region.
36136 2017-02-07 Andrew Pinski <apinski@cavium.com>
36138 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
36140 2017-02-07 Andrew Pinski <apinski@cavium.com>
36142 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
36143 (thunderxt88): Likewise.
36144 (thunderxt81): Disable LSE and change v8.1 to v8.
36145 (thunderxt83): Likewise.
36147 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36148 Richard Biener <rguenther@suse.de>
36150 PR middle-end/79399
36151 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
36152 type from int to size_t.
36153 * ira-costs.c (struct_costs_size): Change type from int to size_t.
36155 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36157 PR rtl-optimization/79386
36158 * cprop.c (bypass_conditional_jumps): Initialize
36159 bypass_last_basic_block already before splitting bbs after
36160 unconditional traps...
36161 (bypass_conditional_jumps): ... rather than here.
36164 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
36165 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
36166 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
36167 fix -masm=intel patterns.
36169 2017-02-07 Richard Biener <rguenther@suse.de>
36171 PR tree-optimization/79256
36172 PR middle-end/79278
36173 * builtins.c (get_object_alignment_2): Use min_align_of_type
36174 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
36175 and ADJUST_FIELD_ALIGN.
36177 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
36179 * doc/tm.texi: Regenerate.
36180 * stor-layout.c (layout_decl): Adjust.
36181 (update_alignment_for_field): Likewise.
36182 (place_field): Likewise.
36183 (min_align_of_type): Likewise.
36184 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
36185 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
36186 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
36187 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
36188 * config/frv/frv.c (frv_adjust_field_align): Likewise.
36189 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
36190 * config/i386/i386.c (x86_field_alignment): Likewise.
36191 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
36192 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
36193 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
36194 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
36195 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
36196 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
36200 2017-01-30 Richard Biener <rguenther@suse.de>
36202 PR tree-optimization/79256
36203 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36204 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36207 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
36209 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
36210 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
36211 builtins to SImode and emit a zero-extend, if necessary.
36213 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36215 * docs/invoke.texi (RISC-V Options): Alphabetize.
36217 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36219 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
36222 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36224 * config/riscv/riscv.c: New file.
36225 * common/config/riscv/riscv-common.c: Likewise.
36226 * config.gcc: Likewise.
36227 * config/riscv/constraints.md: Likewise.
36228 * config/riscv/elf.h: Likewise.
36229 * config/riscv/generic.md: Likewise.
36230 * config/riscv/linux.h: Likewise.
36231 * config/riscv/multilib-generator: Likewise.
36232 * config/riscv/peephole.md: Likewise.
36233 * config/riscv/pic.md: Likewise.
36234 * config/riscv/predicates.md: Likewise.
36235 * config/riscv/riscv-builtins.c: Likewise.
36236 * config/riscv/riscv-c.c: Likewise.
36237 * config/riscv/riscv-ftypes.def: Likewise.
36238 * config/riscv/riscv-modes.def: Likewise.
36239 * config/riscv/riscv-opts.h: Likewise.
36240 * config/riscv/riscv-protos.h: Likewise.
36241 * config/riscv/riscv.h: Likewise.
36242 * config/riscv/riscv.md: Likewise.
36243 * config/riscv/riscv.opt: Likewise.
36244 * config/riscv/sync.md: Likewise.
36245 * config/riscv/t-elf-multilib: Likewise.
36246 * config/riscv/t-linux: Likewise.
36247 * config/riscv/t-linux-multilib: Likewise.
36248 * config/riscv/t-riscv: Likewise.
36249 * configure.ac: Likewise.
36250 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
36251 Waterman as RISC-V maintainers.
36252 * doc/install.texi: Add RISC-V entries.
36253 * doc/invoke.texi: Add RISC-V options section.
36254 * doc/md.texi: Add RISC-V constraints section.
36255 * configure: Regenerated.
36257 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
36260 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
36261 false values to be constant vectors with all 0 or all 1 bits set.
36262 (vcondu<mode><mode>): Likewise.
36263 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
36265 (fpmask_comparison_operator): Update comment.
36266 (vecint_comparison_operator): New predicate.
36267 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
36268 vector conditionals when the true and false values are constant
36269 vectors with all 0 bits or all 1 bits set.
36271 2017-02-06 Martin Sebor <msebor@redhat.com>
36273 PR tree-optimization/79376
36274 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
36276 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
36278 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
36279 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
36280 to simplify split condition.
36282 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36284 * omp-expand.c (oxpand_omp_atomic_fetch_op,
36285 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
36288 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
36290 PR rtl-optimization/68664
36291 * target.def (can_speculate_insn): New hook.
36292 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
36293 * doc/tm.texi: Regenerate.
36294 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
36295 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
36296 (rs6000_sched_can_speculate_insn): New function.
36298 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36300 PR tree-optimization/79284
36301 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
36302 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
36303 vectorizable_mask_load_store, vectorizable_operation,
36304 vect_is_simple_cond, get_same_sized_vectype): Use it instead
36305 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
36306 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
36307 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
36308 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
36309 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
36310 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
36311 is_gimple_assign (stmt). Replace another such test with
36312 is_gimple_assign (stmt).
36314 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36317 * config/avr/avr.c (rtl-iter.h): Include it.
36318 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
36319 (avr_legitimate_combined_insn): ...and implementation.
36321 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
36323 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
36324 * config/s390/s390.c (s390_const_operand_ok)
36325 (s390_canonicalize_comparison, s390_extract_part)
36326 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
36327 (s390_contiguous_bitmask_p, s390_rtx_costs)
36328 (legitimize_pic_address): Likewise.
36329 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
36330 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
36331 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
36332 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
36333 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
36335 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36337 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
36338 REGNO($0) == REGNO($1).
36340 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36342 * config/s390/linux.h(SIZE_TYPE): Add comment.
36344 2017-02-06 Julian Brown <julian@codesourcery.com>
36345 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36346 Virendra Pathak <virendra.pathak@broadcom.com>
36348 * config/aarch64/aarch64-cores.def: Change the scheduler
36350 * config/aarch64/aarch64.md: Include thunderx2t99.md.
36351 * config/aarch64/thunderx2t99.md: New file.
36353 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36355 * doc/standards.texi (Go Language): Update link to language
36358 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36360 * tree-eh.c (lower_resx): Sanitize profile.
36361 (cleanup_empty_eh_move_lp): Likewise.
36363 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36366 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
36368 * cfgloopmanip.h (loop_version): Update prototype.
36369 * modulo-sched.c (sms_schedule): Update call of loop_version.
36370 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
36371 * tree-parloops.c (gen_parallel_loop): Likewise.
36372 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
36373 * tree-ssa-loop-split.c (split_loop): Likewise.
36374 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
36375 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
36377 2017-02-05 Martin Liska <mliska@suse.cz>
36380 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
36382 (print_operand_address): Initialize a struct to zero.
36384 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36386 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
36387 garbage collector only in textual form.
36389 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36391 * doc/extend.texi (x86 specific memory model extensions for
36392 transactional memory): Simplify a phrase.
36394 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
36397 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
36398 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
36399 (atomic_storedi_1): Likewise.
36401 2017-02-04 Jakub Jelinek <jakub@redhat.com>
36403 PR tree-optimization/79338
36404 * tree-parloops.c (gather_scalar_reductions): Don't call
36405 vect_analyze_loop_form for loop->inner before destroying loop's
36408 2017-02-03 Martin Sebor <msebor@redhat.com>
36410 PR tree-optimization/79327
36411 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
36412 when precision has resulted in leading zeros.
36413 (format_integer): Adjust the likely counter to assume an unknown
36414 argument that may be zero is non-zero.
36416 2017-02-03 Jason Merrill <jason@redhat.com>
36419 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
36420 avoid copying non-taken branch.
36422 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36424 PR tree-optimization/79340
36425 * tree-vect-loop.c (vectorizable_reduction): Release
36426 vec_defs elements after safe_splicing them into other vectors.
36429 PR tree-optimization/79327
36430 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
36431 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
36433 (format_integer): Use wide_int_to_tree instead of build_int_cst
36434 + to_?hwi. If argmin is NULL, just set argmin and argmax to
36435 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
36436 of shortest and longest sequence.
36438 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
36440 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
36441 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
36443 2017-02-03 Walter Lee <walt@tilera.com>
36446 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
36447 after initial stackframe link reg save.
36448 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
36450 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36453 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
36454 wu for stxssp alternative.
36456 2017-02-03 Martin Sebor <msebor@redhat.com>
36458 PR tree-optimization/79352
36459 * gimple-fold.c (get_range_strlen): Add argument.
36460 (get_range_strlen): Change return type to bool.
36461 (get_maxval_strlen): Pass in a dummy argument.
36462 * gimple-fold.h (get_range_strlen): Change return type to bool.
36463 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
36464 * tree.h (array_at_struct_end_p): Add argument.
36465 * tree.c (array_at_struct_end_p): Handle it.
36467 2017-02-03 Martin Liska <mliska@suse.cz>
36470 * multiple_target.c (create_dispatcher_calls): Redirect edge
36471 from a caller of a dispatcher.
36472 (expand_target_clones): Make the clones local.
36473 (ipa_target_clone): Do both target clones and resolvers.
36474 (ipa_dispatcher_calls): Remove the pass.
36475 (pass_dispatcher_calls::gate): Likewise.
36476 (make_pass_dispatcher_calls): Likewise.
36477 * passes.def (pass_target_clone): Put as very first IPA early
36480 2017-02-03 Martin Liska <mliska@suse.cz>
36482 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
36483 in case of a function with ifunc attribute.
36485 2017-02-03 Martin Liska <mliska@suse.cz>
36487 * cgraph.c (cgraph_node::dump): Dump function version info.
36488 * symtab.c (symtab_node::dump_base): Add missing new line.
36490 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36492 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
36493 (ifcombine_ifandif): Use it.
36495 2017-02-03 Martin Liska <mliska@suse.cz>
36497 * doc/invoke.texi: Document default value for
36498 use-after-scope-direct-emission-threshold.
36500 2017-02-03 Martin Liska <mliska@suse.cz>
36502 PR tree-optimization/79339
36503 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
36504 (format_floating): Likewise.
36506 2017-02-03 Martin Liska <mliska@suse.cz>
36509 * ipa-prop.c (ipa_node_params_t::insert): Remove current
36511 (ipa_node_params_t::remove): Likewise.
36512 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
36513 initialization from removed ipa_node_params_t::insert.
36514 (ipa_node_params::~ipa_node_params): Move from removed
36515 ipa_node_params_t::release.
36516 * symbol-summary.h (symbol_summary::m_released): New member.
36517 Do not release a summary twice. Do not allow to call finalizer
36518 for types of a summary that live in GGC memory.
36520 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36522 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
36525 2017-02-02 Martin Sebor <msebor@redhat.com>
36527 PR middle-end/79275
36528 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
36529 (format_string): Tighten up the range of output for non-constant
36530 strings and correct the expected range for wide non-constant strings.
36532 2017-02-02 Martin Sebor <msebor@redhat.com>
36534 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
36536 PR middle-end/32003
36537 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
36539 (-fdump-tree-@var): Add to index and document how to come up
36540 with pass-specific option and dump file names.
36541 (-fdump-passes): Clarify where to look for output.
36543 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36545 PR middle-end/77445
36546 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
36547 statistics of the analyzed path; allow threading for speed when
36548 any of BBs along the path are optimized for speed.
36550 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
36552 PR middle-end/78468
36553 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
36554 settings of the virtual registers.
36557 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
36559 * explow.c (get_dynamic_stack_size): Take known alignment of stack
36560 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
36563 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36565 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
36566 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
36568 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36570 * config/s390/s390.md: Add missing comments with the expanded
36572 * config/s390/vector.md: Likewise.
36573 * config/s390/vx-builtins.md: Likewise.
36575 2017-02-02 Jakub Jelinek <jakub@redhat.com>
36578 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
36579 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
36580 conditions on a single line.
36582 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36584 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36585 __S390_VX__ to __VX__.
36587 2017-02-01 Andrew Pinski <apinski@cavium.com>
36589 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
36590 stmt_info to record_stmt_cost.
36591 (vect_get_known_peeling_cost): Pass stmt_info if known to
36593 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
36594 cpu_vector_cost field into
36595 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
36596 field into vec_int_stmt_cost and vec_fp_stmt_cost.
36597 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
36598 splitting of scalar_stmt_cost and vec_stmt_cost.
36599 (thunderx_vector_cost): Likewise.
36600 (cortexa57_vector_cost): LIkewise.
36601 (exynosm1_vector_cost): Likewise.
36602 (xgene1_vector_cost): Likewise.
36603 (thunderx2t99_vector_cost): Improve after the splitting of the two
36605 (aarch64_builtin_vectorization_cost): Update for the splitting of
36606 scalar_stmt_cost and vec_stmt_cost.
36608 2017-02-01 Torvald Riegel <triegel@redhat.com>
36609 Richard Henderson <rth@redhat.com>
36611 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
36612 conditional on existance of a fast atomic load.
36613 * optabs-query.c (can_atomic_load_p): New function.
36614 * optabs-query.h (can_atomic_load_p): Declare it.
36615 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
36616 no fast atomic load is available for the particular size of access.
36617 (expand_atomic_compare_and_swap): Likewise.
36618 (expand_atomic_load): Likewise.
36619 (expand_atomic_store): Likewise.
36620 (expand_atomic_fetch_op): Likewise.
36621 * testsuite/lib/target-supports.exp
36622 (check_effective_target_sync_int_128): Remove x86 because it provides
36623 no fast atomic load.
36624 (check_effective_target_sync_int_128_runtime): Likewise.
36626 2017-02-01 Richard Biener <rguenther@suse.de>
36628 * graphite.c: Include tree-vectorizer.h for find_loop_location.
36629 (graphite_transform_loops): Provide opt-info for optimized nests.
36630 * tree-parloop.c (parallelize_loops): Provide opt-info for
36631 parallelized loops.
36633 2017-02-01 Richard Biener <rguenther@suse.de>
36635 PR middle-end/79315
36636 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
36637 was not set before.
36639 2017-02-01 Richard Biener <rguenther@suse.de>
36641 PR tree-optimization/71824
36642 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36643 Verify the loops are valid in the merged SESE region.
36644 (scop_detection::can_represent_loop_1): Check analyzing the
36645 evolution of the number of iterations in the region succeeds.
36647 2017-01-31 Ian Lance Taylor <iant@golang.org>
36649 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
36650 REG_ARGS_SIZE note to 32-bit push insns and call insn.
36652 2017-01-31 David Malcolm <dmalcolm@redhat.com>
36654 PR preprocessor/79210
36655 * input.c (get_substring_ranges_for_loc): Replace line_width
36656 assertion with error-handling.
36658 2017-01-31 Richard Biener <rguenther@suse.de>
36660 PR tree-optimization/77318
36661 * graphite-sese-to-poly.c (extract_affine): Fix assert.
36662 (create_pw_aff_from_tree): Take loop parameter.
36663 (add_condition_to_pbb): Pass loop of the condition to
36664 create_pw_aff_from_tree.
36666 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36668 * config/s390/s390.c (s390_asan_shadow_offset): New function.
36669 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
36671 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
36675 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
36677 (convert_int_to_float128): Likewise.
36678 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
36679 (convert_int_to_float128): Likewise.
36680 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
36681 (UNSPEC_IEEE128_CONVERT): Likewise.
36682 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
36683 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
36684 Use local variables for IBM extended format.
36685 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
36686 (fix_trunc<mode>si2_fprs): Likewise.
36687 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
36688 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
36689 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
36690 to know that we can now have integers of all sizes in vector
36692 (fix<uns>_<mode>di2_hw): Likewise.
36693 (float<uns>_<mode>si2_hw): Likewise.
36694 (fix_<mode>si2_hw): Likewise.
36695 (fixuns_<mode>si2_hw): Likewise.
36696 (float<uns>_<mode>di2_hw): Likewise.
36697 (float_<mode>di2_hw): Likewise.
36698 (float_<mode>si2_hw): Likewise.
36699 (floatuns_<mode>di2_hw): Likewise.
36700 (floatuns_<mode>si2_hw): Likewise.
36701 (xscvqp<su>wz_<mode>): Delete, no longer used.
36702 (xscvqp<su>dz_<mode>): Likewise.
36703 (xscv<su>dqp_<mode>): Likewise.
36704 (ieee128_mfvsrd_64bit): Likewise.
36705 (ieee128_mfvsrd_32bit): Likewise.
36706 (ieee128_mfvsrwz): Likewise.
36707 (ieee128_mtvsrw): Likewise.
36708 (ieee128_mtvsrd_64bit): Likewise.
36709 (ieee128_mtvsrd_32bit): Likewise.
36711 2017-01-31 Martin Liska <mliska@suse.cz>
36714 * ipa-prop.c (ipa_free_all_node_params): Call release method
36715 instead of ~sumbol_summary to not to trigger double times
36718 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
36720 PR tree-optimization/71691
36721 * bitmap.h (class auto_bitmap): New.
36722 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
36723 is_maybe_undefined instead of ssa_undefined_value_p.
36725 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36727 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36728 __S390_ARCH_LEVEL__ to __ARCH__.
36730 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36732 PR tree-optimization/79267
36733 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
36734 if should_remove_lhs_p is true.
36736 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
36739 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
36740 (add_alignment_attribute): New.
36741 (base_type_die): Add alignment attribute.
36742 (subrange_type_die): Likewise.
36743 (modified_type_die): Likewise.
36744 (gen_array_type_die): Likewise.
36745 (gen_descr_array_type_die: Likewise.
36746 (gen_enumeration_type_die): Likewise.
36747 (gen_subprogram_die): Likewise.
36748 (gen_variable_die): Likewise.
36749 (gen_field_die): Likewise.
36750 (gen_ptr_to_mbr_type_die): Likewise.
36751 (gen_struct_or_union_type_die): Likewise.
36752 (gen_subroutine_type_die): Likewise.
36753 (gen_typedef_die): Likewise.
36754 (base_type_cmp): Compare alignment attribute.
36756 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36759 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
36760 (setb_unsigned) New pattern for setb with CCUNS.
36761 * config/rs6000/rs6000.c (expand_block_compare): Use a different
36762 subfc./subfe sequence to avoid overflow problems. Generate a
36763 shorter sequence with cmpld/setb for power9.
36764 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
36765 for generating subfc. instruction.
36766 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
36767 now uses this instruction.
36769 2017-01-30 Ian Lance Taylor <iant@google.com>
36772 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
36773 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
36775 2017-01-30 Martin Sebor <msebor@redhat.com>
36777 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
36778 Move constant to the right of a relational operator.
36779 (get_mpfr_format_length, format_character, format_string): Ditto.
36780 (should_warn_p, maybe_warn): Same.
36782 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
36784 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
36787 * asan.c (get_translation_unit_decl): Remove function.
36788 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
36790 2017-01-30 Martin Liska <mliska@suse.cz>
36792 PR gcov-profile/79259
36793 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
36794 -fprofile-generate.
36796 2017-01-30 Martin Liska <mliska@suse.cz>
36799 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
36800 Initialize variables with NULL value.
36802 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
36805 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
36807 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
36809 2017-01-30 Richard Biener <rguenther@suse.de>
36811 PR tree-optimization/79276
36812 * tree-vrp.c (process_assert_insertions): Properly adjust common
36813 when removing a duplicate.
36815 2017-01-30 Richard Biener <rguenther@suse.de>
36817 PR tree-optimization/79256
36818 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36819 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36821 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
36823 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
36826 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
36827 ("*r<noxa>sbg_<mode>_sll_bitmask")
36828 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
36829 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
36830 Use contiguous_bitmask_nowrap_operand.
36832 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36835 * config/rs6000/altivec.h (vec_xl): Revise #define.
36836 (vec_xst): Likewise.
36838 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
36840 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
36842 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
36844 PR rtl-optimization/79194
36845 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
36846 traps before call to bypass_conditional_jumps.
36848 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36850 PR tree-optimization/71374
36851 * lra-constraints.c (check_conflict_input_operands): New.
36852 (match_reload): Use it.
36854 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36857 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
36858 account to calculate conflict_set.
36860 2017-01-27 Bin Cheng <bin.cheng@arm.com>
36862 PR rtl-optimization/78559
36863 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
36864 other_insn in combine.
36866 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
36868 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
36869 uint16_type_node for BT_UINT16.
36871 2017-01-27 David Malcolm <dmalcolm@redhat.com>
36873 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
36874 "RTL Tests" to menu.
36875 (GIMPLE Tests): New node.
36876 (RTL Tests): New node.
36878 2017-01-27 Richard Biener <rguenther@suse.de>
36880 PR tree-optimization/79245
36881 * tree-loop-distribution.c (distribute_loop): Apply cost
36882 modeling also to detected patterns.
36884 2017-01-27 Richard Biener <rguenther@suse.de>
36886 PR tree-optimization/71433
36887 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
36888 (compare_assert_loc): New function.
36889 (process_assert_insertions): Sort and optimize assert locations
36890 to remove duplicates and push down identical assertions on
36891 edges to their destination block.
36893 2017-01-27 Richard Biener <rguenther@suse.de>
36895 PR tree-optimization/79244
36896 * tree-vrp.c (remove_range_assertions): Forcefully propagate
36897 out SSA names even if abnormal.
36899 2017-01-27 Jakub Jelinek <jakub@redhat.com>
36901 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
36902 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
36903 instead of MPFR_RNDN.
36905 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
36908 * arm.c (arm_option_override): Don't call build_target_option_node
36909 until after doing all option overrides.
36910 (arm_valid_target_attribute_tree): Likewise.
36912 2017-01-27 Martin Liska <mliska@suse.cz>
36914 * doc/invoke.texi (-fprofile-arcs): Document profiling support
36915 for {cd}tors and C++ {cd}tors.
36917 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
36919 * config/s390/s390.md ("*setmem_long_and")
36920 ("*setmem_long_and_31z"): Use zero_extend instead of and.
36922 2017-01-26 Martin Sebor <msebor@redhat.com>
36924 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
36927 2017-01-26 Martin Sebor <msebor@redhat.com>
36929 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36930 HAVE_DFmode before using XFmode or DFmode.
36931 (parse_directive): Avoid using the z length modifier to avoid
36932 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36934 PR middle-end/78703
36935 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36936 to accept adjustment as an array.
36937 (get_int_range): New function.
36938 (struct directive): Make width and prec arrays.
36939 (directive::set_width, directive::set_precision): Call get_int_range.
36940 (format_integer, format_floating): Handle width and precision ranges.
36941 (format_string, parse_directive): Same.
36943 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36946 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36947 comdat_type_p set, just clone them, but keep the children in the
36951 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36952 which have direct callers with -fvar-tracking-assignments enabled
36954 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36955 inside of type units.
36957 2017-01-26 Martin Sebor <msebor@redhat.com>
36959 PR middle-end/78703
36960 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36962 (struct format_result): Replace number_chars, number_chars_min,
36963 and number_chars_max with a single member of struct result_range.
36965 (format_result::operator+=): Adjust.
36966 (struct fmtresult): Remove bounded. Handle likely and unlikely
36968 (fmtresult::adjust_for_width_or_precision): New function.
36969 (fmtresult:type_max_digits): New function.
36970 (bytes_remaining): Handle likely and unlikely counters.
36971 (min_bytes_remaining): Remove.
36972 (format_percent): Simplify.
36973 (format_integer, format_floating): Set likely and unlikely counters.
36974 (get_string_length, format_character, format_string): Same.
36975 (format_plain, should_warn_p): New function.
36976 (maybe_warn): Call should_warn_p. Update diagnostic messages
36977 and handle those for all directives, including plain strings.
36978 (format_directive): Handle likely and unlikely counters.
36979 Remove unnecessary quoting from diagnostics. Add an informational
36981 (add_bytes): Remove.
36982 (pass_sprintf_length::compute_format_length): Simplify.
36983 (try_substitute_return_value): Handle likely and unlikely counters.
36985 2017-01-26 Carl Love <cel@us.ibm.com>
36987 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36988 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36990 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36993 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36994 endianess for subregs into account.
36995 * lra-constraints.c (lra_constraints): Do risky transformations
36996 always on the first iteration.
36997 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36998 last_call_used_reg_set.
36999 (process_bb_lives): Define and use last_call_used_reg_set.
37000 * lra.c (lra): Always continue after lra_constraints on the first
37003 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
37005 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
37007 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
37009 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37011 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
37012 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
37013 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
37014 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
37015 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
37016 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
37017 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
37018 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
37019 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
37021 2017-01-26 Marek Polacek <polacek@redhat.com>
37024 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
37025 for the third operand.
37027 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37029 PR middle-end/79236
37030 * omp-low.c (struct omp_context): Add simt_stmt field.
37031 (scan_omp_for): Return omp_context *.
37032 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
37033 context to the _simt_ SIMD stmt.
37034 (lower_omp_for): For combined SIMD with sibling _simt_
37035 SIMD, make sure to use the same decls in _looptemp_
37036 clauses as in the sibling.
37038 2017-01-26 David Sherwood <david.sherwood@arm.com>
37040 PR middle-end/79212
37041 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
37044 2017-01-26 Jakub Jelinek <jakub@redhat.com>
37047 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
37048 emit fld b; fld a; if possible.
37050 * brig-builtins.def: Update copyright years.
37051 * config/arm/arm_acle_builtins.def: Update copyright years.
37053 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
37056 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
37057 constraint instead of o for the stxsd instruction.
37059 2017-01-25 Carl Love <cel@us.ibm.com>
37061 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
37062 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
37064 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
37066 * doc/invoke.texi (C++ Dialect Options): Fix typo.
37068 2017-01-25 Richard Biener <rguenther@suse.de>
37070 PR tree-optimization/69264
37071 * target.def (vector_alignment_reachable): Improve documentation.
37072 * doc/tm.texi: Regenerate.
37073 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
37075 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
37076 earlier changes with respect to TYPE_USER_ALIGN.
37077 (vector_alignment_reachable_p): Likewise. Improve dumping.
37079 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37082 * config/arm/arm.md (xordi3): Force constant operand into a register
37085 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37087 * doc/invoke.texi (-fstore-merging): Correct default optimization
37088 levels at which it is enabled.
37089 (-O): Move -fstore-merging from list to...
37092 2017-01-25 Richard Biener <rguenther@suse.de>
37095 * omp-expand.c: Include debug.h.
37096 (expand_omp_taskreg): Make sure to generate early debug before
37097 outlining anything from a function.
37098 (expand_omp_target): Likewise.
37099 (grid_expand_target_grid_body): Likewise.
37101 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
37104 * asan.c (get_translation_unit_decl): New function.
37105 (asan_add_global): Extract modules file name from globals
37106 TRANSLATION_UNIT_DECL name.
37108 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
37111 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
37112 for long calls with APCS frame and VFP.
37114 2017-01-24 David Malcolm <dmalcolm@redhat.com>
37116 * cfg.c (original_copy_tables_initialized_p): New function.
37117 * cfg.h (original_copy_tables_initialized_p): New decl.
37118 * cfgrtl.c (relink_block_chain): Guard the call to
37119 free_original_copy_tables with a call to
37120 original_copy_tables_initialized_p.
37121 * cgraph.h (symtab_node::native_rtl_p): New decl.
37122 * cgraphunit.c (symtab_node::native_rtl_p): New function.
37123 (symtab_node::needed_p): Don't assert for early assembly output
37124 for __RTL functions.
37125 (cgraph_node::finalize_function): Set "force_output" for __RTL
37127 (cgraph_node::analyze): Bail out early for __RTL functions.
37128 (analyze_functions): Update assertion to support __RTL functions.
37129 (cgraph_node::expand): Bail out early for __RTL functions.
37130 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
37132 * function.h (struct function): Update comment for field
37134 * gimple-expr.c: Include "tree-pass.h".
37135 (gimple_has_body_p): Return false for __RTL functions.
37136 * Makefile.in (OBJS): Add run-rtl-passes.o.
37137 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
37139 (gcc::pass_manager::get_clean_slate): New accessor.
37140 * passes.c: Include "insn-addr.h".
37141 (should_skip_pass_p): Add logging. Update logic for running
37142 "expand" to be compatible with both __GIMPLE and __RTL. Guard
37143 property-provider override so it is only done for gimple passes.
37145 (skip_pass): New function.
37146 (execute_one_pass): Call skip_pass when skipping passes.
37147 * read-md.c (md_reader::read_char): Support filtering
37148 the input to a subset of line numbers.
37149 (md_reader::md_reader): Initialize fields
37150 m_first_line and m_last_line.
37151 (md_reader::read_file_fragment): New function.
37152 * read-md.h (md_reader::read_file_fragment): New decl.
37153 (md_reader::m_first_line): New field.
37154 (md_reader::m_last_line): New field.
37155 * read-rtl-function.c (function_reader::create_function): Only
37156 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
37157 curr_properties. Set DECL_INITIAL to a dummy block.
37158 (read_rtl_function_body_from_file_range): New function.
37159 * read-rtl-function.h (read_rtl_function_body_from_file_range):
37161 * run-rtl-passes.c: New file.
37162 * run-rtl-passes.h: New file.
37164 2017-01-24 Jeff Law <law@redhat.com>
37166 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
37169 2017-01-24 Bin Cheng <bin.cheng@arm.com>
37171 PR tree-optimization/79159
37172 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
37173 (record_nonwrapping_iv): Improve boundary using above function if no
37174 value range information.
37176 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
37177 Martin Jambor <mjambor@suse.cz>
37179 * brig-builtins.def: New file.
37180 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
37181 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
37182 (DEF_HSAIL_SAT_BUILTIN): Likewise.
37183 (DEF_HSAIL_INTR_BUILTIN): Likewise.
37184 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
37185 * builtin-types.def (BT_INT8): New.
37186 (BT_INT16): Likewise.
37187 (BT_UINT8): Likewise.
37188 (BT_UINT16): Likewise.
37189 (BT_FN_ULONG): Likewise.
37190 (BT_FN_UINT_INT): Likewise.
37191 (BT_FN_UINT_ULONG): Likewise.
37192 (BT_FN_UINT_LONG): Likewise.
37193 (BT_FN_UINT_PTR): Likewise.
37194 (BT_FN_ULONG_PTR): Likewise.
37195 (BT_FN_INT8_FLOAT): Likewise.
37196 (BT_FN_INT16_FLOAT): Likewise.
37197 (BT_FN_UINT32_FLOAT): Likewise.
37198 (BT_FN_UINT16_FLOAT): Likewise.
37199 (BT_FN_UINT8_FLOAT): Likewise.
37200 (BT_FN_UINT64_FLOAT): Likewise.
37201 (BT_FN_UINT16_UINT32): Likewise.
37202 (BT_FN_UINT32_UINT16): Likewise.
37203 (BT_FN_UINT16_UINT16_UINT16): Likewise.
37204 (BT_FN_INT_PTR_INT): Likewise.
37205 (BT_FN_UINT_PTR_UINT): Likewise.
37206 (BT_FN_LONG_PTR_LONG): Likewise.
37207 (BT_FN_ULONG_PTR_ULONG): Likewise.
37208 (BT_FN_VOID_UINT64_UINT64): Likewise.
37209 (BT_FN_UINT8_UINT8_UINT8): Likewise.
37210 (BT_FN_INT8_INT8_INT8): Likewise.
37211 (BT_FN_INT16_INT16_INT16): Likewise.
37212 (BT_FN_INT_INT_INT): Likewise.
37213 (BT_FN_UINT_FLOAT_UINT): Likewise.
37214 (BT_FN_FLOAT_UINT_UINT): Likewise.
37215 (BT_FN_ULONG_UINT_UINT): Likewise.
37216 (BT_FN_ULONG_UINT_PTR): Likewise.
37217 (BT_FN_ULONG_ULONG_ULONG): Likewise.
37218 (BT_FN_UINT_UINT_UINT): Likewise.
37219 (BT_FN_VOID_UINT_PTR): Likewise.
37220 (BT_FN_UINT_UINT_PTR: Likewise.
37221 (BT_FN_UINT32_UINT64_PTR): Likewise.
37222 (BT_FN_INT_INT_UINT_UINT): Likewise.
37223 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
37224 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
37225 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
37226 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
37227 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
37228 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
37229 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
37230 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
37231 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
37232 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
37233 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
37234 * doc/frontends.texi: List BRIG FE.
37235 * doc/install.texi (Testing): Add BRIG tesring requirements.
37236 * doc/invoke.texi (Overall Options): Mention BRIG.
37237 * doc/standards.texi (Standards): Doucment BRIG HSA version.
37239 2017-01-24 Richard Biener <rguenther@suse.de>
37241 PR translation/79208
37242 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
37244 2017-01-24 Martin Jambor <mjambor@suse.cz>
37247 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
37248 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
37249 and known_contexts.
37251 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
37253 PR middle-end/79123
37254 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
37255 casts from signed to unsigned really don't have a range.
37257 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
37259 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
37260 GMP_RNDx for compatiblity.
37262 2017-01-24 Martin Liska <mliska@suse.cz>
37265 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
37266 that would prevent us to call alloca with -1 as argument.
37268 2017-01-24 Jakub Jelinek <jakub@redhat.com>
37270 * dwarf2out.c (output_compilation_unit_header, output_file_names):
37271 Avoid -Wformat-security warning.
37273 2017-01-23 Andrew Pinski <apinski@cavium.com>
37275 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
37278 2017-01-23 Martin Sebor <msebor@redhat.com>
37280 PR middle-end/78703
37281 * gimple-ssa-sprintf.c (warn_level): New global.
37282 (format_integer): Use it here and throughout the rest of the file.
37283 Use the same switch to compute sign as base.
37284 (maybe_warn): New function.
37285 (format_directive): Factor out warnings into maybe_warn.
37286 Add debugging output. Use warn_level.
37287 (add_bytes): Use warn_level.
37288 (pass_sprintf_length::compute_format_length): Add debugging output.
37289 (try_substitute_return_value): Same.
37290 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
37292 PR middle-end/78703
37293 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
37294 (struct fmtresult, format_integer, format_floating): Adjust.
37295 (fmtresult::fmtresult): Set max correctly in two argument ctor.
37296 (get_string_length, format_string,format_directive): Same.
37297 (pass_sprintf_length::compute_format_length): Same.
37298 (try_substitute_return_value): Simplify slightly.
37300 PR middle-end/78703
37301 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
37302 (fmtresult::operator+=): Outlined.
37303 (struct fmtresult): Add ctors.
37304 (struct conversion_spec): Rename...
37305 (struct directive): ...to this. Add and remove data members.
37306 (directive::set_width, directive::set_precision): New functions.
37307 (format_percent): Use fmtresult ctor.
37308 (get_width_and_precision): Remove.
37309 (format_integer): Make naming changes. Avoid computing width and
37311 (format_floating): Same. Adjust indentation.
37312 (format_character, format_none): New functions.
37313 (format_string): Moved character handling to format_character.
37314 (format_directive): Remove arguments, change return type.
37315 (parse_directive): New function.
37316 (pass_sprintf_length::compute_format_length): Move directive
37317 parsing to parse_directive.
37319 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37321 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
37322 (assign_assembler_name_if_needed): ... this.
37323 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
37324 (assign_assembler_name_if_needed): ... this.
37325 (free_lang_data_in_cgraph): Adjust callers.
37326 * cgraphunit.c (cgraph_node::analyze): Likewise.
37327 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
37330 2017-01-23 Richard Biener <rguenther@suse.de>
37332 PR tree-optimization/79088
37333 PR tree-optimization/79188
37334 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
37335 resetting loop bounds after last path deletion. Reset loop
37336 bounds of the target loop, make code match the comments.
37337 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
37338 Make sure loops need no fixups.
37340 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
37342 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
37343 exponent support with double type for first argument.
37344 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
37345 type returned by __builtin_vec_extract_sig,
37346 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
37347 functions from "vector int" to "vector unsigned int" or from
37348 "vector long long int" to "vector unsigned long long int".
37349 Changed type returned by __builtin_vec_extract_exp,
37350 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
37351 functions from "vector int" to "vector unsigned int" or from
37352 "vector long long int" to "vector unsigned long long int".
37353 Changed return type of __builtin_vec_test_data_class,
37354 __builtin_vec_test_data_class_sp, and
37355 __builtin_vec_test_data_class_dp from "vector int" to
37356 "vector bool int" or from "vector long long int" to "vector bool
37357 long long int" and changed second argument type from "unsigned
37358 int" to "int". Added new overloaded function forms "vector float
37359 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
37360 "vector float __builtin_vec_insert_exp_sp (vector float, vector
37361 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
37362 double, vector unsigned long long int)" and "vector double
37363 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
37364 long int)". Changed return type of
37365 __builtin_scalar_test_data_class and
37366 __builtin_scalar_test_data_class_sp and
37367 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
37368 int" and changed second argument from "unsigned int" to "int".
37369 Changed type returned by __builtin_scalar_test_neg,
37370 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
37371 from "int" to "bool int". Added new overloaded function form
37372 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
37373 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
37374 exponent double-precision with floating point first argument.
37375 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
37376 documentation of scalar_test_data_class, scalar_test_neg,
37377 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
37378 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
37379 vec_test_data_class built-in functions to reflect refinements in
37380 their type signatures.
37382 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
37384 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
37386 (aarch64_elf_asm_destructor): Likewise.
37388 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
37390 PR rtl-optimization/78634
37391 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
37392 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
37393 * ifcvt.c (noce_try_cmove): Add missing cost check.
37395 PR rtl-optimization/71724
37396 * combine.c (if_then_else_cond): Look for situations where it is
37397 beneficial to undo the work of one of the recursive calls.
37399 2017-01-23 Bin Cheng <bin.cheng@arm.com>
37401 PR tree-optimization/70754
37402 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
37403 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
37404 combined stmt before it if not NULL.
37405 (combine_chains): Process refs reversely and compute dominance point
37408 2017-01-23 Martin Liska <mliska@suse.cz>
37410 PR tree-optimization/79196
37411 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
37412 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
37414 (strlen_optimize_stmt): Call the renamed function.
37416 2017-01-23 Michael Matz <matz@suse.de>
37418 PR tree-optimization/78384
37419 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
37421 2017-01-23 Richard Biener <rguenther@suse.de>
37423 PR tree-optimization/79186
37424 * tree-vrp.c (register_new_assert_for): Make sure we've seen
37425 both incoming edges before moving an assert.
37427 2017-01-23 Martin Jambor <mjambor@suse.cz>
37429 * ipa-prop.c (load_from_param_1): Removed.
37430 (load_from_unmodified_param): Bits from load_from_param_1 put back
37432 (load_from_param): Removed.
37433 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
37434 with stmt. Reverted back to use of load_from_unmodified_param.
37436 2017-01-23 Martin Jambor <mjambor@suse.cz>
37439 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
37440 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
37441 field a pointer to garbage collected vector, mark lattices and
37442 ipcp_orig_node with GTY((skip)).
37443 (ipa_get_param_count): Adjust to descriptors being a pointer.
37444 (ipa_get_param): Likewise.
37445 (ipa_get_type): Likewise.
37446 (ipa_get_param_move_cost): Likewise.
37447 (ipa_set_param_used): Likewise.
37448 (ipa_get_controlled_uses): Likewise.
37449 (ipa_set_controlled_uses): Likewise.
37450 (ipa_is_param_used): Likewise.
37451 (ipa_node_params_t): Move into garbage collector. New methods insert
37453 (ipa_node_params_sum): Annotate wth GTY(()).
37454 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
37456 (ipa_load_from_parm_agg): Adjust declaration.
37457 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
37458 * ipa-profile.c (ipa_profile): Likewise.
37459 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
37460 (ipa_populate_param_decls): Make descriptors parameter garbage
37462 (ipa_dump_param): Adjust to descriptors being a pointer.
37463 (ipa_alloc_node_params): Likewise.
37464 (ipa_initialize_node_params): Likewise.
37465 (load_from_param_1): Make descriptors parameter garbage collected.
37466 (load_from_unmodified_param): Likewise.
37467 (load_from_param): Likewise.
37468 (ipa_load_from_parm_agg): Likewise.
37469 (ipa_node_params::~ipa_node_params): Removed.
37470 (ipa_free_all_node_params): Remove call to delete operator.
37471 (ipa_node_params_t::insert): New.
37472 (ipa_node_params_t::remove): Likewise.
37473 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
37474 copy known_csts and known_contexts vectors.
37475 (ipa_read_node_info): Adjust to descriptors being a pointer.
37476 (ipcp_modif_dom_walker): Make m_descriptors field garbage
37478 (ipcp_transform_function): Make descriptors variable garbage
37481 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
37483 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
37484 * config/i386/avx512dqintrin.h: Ditto.
37485 * config/i386/avx512fintrin.h: Ditto.
37486 * config/i386/i386.c: Handle new builtins.
37487 * config/i386/i386-builtin.def: Add new builtins.
37488 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
37489 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
37491 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37492 Martin Liska <mliska@suse.cz>
37494 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
37495 * asan.c (asan_expand_poison_ifn): Support stores and use
37496 appropriate ASAN report function.
37497 * internal-fn.c (expand_ASAN_POISON_USE): New function.
37498 * internal-fn.def (ASAN_POISON_USE): Declare.
37499 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
37500 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
37501 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
37502 ASAN_POISON calls w/o LHS.
37503 * tree-ssa.c (execute_update_addresses_taken): Create clobber
37504 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
37505 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
37506 * gimplify.c (asan_poison_variables): Add attribute
37507 use_after_scope_memory to variables that really needs to live
37509 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
37510 having the attribute.
37512 2017-01-23 Martin Liska <mliska@suse.cz>
37514 * asan.c (create_asan_shadow_var): New function.
37515 (asan_expand_poison_ifn): Likewise.
37516 * asan.h (asan_expand_poison_ifn): New declaration.
37517 * internal-fn.c (expand_ASAN_POISON): Likewise.
37518 * internal-fn.def (ASAN_POISON): New builtin.
37519 * sanopt.c (pass_sanopt::execute): Expand
37520 asan_expand_poison_ifn.
37521 * tree-inline.c (copy_decl_for_dup_finish): Make function
37523 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
37524 * tree-ssa.c (is_asan_mark_p): New function.
37525 (execute_update_addresses_taken): Rewrite local variables
37526 (identified just by use-after-scope as addressable) into SSA.
37528 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37530 * doc/install.texi (Specific): opensource.apple.com uses https
37531 now. Remove trailing slash.
37533 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37535 * README.Portability: Remove note on an Irix compatibility issue.
37537 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
37539 * gcov.c (INCLUDE_ALGORITHM): Define.
37540 (INCLUDE_VECTOR): Define.
37541 No longer include <vector> and <algorithm> directly.
37543 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
37545 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
37547 * doc/invoke.texi (Code Gen Options): Ditto.
37549 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
37552 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
37554 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
37556 rtl-optimization/79125
37557 * cprop.c (local_cprop_pass): Handle cases where we make an
37558 unconditional trap.
37560 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
37564 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
37565 read from, for big endian.
37567 2017-01-20 Jiong Wang <jiong.wang@arm.com>
37569 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
37570 register pauth builtins for LP64 only.
37572 2017-01-20 Marek Polacek <polacek@redhat.com>
37575 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
37578 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37580 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
37582 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
37583 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
37584 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
37586 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37589 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
37590 in big-endian mode when they are not a single duplicated value.
37592 2017-01-20 Richard Biener <rguenther@suse.de>
37594 * BASE-VER: Bump to 7.0.1.
37596 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37598 * omp-low.c (omplow_simd_context): New struct. Use it...
37599 (lower_rec_simd_input_clauses): ...here and...
37600 (lower_rec_input_clauses): ...here to hold common data. Adjust all
37601 references to idx, lane, max_vf, is_simt.
37603 2017-01-20 Graham Markall <graham.markall@embecosm.com>
37605 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
37608 2017-01-20 Martin Jambor <mjambor@suse.cz>
37610 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
37611 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
37613 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
37614 (GTFILES): Rename hsa.c to hsa-common.c.
37615 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
37616 * hsa-dump.c: Likewise.
37617 * hsa-gen.c: Likewise.
37618 * hsa-regalloc.c: Likewise.
37619 * ipa-hsa.c: Likewise.
37620 * omp-expand.c: Likewise.
37621 * omp-low.c: Likewise.
37622 * toplev.c: Likewise.
37624 2017-01-20 Marek Polacek <polacek@redhat.com>
37627 * doc/invoke.texi: Document -Wduplicated-branches.
37628 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
37629 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
37630 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
37631 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
37632 return 0 only when not OEP_LEXICOGRAPHIC.
37633 (fold_build_cleanup_point_expr): Use the expression
37634 location when building CLEANUP_POINT_EXPR.
37635 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
37636 * tree.c (add_expr): Handle error_mark_node.
37638 2017-01-20 Martin Liska <mliska@suse.cz>
37641 * tree-profile.c (init_ic_make_global_vars): Do not call
37643 (gimple_init_gcov_profiler): Likewise.
37645 2017-01-20 Martin Liska <mliska@suse.cz>
37648 * cgraph.h (maybe_create_reference): Remove argument and
37650 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
37652 * ipa-cp.c (create_specialized_node): Likewise.
37653 * symtab.c (symtab_node::maybe_create_reference): Handle
37654 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
37656 2017-01-20 Martin Liska <mliska@suse.cz>
37658 * read-rtl-function.c (function_reader::create_function): Use
37659 build_decl instread of build_decl_stat.
37661 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
37663 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
37664 * config/i386/avx512dqintrin.h: Ditto.
37665 * config/i386/avx512fintrin.h: Ditto.
37666 * config/i386/i386-builtin-types.def: Add new types.
37667 * config/i386/i386.c: Handle new types.
37668 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
37669 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
37670 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
37671 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
37672 (__builtin_ia32_kshiftridi): New.
37673 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
37675 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
37679 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
37680 define to rs6000_init_stack_protect_guard.
37681 (rs6000_init_stack_protect_guard): New function.
37683 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37684 Yunqiang Su <yunqiang.su@imgtec.com>
37686 * config.gcc (supported_defaults): Add madd4.
37687 (with_madd4): Add validation.
37688 (all_defaults): Add madd4.
37689 * config/mips/mips.opt (mmadd4): New option.
37690 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37692 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37694 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
37695 (ISA_HAS_FUSED_MADD4): Likewise.
37696 * doc/invoke.texi (-mmadd4): Document the new option.
37697 * doc/install.texi (--with-madd4): Document the new option.
37699 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37701 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
37702 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
37703 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
37704 (aarch64_init_pauth_hint_builtins): New.
37705 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
37706 (aarch64_expand_builtin): Expand new builtins.
37708 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37710 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
37711 * combine-stack-adj.c (no_unhandled_cfa): Handle
37712 REG_CFA_TOGGLE_RA_MANGLE.
37713 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
37714 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
37715 info for return address signing.
37716 (aarch64_expand_epilogue): Likewise.
37718 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37720 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
37721 * config/aarch64/aarch64-protos.h
37722 (aarch64_return_address_signing_enabled): New declaration.
37723 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
37725 (aarch64_expand_prologue): Sign return address before it's pushed onto
37727 (aarch64_expand_epilogue): Authenticate return address fetched from
37729 (aarch64_override_options): Sanity check for ILP32 and ISA level.
37730 (aarch64_attributes): New function attributes for "sign-return-address".
37731 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
37732 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
37733 ("*do_return"): Generate combined instructions according to key index.
37734 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
37735 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
37737 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
37738 * config/aarch64/aarch64.opt (msign-return-address=): New.
37739 * doc/extend.texi (AArch64 Function Attributes): Documents
37740 "sign-return-address=".
37741 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
37743 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37745 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
37746 overall option summary.
37748 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37750 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
37751 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
37752 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
37753 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
37755 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
37757 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
37758 -mpower9-minmax by default for -mcpu=power9.
37759 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
37760 128-bit floating point.
37762 2017-01-20 Alan Modra <amodra@gmail.com>
37764 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
37765 optimizing for size.
37767 2017-01-20 Alan Modra <amodra@gmail.com>
37770 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
37771 for strcmp and strncmp from corresponding builtin decl.
37773 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37775 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
37776 instead of i386/rtems-64.h.
37777 * config/i386/rtems-64.h: Remove.
37779 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37783 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
37785 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
37787 2017-01-19 Tamar Christina <tamar.christina@arm.com>
37789 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
37790 Change int to HOST_WIDE_INT.
37791 * config/aarch64/aarch64-protos.h
37792 (aarch64_simd_gen_const_vector_dup): Likewise.
37793 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
37795 2017-01-19 David Malcolm <dmalcolm@redhat.com>
37797 * langhooks-def.h (lhd_type_for_size): New decl.
37798 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
37799 * langhooks.c (lhd_type_for_size): New function, taken from
37802 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
37804 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
37805 define_bypass for CR latency.
37806 (power9-cracked-alu): Update bypass latency and remove power9-branch.
37807 (power9-alu2): Add define_bypass for CR latency.
37809 (power9-mul): Update insn latency.
37810 (power9-mul-compare): Update insn latency, bypass latency and remove
37813 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37815 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
37817 * config/aarch64/aarch64.md
37818 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
37819 aarch64_nopcrelative_literal_loads.
37820 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
37822 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
37824 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
37825 TARGET_LOONGSON_3A.
37826 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
37828 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
37831 * config.gcc (supported_defaults): Add lxc1-sxc1.
37832 (with_lxc1_sxc1): Add validation.
37833 (all_defaults): Add lxc1-sxc1.
37834 * config/mips/mips.opt (mlxc1-sxc1): New option.
37835 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37837 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37838 __mips_no_lxc1_sxc1.
37839 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
37840 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
37841 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
37843 2017-01-19 Richard Biener <rguenther@suse.de>
37845 PR tree-optimization/72488
37846 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
37847 sure to restore SSA info.
37848 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
37850 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
37852 PR rtl-optimization/79121
37853 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
37854 of the inner type when shifting an extended value.
37856 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37859 * symtab.c (symtab_node::equal_address_to): Fix comparing of
37860 interposable aliases.
37862 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
37865 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
37866 Use the evmergelohi instruction.
37867 (mov_si<mode>_e500_subreg4_2_le): Likewise.
37868 (mov_sitf_e500_subreg8_2_be): Likewise.
37869 (mov_sitf_e500_subreg12_2_le): Likewise.
37870 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
37871 (mov_si<mode>_e500_subreg4_2_be): Likewise.
37872 (mov_sitf_e500_subreg8_2_le): Likewise.
37873 (mov_sitf_e500_subreg12_2_be): Likewise.
37875 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37877 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
37878 attribute from vecsimple to vecperm.
37879 (altivec_vbpermq2): Likewise.
37881 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37884 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
37886 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37887 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
37888 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
37889 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
37890 case where N arg is SIZE_MAX.
37891 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
37892 (cmpstrsi): Add pattern.
37894 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
37896 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
37897 __builtin_vec_revb builtins.
37898 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
37899 built-in functions to support generation of the ISA 3.0 XXBR<x>
37900 vector byte reverse instructions.
37901 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
37902 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
37903 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
37904 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
37905 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
37906 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
37907 (P9V_BUILTIN_VEC_REVB): Likewise.
37908 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
37909 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
37910 (p9_xxbrq_v16qi): Likewise.
37911 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
37912 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
37913 (p9_xxbrh_v8hi): Likewise.
37914 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
37915 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
37916 vec_revb built-in functions.
37918 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
37920 PR rtl-optimization/78952
37921 * config/i386/i386.md (any_extract): New code iterator.
37922 (*insvqi_2): Use any_extract for source operand.
37923 (*insvqi_3): Use any_shiftrt for source operand.
37925 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
37927 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37929 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37931 2017-01-18 Matthias Klose <doko@ubuntu.com>
37933 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37935 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37937 * config/rs6000/altivec.h (vec_bperm): Change #define.
37938 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37939 (altivec_vbpermq2): New define_insn.
37940 (altivec_vbpermd): Likewise.
37941 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37942 function interface.
37943 (VBPERMD): Likewise.
37944 (VBPERM): New polymorphic function interface.
37945 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37946 Add entries for P9V_BUILTIN_VEC_VBPERM.
37947 * doc/extend.texi: Add interfaces for vec_bperm.
37949 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37951 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37952 first letter of error messages.
37953 (s390_resolve_overloaded_builtin): Likewise.
37954 * config/s390/s390.c (s390_expand_builtin): Likewise.
37955 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37956 (s390_valid_target_attribute_inner_p): Likewise.
37957 * config/s390/s390.md ("tabort"): Likewise.
37959 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37961 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37962 (ISA_AVOID_DIV_HILO): New macro.
37963 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37964 (ISA_HAS_DDIV): Likewise.
37966 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37968 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37970 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37972 * doc/invoke.texi (fabi-version): Spelling fix.
37974 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37977 * doc/invoke.texi (fabi-version): Mention mangling fix for
37980 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37983 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37985 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37988 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37989 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37991 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37993 (stack_protect_test): Ditto.
37994 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37995 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37997 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37998 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37999 -mstack-protector-guard-offset=.
38000 (RS/6000 and PowerPC Options): Ditto.
38002 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38004 * config/i386/i386.h (MASK_CLASS_P): New define.
38005 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
38006 there are no registers from different register sets also when
38007 mask registers are used. Update function comment.
38008 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
38009 to (*k/*r) and (*k/*km) alternatives.
38011 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
38013 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
38014 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
38015 (EH_RETURN_HANDLER_RTX): New define.
38016 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
38017 Force frame pointer in EH return functions.
38018 (aarch64_expand_epilogue): Add barrier for eh_return.
38019 (aarch64_final_eh_return_addr): Remove.
38020 (aarch64_eh_return_handler_rtx): New function.
38021 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
38023 (aarch64_eh_return_handler_rtx): New prototype.
38025 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38027 * config/rs6000/altivec.h (vec_rlmi): New #define.
38028 (vec_vrlnm): Likewise.
38029 (vec_rlnm): Likewise.
38030 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
38031 (UNSPEC_VRLNM): Likewise.
38032 (VIlong): New mode iterator.
38033 (altivec_vrl<VI_char>mi): New define_insn.
38034 (altivec_vrl<VI_char>nm): Likewise.
38035 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
38037 (VRLDNM): Likewise.
38038 (RLNM): New polymorphic function entry.
38039 (VRLWMI): New monomorphic function entry.
38040 (VRLDMI): Likewise.
38041 (RLMI): New polymorphic function entry.
38042 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
38043 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
38044 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
38047 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38050 * dwarf2out.c (field_byte_offset): Restore the
38051 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
38052 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
38053 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
38056 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38059 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
38061 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38064 * dwarf2out.c (add_data_member_location_attribute): For constant
38065 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
38066 instead of DW_AT_data_member_location, DW_AT_bit_offset and
38067 DW_AT_byte_size attributes.
38069 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
38071 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
38072 after forcing to constant memory when the code model is medium.
38074 2017-01-17 Julia Koval <julia.koval@intel.com>
38077 * config/i386/avx512fintrin.h
38078 (_mm512_i32gather_ps): Change __addr type to void const*.
38079 (_mm512_mask_i32gather_ps): Ditto.
38080 (_mm512_i32gather_pd): Ditto.
38081 (_mm512_mask_i32gather_pd): Ditto.
38082 (_mm512_i64gather_ps): Ditto.
38083 (_mm512_mask_i64gather_ps): Ditto.
38084 (_mm512_i64gather_pd): Ditto.
38085 (_mm512_mask_i64gather_pd): Ditto.
38086 (_mm512_i32gather_epi32): Ditto.
38087 (_mm512_mask_i32gather_epi32): Ditto.
38088 (_mm512_i32gather_epi64): Ditto.
38089 (_mm512_mask_i32gather_epi64): Ditto.
38090 (_mm512_i64gather_epi32): Ditto.
38091 (_mm512_mask_i64gather_epi32): Ditto.
38092 (_mm512_i64gather_epi64): Ditto.
38093 (_mm512_mask_i64gather_epi64): Ditto.
38094 (_mm512_i32scatter_ps): Change __addr type to void*.
38095 (_mm512_mask_i32scatter_ps): Ditto.
38096 (_mm512_i32scatter_pd): Ditto.
38097 (_mm512_mask_i32scatter_pd): Ditto.
38098 (_mm512_i64scatter_ps): Ditto.
38099 (_mm512_mask_i64scatter_ps): Ditto.
38100 (_mm512_i64scatter_pd): Ditto.
38101 (_mm512_mask_i64scatter_pd): Ditto.
38102 (_mm512_i32scatter_epi32): Ditto.
38103 (_mm512_mask_i32scatter_epi32): Ditto.
38104 (_mm512_i32scatter_epi64): Ditto.
38105 (_mm512_mask_i32scatter_epi64): Ditto.
38106 (_mm512_i64scatter_epi32): Ditto.
38107 (_mm512_mask_i64scatter_epi32): Ditto.
38108 (_mm512_i64scatter_epi64): Ditto.
38109 (_mm512_mask_i64scatter_epi64): Ditto.
38110 * config/i386/avx512pfintrin.h
38111 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
38112 (_mm512_mask_prefetch_i32gather_ps): Ditto.
38113 (_mm512_mask_prefetch_i64gather_pd): Ditto.
38114 (_mm512_mask_prefetch_i64gather_ps): Ditto.
38115 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
38116 (_mm512_prefetch_i32scatter_ps): Ditto.
38117 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
38118 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
38119 (_mm512_prefetch_i64scatter_pd): Ditto.
38120 (_mm512_prefetch_i64scatter_ps): Ditto.
38121 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
38122 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
38123 * config/i386/avx512vlintrin.h
38124 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
38125 (_mm_mmask_i32gather_ps): Ditto.
38126 (_mm256_mmask_i32gather_pd): Ditto.
38127 (_mm_mmask_i32gather_pd): Ditto.
38128 (_mm256_mmask_i64gather_ps): Ditto.
38129 (_mm_mmask_i64gather_ps): Ditto.
38130 (_mm256_mmask_i64gather_pd): Ditto.
38131 (_mm_mmask_i64gather_pd): Ditto.
38132 (_mm256_mmask_i32gather_epi32): Ditto.
38133 (_mm_mmask_i32gather_epi32): Ditto.
38134 (_mm256_mmask_i32gather_epi64): Ditto.
38135 (_mm_mmask_i32gather_epi64): Ditto.
38136 (_mm256_mmask_i64gather_epi32): Ditto.
38137 (_mm_mmask_i64gather_epi32): Ditto.
38138 (_mm256_mmask_i64gather_epi64): Ditto.
38139 (_mm_mmask_i64gather_epi64): Ditto.
38140 (_mm256_i32scatter_ps): Change __addr type to void*.
38141 (_mm256_mask_i32scatter_ps): Ditto.
38142 (_mm_i32scatter_ps): Ditto.
38143 (_mm_mask_i32scatter_ps): Ditto.
38144 (_mm256_i32scatter_pd): Ditto.
38145 (_mm256_mask_i32scatter_pd): Ditto.
38146 (_mm_i32scatter_pd): Ditto.
38147 (_mm_mask_i32scatter_pd): Ditto.
38148 (_mm256_i64scatter_ps): Ditto.
38149 (_mm256_mask_i64scatter_ps): Ditto.
38150 (_mm_i64scatter_ps): Ditto.
38151 (_mm_mask_i64scatter_ps): Ditto.
38152 (_mm256_i64scatter_pd): Ditto.
38153 (_mm256_mask_i64scatter_pd): Ditto.
38154 (_mm_i64scatter_pd): Ditto.
38155 (_mm_mask_i64scatter_pd): Ditto.
38156 (_mm256_i32scatter_epi32): Ditto.
38157 (_mm256_mask_i32scatter_epi32): Ditto.
38158 (_mm_i32scatter_epi32): Ditto.
38159 (_mm_mask_i32scatter_epi32): Ditto.
38160 (_mm256_i32scatter_epi64): Ditto.
38161 (_mm256_mask_i32scatter_epi64): Ditto.
38162 (_mm_i32scatter_epi64): Ditto.
38163 (_mm_mask_i32scatter_epi64): Ditto.
38164 (_mm256_i64scatter_epi32): Ditto.
38165 (_mm256_mask_i64scatter_epi32): Ditto.
38166 (_mm_i64scatter_epi32): Ditto.
38167 (_mm_mask_i64scatter_epi32): Ditto.
38168 (_mm256_i64scatter_epi64): Ditto.
38169 (_mm256_mask_i64scatter_epi64): Ditto.
38170 (_mm_i64scatter_epi64): Ditto.
38171 (_mm_mask_i64scatter_epi64): Ditto.
38172 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
38173 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
38174 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
38175 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
38176 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
38177 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
38178 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
38179 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
38180 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
38181 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
38182 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
38183 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
38184 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
38185 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
38186 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
38187 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
38188 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
38189 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
38190 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
38191 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
38192 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
38193 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
38194 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
38195 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
38196 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
38197 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
38198 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
38199 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
38200 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
38201 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
38202 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
38203 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
38204 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
38205 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
38206 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
38207 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
38208 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
38209 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
38210 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
38211 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
38212 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
38213 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
38214 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
38215 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
38216 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
38217 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
38218 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
38219 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
38220 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
38221 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
38222 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
38223 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
38224 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
38225 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
38226 definitions accordingly.
38228 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
38229 Kuan-Lin Chen <kuanlinchentw@gmail.com>
38232 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
38235 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
38238 * ira-conflicts.c (ira_build_conflicts): Update total conflict
38239 hard regs for inner regno.
38241 2017-01-17 Martin Liska <mliska@suse.cz>
38244 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
38245 assumption and add comment.
38247 2017-01-17 Nathan Sidwell <nathan@acm.org>
38249 * ipa-visibility.c (localize_node): New function, broken out of ...
38250 (function_and_variable_visibility): ... here. Call it.
38252 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38254 PR middle-end/77445
38255 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
38256 correctly set frequency of oudgoing edge.
38257 (duplicate_thread_path): Fix profile updating.
38259 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38262 * configure.ac: Add GCC_BASE_VER.
38263 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
38264 version from BASE-VER file.
38265 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
38266 (gcc.o): Depend on $(BASEVER).
38267 * common.opt (dumpfullversion): New option.
38268 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
38269 * doc/invoke.texi: Document -dumpfullversion.
38270 * doc/install.texi: Document --with-gcc-major-version-only.
38271 * configure: Regenerated.
38273 2017-01-17 Richard Biener <rguenther@suse.de>
38275 PR tree-optimization/71433
38276 * tree-vrp.c (register_new_assert_for): Merge same asserts
38277 on all incoming edges.
38278 (process_assert_insertions_for): Handle insertions at the
38281 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
38283 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
38284 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
38286 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
38289 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
38292 2017-01-17 Alan Modra <amodra@gmail.com>
38295 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
38296 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
38297 symbolic stack limit when pic.
38299 2017-01-16 Martin Sebor <msebor@redhat.com>
38301 PR tree-optimization/78608
38302 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
38304 2017-01-16 Jeff Law <law@redhat.com>
38307 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
38308 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
38309 for several include directories that may be relative to sysroot.
38310 * config/i386/x-mingw32 (gplus_includedir): Define.
38311 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
38312 (native_system_includedir): Likewise.
38313 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
38314 override if TARGET_SYSTEM_ROOT is defined.
38315 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
38317 PR tree-optimization/79090
38318 PR tree-optimization/33562
38319 PR tree-optimization/61912
38320 PR tree-optimization/77485
38321 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
38322 and computed trims into the dump file.
38324 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38326 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
38328 2017-01-16 Jakub Jelinek <jakub@redhat.com>
38331 * gimplify.c (gimplify_init_constructor): If want_value and
38332 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
38336 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
38337 sequence. Formatting fixes.
38338 (doloop_optimize): Formatting fixes.
38341 * gcc.c (debug_level_greater_than_spec_func): New function.
38342 (static_spec_functions): Add debug-level-gt spec function.
38343 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
38345 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38346 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38347 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
38348 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
38349 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
38350 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
38352 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
38354 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
38355 QImode fixups to general and mask registers only.
38357 2017-01-16 Carl Love <cel@us.ibm.com>
38359 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
38360 for built-in functions
38361 vector signed char vec_nabs (vector signed char)
38362 vector signed short vec_nabs (vector signed short)
38363 vector signed int vec_nabs (vector signed int)
38364 vector signed long long vec_nabs (vector signed long long)
38365 vector float vec_nabs (vector float)
38366 vector double vec_nabs (vector double)
38367 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
38369 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
38370 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
38371 * doc/extend.texi: Update the documentation file for the new built-in
38374 2017-01-16 Martin Sebor <msebor@redhat.com>
38376 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
38379 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38381 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
38382 UNSPEC_VSX__XXSPLTD to require special splat handling.
38384 2017-01-16 David Malcolm <dmalcolm@redhat.com>
38387 * system.h: Poison strndup.
38389 2017-01-16 Alan Modra <amodra@gmail.com>
38392 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
38395 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
38397 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
38399 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
38401 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
38402 call recog here. Assert that INSN_CODE (insn) is non-negative.
38404 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38407 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
38409 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
38410 in the currently scheduled RTL fragment.
38412 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38414 PR rtl-optimization/78751
38415 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
38418 2017-01-14 Jeff Law <law@redhat.com>
38420 PR tree-optimization/79090
38421 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
38422 variable length stores.
38423 (compute_trims): Delete dead assignment to *trim_tail.
38424 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
38427 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
38429 PR rtl-optimization/78626
38430 PR rtl-optimization/78727
38431 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
38432 of a block, and split such blocks after everything else is finished.
38434 2017-01-14 Alan Modra <amodra@gmail.com>
38437 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
38438 target legitimate_combined_insn.
38439 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
38440 (rs6000_legitimate_combined_insn): New function.
38441 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
38443 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
38444 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
38445 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
38447 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
38449 * doc/frontends.texi (G++ and GCC): Remove references to Java.
38451 2017-01-13 Jeff Law <law@redhat.com>
38453 PR tree-optimization/33562
38454 PR tree-optimization/61912
38455 PR tree-optimization/77485
38456 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
38458 (delete_dead_assignment): Likewise.
38459 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
38460 statement to delete_dead_call and delete_dead_assignment.
38462 2017-01-13 David Malcolm <dmalcolm@redhat.com>
38465 * substring-locations.c (format_warning_va): Strengthen case 1 so
38466 that both endpoints of the substring must be within the format
38467 range for just the substring to be printed.
38469 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
38471 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
38472 * config/i386/i386.c (ix86_target_string): Add missing options
38473 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
38474 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
38475 flags_other and ix86_target_other to flags2_other. Display unknown
38477 (ix86_valid_target_attribute_inner_p): Add missing options and
38478 reorder options by implied ISAs, as in ix86_target_string.
38480 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38482 * hash-table.h (hash_table::too_empty_p): New function.
38483 (hash_table::expand): Use it.
38484 (hash_table::traverse): Likewise.
38485 (hash_table::empty_slot): Use sizeof (value_type) instead of
38486 sizeof (PTR) to convert bytes to elements. Shrink the table
38487 if the current size is excessive for the current number of
38490 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38492 * ira-costs.c (record_reg_classes): Break from the inner loop
38493 early once alt_fail is known to be true. Update outer loop
38494 handling accordingly.
38496 2017-01-13 Jeff Law <law@redhat.com>
38498 * tree-ssa-dse.c (decrement_count): New function.
38499 (increment_start_addr, maybe_trim_memstar_call): Likewise.
38500 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
38501 when we know the partially dead statement is a mem* function.
38503 PR tree-optimization/61912
38504 PR tree-optimization/77485
38505 * tree-ssa-dse.c: Include expr.h.
38506 (maybe_trim_constructor_store): New function.
38507 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
38509 PR tree-optimization/33562
38510 PR tree-optimization/61912
38511 PR tree-optimization/77485
38512 * doc/invoke.texi: Document new dse-max-object-size param.
38513 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
38514 * tree-ssa-dse.c: Include params.h.
38515 (dse_store_status): New enum.
38516 (initialize_ao_ref_for_dse): New, partially extracted from
38518 (valid_ao_ref_for_dse, normalize_ref): New.
38519 (setup_live_bytes_from_ref, compute_trims): Likewise.
38520 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
38521 (maybe_trim_partially_dead_store): Likewise.
38522 (maybe_trim_complex_store): Likewise.
38523 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
38524 Track what bytes live from the original store. Return tri-state
38525 for dead, partially dead or live.
38526 (dse_dom_walker): Add constructor, destructor and new private members.
38527 (delete_dead_call, delete_dead_assignment): New extracted from
38529 (dse_optimize_stmt): Make a member of dse_dom_walker.
38530 Use initialize_ao_ref_for_dse.
38532 PR tree-optimization/33562
38533 PR tree-optimization/61912
38534 PR tree-optimization/77485
38535 * sbitmap.h (bitmap_count_bits): Prototype.
38536 (bitmap_clear_range, bitmap_set_range): Likewise.
38537 * sbitmap.c (bitmap_clear_range): New function.
38538 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
38540 2017-01-13 Martin Liska <mliska@suse.cz>
38543 * function.c (set_cfun): Add new argument force.
38544 * function.h (set_cfun): Likewise.
38545 * ipa-inline-transform.c (inline_call): Use the function when
38546 strict alising from is dropped for function we inline to.
38548 2017-01-13 Richard Biener <rguenther@suse.de>
38550 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
38551 for dumping GIMPLE INTEGER_CSTs.
38553 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
38555 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
38556 to 201112L since C++17.
38558 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
38561 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
38562 if -fsanitize=kernel-address is present.
38564 2017-01-13 Richard Biener <rguenther@suse.de>
38566 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
38567 as _Literal ( type ) number in case usual suffixes do not
38568 preserve all information.
38570 2017-01-13 Richard Biener <rguenther@suse.de>
38572 PR tree-optimization/77283
38573 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
38574 and ssa-iterators.h.
38575 (is_feasible_trace): Implement a cost model based on joiner
38578 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
38581 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
38582 char or short to __float128/_Float128 directly.
38584 2017-01-12 Martin Sebor <msebor@redhat.com>
38586 to -Wformat-overflow.
38587 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
38588 (min_bytes_remaining): Same.
38589 (get_string_length): Same.
38590 (format_string): Same.
38591 (format_directive): Same.
38593 (pass_sprintf_length::handle_gimple_call): Same.
38595 2017-01-12 Jakub Jelinek <jakub@redhat.com>
38597 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
38598 info.nowrite calls with no lhs that can't throw. Return bool
38599 whether gsi_remove has been called or not.
38600 (pass_sprintf_length::handle_gimple_call): Return bool whether
38601 try_substitute_return_value called gsi_remove. Formatting fix.
38602 (pass_sprintf_length::execute): Don't use gsi_remove if
38603 handle_gimple_call returned true.
38606 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
38607 be removed due to side-effects, don't remove following barrier nor
38608 turn the successor edge into fallthru edge.
38610 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38613 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
38614 element-reversing loads and stores as not swappable.
38616 2017-01-12 Nathan Sidwell <nathan@acm.org>
38617 Nicolai Stange <nicstange@gmail.com>
38619 * combine.c (try_combine): Don't ignore result of overlap checking
38620 loop. Combine overlap & asm check into single loop.
38622 2017-01-12 Richard Biener <rguenther@suse.de>
38624 * tree-pretty-print.c (dump_generic_node): Provide -gimple
38625 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
38627 2017-01-12 Richard Biener <rguenther@suse.de>
38629 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
38630 and TS_TARGET_OPTION directly derive from TS_BASE.
38631 * tree-core.h (tree_optimization_option): Derive from tree_base.
38632 (tree_target_option): Likewise.
38634 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
38636 * config/i386/i386.c (memory_address_length): Increase len
38637 only when rip_relative_addr_p returns false.
38639 2017-01-11 Julia Koval <julia.koval@intel.com>
38641 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
38642 (OPTION_MASK_ISA_SGX_SET): New.
38643 (ix86_handle_option): Handle OPT_msgx.
38644 * config.gcc: Added sgxintrin.h.
38645 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
38646 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
38647 * config/i386/i386.c (ix86_target_string): Add -msgx.
38649 (ix86_option_override_internal): Handle new options.
38650 (ix86_valid_target_attribute_inner_p): Add sgx.
38651 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
38652 * config/i386/i386.opt: Add msgx.
38653 * config/i386/sgxintrin.h: New file.
38654 * config/i386/x86intrin.h: Add sgxintrin.h.
38656 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38659 * fold-const.c (maybe_nonzero_address): Return 1 for function
38661 (tree_single_nonzero_warnv_p): Don't handle function local objects
38665 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
38668 2017-01-11 David Malcolm <dmalcolm@redhat.com>
38671 * opts.c: Include "spellcheck.h"
38672 (struct string_fragment): New struct.
38673 (struct edit_distance_traits<const string_fragment &>): New
38675 (get_closest_sanitizer_option): New function.
38676 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
38678 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38680 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
38682 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
38683 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
38684 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
38685 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
38686 for initial die_offset if dwarf_split_debug_info.
38687 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
38688 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
38690 (output_skeleton_debug_sections): Formatting fix. Use
38691 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
38692 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
38694 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
38696 * config/arm/cortex-a53.md: Add bypasses for
38697 cortex_a53_r2f_cvt.
38698 (cortex_a53_r2f): Only use for transfers.
38699 (cortex_a53_f2r): Likewise.
38700 (cortex_a53_r2f_cvt): Add reservation for conversions.
38701 (cortex_a53_f2r_cvt): Likewise.
38703 2017-01-11 Tamar Christina <tamar.christina@arm.com>
38705 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
38706 to all inlined functions, change static to extern.
38708 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
38711 * config/arm/arm.c (legitimize_pic_address): Handle reference to
38713 (arm_assemble_integer): Likewise.
38715 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
38717 * config.gcc: Use new awk script to check CPU, FPU and architecture
38718 parameters for --with-... options.
38719 * config/arm/parsecpu.awk: New file
38720 * config/arm/arm-cpus.in: New file.
38721 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
38723 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
38725 * config/arm/t-arm: Update dependency rules.
38726 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
38727 of processing .def files.
38728 * config/arm/genopt.sh: Deleted.
38729 * config/arm/gentune.sh: Deleted.
38730 * config/arm/arm-cores.def: Deleted.
38731 * config/arm/arm-arches.def: Deleted.
38732 * config/arm/arm-fpus.def: Deleted.
38733 * config/arm/arm-tune.md: Regenerated.
38734 * config/arm/arm-tables.opt: Regenerated.
38735 * config/arm/arm-cpu.h: New generated file.
38736 * config/arm/arm-cpu-data.h: New generated file.
38737 * config/arm/arm-cpu-cdata.h: New generated file.
38739 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
38742 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
38744 (input_varpool_node): Unpack dynamically_initialized bit.
38746 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
38748 PR rtl-optimization/79032
38749 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
38750 the alignment of the adjusted memory reference against that of MODE,
38751 instead of the alignment of the original memory reference.
38753 2017-01-11 Martin Jambor <mjambor@suse.cz>
38755 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
38757 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
38758 decorated functions.
38760 2017-01-11 Richard Biener <rguenther@suse.de>
38762 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
38763 set range/nonnull info for PHI results. Do not set it on
38764 stmts marked for removal.
38766 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
38768 * expr.c (store_field): In the bitfield case, fetch the return value
38769 from the registers before applying a single big-endian adjustment.
38770 Always do a final load for a BLKmode value not larger than a word.
38772 2017-01-10 David Malcolm <dmalcolm@redhat.com>
38775 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38776 that we correctly handle column numbers greater than
38777 LINE_MAP_MAX_COLUMN_NUMBER.
38779 2017-01-10 Martin Sebor <msebor@redhat.com>
38781 PR middle-end/78245
38782 * gimple-ssa-sprintf.c (get_destination_size): Call
38783 {init,fini}object_sizes.
38784 * tree-object-size.c (addr_object_size): Adjust.
38785 (pass_through_call): Adjust.
38786 (pass_object_sizes::execute): Adjust.
38787 * tree-object-size.h (fini_object_sizes): Declare.
38789 2017-01-10 Martin Sebor <msebor@redhat.com>
38791 PR tree-optimization/78775
38792 * builtins.c (get_size_range): Move...
38793 * calls.c: ...to here.
38794 (alloc_max_size): Accept zero argument.
38795 (operand_signed_p): Remove.
38796 (maybe_warn_alloc_args_overflow): Call get_size_range.
38797 * calls.h (get_size_range): Declare.
38799 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
38801 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
38802 from TI's devices.csv file as of September 2016.
38803 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
38805 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
38807 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
38808 * doc/invoke.texi: Likewise.
38809 * doc/md.texi: Likewise.
38810 * doc/objc.texi: Likewise.
38812 2017-01-10 Joshua Conner <joshconner@google.com>
38814 * config/arm/fuchsia-elf.h: New file.
38815 * config/fuchsia.h: New file.
38816 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
38817 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
38819 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
38821 2016-01-10 Richard Biener <rguenther@suse.de>
38823 PR tree-optimization/79034
38824 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
38825 Propagate out degenerate PHIs in the joiner.
38827 2017-01-10 Martin Liska <mliska@suse.cz>
38829 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
38830 (sort_congruence_classes_by_decl_uid): Likewise.
38831 (sort_congruence_class_groups_by_decl_uid): Likewise.
38832 (sem_item_optimizer::merge_classes): Sort class, groups in these
38833 classes and members in the groups by DECL_UID of declarations.
38834 This would make merge operations stable.
38836 2017-01-10 Martin Liska <mliska@suse.cz>
38838 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
38839 usage of m_classes_vec.
38840 (sem_item_optimizer::~sem_item_optimizer): Likewise.
38841 (sem_item_optimizer::get_group_by_hash): Likewise.
38842 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
38843 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
38844 (sem_item_optimizer::verify_classes): Likewise.
38845 (sem_item_optimizer::process_cong_reduction): Likewise.
38846 (sem_item_optimizer::dump_cong_classes): Likewise.
38847 (sem_item_optimizer::merge_classes): Likewise.
38848 * ipa-icf.h (congruence_class_hash): Rename from
38849 congruence_class_group_hash. Remove declaration of m_classes_vec.
38851 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
38853 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
38854 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
38855 * config.gcc: Add avx512vpopcntdqintrin.h.
38856 * config/i386/avx512vpopcntdqintrin.h: New.
38857 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
38858 * config/i386/i386-builtin-types.def: Add new types.
38859 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
38860 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
38861 __builtin_ia32_vpopcountq_v8di_mask): New.
38862 * config/i386/i386-c.c (ix86_target_macros_internal): Define
38863 __AVX512VPOPCNTDQ__.
38864 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
38865 (PTA_AVX512VPOPCNTDQ): Define.
38866 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
38867 TARGET_AVX512VPOPCNTDQ_P): Define.
38868 * config/i386/i386.opt: Add mavx512vpopcntdq.
38869 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
38870 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
38872 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38874 PR middle-end/77484
38875 * predict.def (PRED_CALL): Set to 67.
38877 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
38879 * expr.c (store_field): In the bitfield case, if the value comes from
38880 a function call and is of an aggregate type returned in registers, do
38881 not modify the field mode; extract the value in all cases if the mode
38882 is BLKmode and the size is not larger than a word.
38884 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
38887 * config/i386/cpuid.h: Fix undefined behavior.
38889 2017-01-04 Jeff Law <law@redhat.com>
38891 PR tree-optimization/79007
38892 PR tree-optimization/67955
38893 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
38894 conservative for pt.null when flag_non_call_exceptions is on.
38896 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38898 PR translation/79019
38899 PR translation/79020
38900 * params.def (PARAM_INLINE_MIN_SPEEDUP,
38901 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
38902 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
38904 * config/avr/avr.opt (maccumulate-args): Likewise.
38905 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
38906 * common.opt (freport-bug): Likewise.
38907 * cif-code.def (CIF_FINAL_ERROR): Likewise.
38908 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
38909 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
38910 translatable string.
38911 * config/i386/i386.c (function_value_32): Likewise.
38912 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
38913 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
38915 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
38916 * common/config/msp430/msp430-common.c (msp430_handle_option):
38918 * symtab.c (symtab_node::verify_base): Likewise.
38919 * opts.c (set_debug_level): Likewise.
38920 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
38921 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
38922 missing whitespace to translatable strings.
38923 * config/avr/avr.md (bswapsi2): Fix typo in comment.
38924 * config/sh/superh.h: Likewise.
38925 * config/i386/xopintrin.h: Likewise.
38926 * config/i386/znver1.md: Likewise.
38927 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38928 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38929 * double-int.h (struct double_int): Likewise.
38930 * double-int.c (div_and_round_double): Likewise.
38931 * wide-int.cc: Likewise.
38932 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38933 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38934 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38935 (crossjumps_occurred): ... this.
38936 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38939 PR tree-optimization/78899
38940 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38941 returning bool return struct loop *, NULL for failure and the new
38943 (versionable_outer_loop_p): Don't version outer loop if it has
38944 dont_vectorized bit set.
38945 (tree_if_conversion): When versioning outer loop, ensure
38946 tree_if_conversion is performed also on the inner loop of the
38947 non-vectorizable outer loop copy.
38948 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38949 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38950 prevent vectorization of it.
38951 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38952 the outer loop vectorization of the non-scalar version is attempted
38953 before vectorization of the inner loop in scalar version. If
38954 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38955 vectorization of its inner loop.
38956 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38957 has 2 inner loops, rename also on edges from bb whose single pred
38958 is outer_loop->header. Fix typo in function comment.
38960 2017-01-09 Martin Sebor <msebor@redhat.com>
38963 * asan.c (asan_emit_stack_protection): Increase local buffer size
38964 to avoid snprintf truncation warning.
38966 2017-01-09 Andrew Pinski <apinski@cavium.com>
38968 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38969 to reference thunderx2t99 for the tuning structure
38970 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38972 (thunderx2t99_extra_costs): This.
38973 * config/aarch64/aarch64-tune.md: Regenerate.
38974 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38975 (vulcan_addrcost_table): This.
38976 (vulcan_regmove_cost): Rename to ...
38977 (thunderx2t99_regmove_cost): This.
38978 (vulcan_vector_cost): Rename to ...
38979 (thunderx2t99_vector_cost): this.
38980 (vulcan_branch_cost): Rename to ...
38981 (thunderx2t99_branch_cost): This.
38982 (vulcan_tunings): Rename to ...
38983 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38984 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38986 2017-01-09 Martin Jambor <mjambor@suse.cz>
38990 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38991 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38992 (propagate_vr_accross_jump_function): Use the above function for all
38993 value range computations for pass-through jump functions and type
38994 converasion from explicit value range values.
38995 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38996 parameters from TYPE_ARG_TYPES.
38997 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38998 (ipa_write_node_info): Stream type of the actual argument.
38999 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
39001 2017-01-09 Martin Liska <mliska@suse.cz>
39004 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
39005 (lookup_compiler): Do not show error message with have_E.
39007 2017-01-09 Jakub Jelinek <jakub@redhat.com>
39009 PR tree-optimization/78938
39010 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
39011 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
39012 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
39013 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
39016 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39018 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
39021 2017-01-09 Richard Biener <rguenther@suse.de>
39023 PR tree-optimization/78997
39024 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
39025 name condition properly.
39027 2017-01-09 Richard Biener <rguenther@suse.de>
39030 * dwarf2out.c (is_cxx): New overload with context.
39031 (is_naming_typedef_decl): Use it.
39033 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39035 * invoke.texi (Option Summary): Correct spacing in option lists
39036 and add line breaks to fix over-long lines.
39038 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39040 PR middle-end/17660
39042 * extend.texi (Common Variable Attributes): Add xref to GCC
39043 Internals manual to explain mode attribute keywords.
39045 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
39048 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
39049 and Preprocessor Options.
39050 (Options for Linking): Document -pthread here....
39051 (RS/6000 and PowerPC Options): ...not here.
39052 (Solaris 2 Options): ...or here.
39053 * doc/cppopts.texi: Document -pthread.
39055 2017-01-08 Martin Sebor <msebor@redhat.com>
39057 PR middle-end/77708
39058 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
39059 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
39060 New member functions.
39061 (format_directive): Used them.
39063 (pass_sprintf_length::handle_gimple_call): Same.
39064 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
39065 to avoid truncation for any argument.
39066 (extract_affine_mul): Same.
39067 * tree.c (get_file_function_name): Same.
39069 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39071 PR middle-end/77484
39072 * predict.def (PRED_INDIR_CALL): Set to 86.
39074 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39076 PR preprocessor/54124
39077 * doc/cppopts.texi: Reformat -d subtable to list the full name
39078 of the options. Add cross-reference to the docs for the general
39079 compiler -d options.
39080 * doc/invoke.texi (Developer Options): Add cross-reference to the
39081 preprocessor-specific -d option documentation.
39083 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39085 PR preprocessor/13498
39086 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
39087 redudant material, and reflect new command-line options.
39088 (System Headers): Likewise.
39090 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39092 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
39093 -isystem, and -idirafter. Copy-edit.
39094 * doc/cppopts.texi: Copy-edit. Remove contradiction about
39095 default for -ftrack-macro-expansion. Delete obsolete and
39096 badly-formatted implementation details about -fdebug-cpp output.
39097 * doc/cppwarnopts.texi: Copy-edit.
39099 2017-01-07 David Malcolm <dmalcolm@redhat.com>
39102 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39103 that the transition from a max line width >= 1<<10 to narrower
39104 lines works correctly.
39106 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
39108 * doc/options.texi (PerFunction): New.
39109 * opt-functions.awk (switch_flags): Map both Optimization and
39110 PerFunction to CL_OPTIMIZATION.
39111 * opth-gen.awk: Test for PerFunction flag along with
39113 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
39114 it only when the latter is present. Skip those that don't in
39115 the hash function generator.
39116 * common.opt (fvar-tracking): Mark as PerFunction instead of
39118 (fvar-tracking-assignments): Likewise.
39119 (fvar-tracking-assignments-toggle): Likewise.
39120 (fvar-tracking-uninit): Likewise.
39122 2017-01-07 Jakub Jelinek <jakub@redhat.com>
39124 PR translation/79018
39125 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
39128 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
39131 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
39132 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
39133 TARGET_LONG_JUMP_TABLE_OFFSETS.
39134 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
39135 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
39136 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
39137 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
39138 * config/m68k/m68k.md (tablejump expander): Likewise.
39139 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
39140 TARGET_LONG_JUMP_TABLE_OFFSETS.
39141 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
39142 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
39144 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
39145 David Holsgrove <david.holsgrove@xilinx.com>
39147 * common/config/microblaze/microblaze-common.c
39148 (TARGET_EXCEPT_UNWIND_INFO): Remove.
39149 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
39151 * config/microblaze/microblaze.c (microblaze_must_save_register)
39152 (microblaze_expand_epilogue, microblaze_return_addr): Handle
39154 (microblaze_eh_return): New function.
39155 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
39156 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
39157 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
39158 * config/microblaze/microblaze.md (eh_return): New pattern.
39160 2017-01-06 Jakub Jelinek <jakub@redhat.com>
39162 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
39163 GCC_DIAGNOSTIC_STRINGIFY): Define.
39165 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
39167 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39169 * config/arm/arm.md (<mcrr>): New.
39171 * config/arm/arm.c (arm_arch5te): New.
39172 (arm_option_override): Set arm_arch5te.
39173 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
39175 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
39176 (arm_mcrr_qualifiers): ... this. New.
39177 (MRRC_QUALIFIERS): Define to...
39178 (arm_mrrc_qualifiers): ... this. New.
39179 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
39181 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
39182 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
39183 (MRRCI, mrrc, MRRC): New.
39184 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
39185 VUNSPEC_MRRC2): New.
39187 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39189 * config/arm/arm.md (<mcr>): New.
39191 * config/arm/arm.c (arm_coproc_builtin_available): Add
39192 support for mcr, mrc, mcr2 and mrc2.
39193 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
39194 (arm_mcr_qualifiers): ... this. New.
39195 (MRC_QUALIFIERS): Define to ...
39196 (arm_mrc_qualifiers): ... this. New.
39197 (MCR_QUALIFIERS): Define to ...
39198 (arm_mcr_qualifiers): ... this. New.
39199 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
39201 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
39202 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
39203 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
39204 VUNSPEC_MRC2): New.
39206 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39208 * config/arm/arm.md (*ldc): New.
39212 * config/arm/arm.c (arm_coproc_builtin_available): Add
39213 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
39214 (arm_coproc_ldc_stc_legitimate_address): New.
39215 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
39216 'qualifier_const_pointer'.
39217 (LDC_QUALIFIERS): Define to...
39218 (arm_ldc_qualifiers): ... this. New.
39219 (STC_QUALIFIERS): Define to...
39220 (arm_stc_qualifiers): ... this. New.
39221 * config/arm/arm-protos.h
39222 (arm_coproc_ldc_stc_legitimate_address): New.
39223 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
39224 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
39225 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
39226 stc2, stcl, stc2l): New.
39227 * config/arm/constraints.md (Uz): New.
39228 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
39229 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
39230 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
39231 VUNSPEC_STC2L): New.
39233 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39235 * config/arm/arm.md (<cdp>): New.
39236 * config/arm/arm.c (neon_const_bounds): Rename this ...
39237 (arm_const_bounds): ... this.
39238 (arm_coproc_builtin_available): New.
39239 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
39240 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
39241 (CDP_QUALIFIERS): Define to...
39242 (arm_cdp_qualifiers): ... this. New.
39244 (arm_expand_builtin_args): Add case for 6 arguments.
39245 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
39246 (arm_const_bounds): ... this.
39247 (arm_coproc_builtin_available): New.
39248 * config/arm/arm_acle.h (__arm_cdp): New.
39250 * config/arm/arm_acle_builtins.def (cdp): New.
39252 * config/arm/iterators.md (CDPI,CDP,cdp): New.
39253 * config/arm/neon.md: Rename all 'neon_const_bounds' to
39254 'arm_const_bounds'.
39255 * config/arm/types.md (coproc): New.
39256 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
39257 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
39258 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
39259 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
39261 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39263 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
39264 (UBINOP_QUALIFIERS): New.
39266 (acle_builtin_data): New. Change comment.
39267 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
39268 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
39269 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
39270 arm_acle_builtins.def.
39271 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
39272 (arm_init_acle_builtins): New.
39273 (CRC32_BUILTIN): Remove.
39274 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
39275 crc32cb, crc32ch and crc32cw.
39276 (arm_init_crc32_builtins): Remove.
39277 (arm_init_builtins): Use arm_init_acle_builtins rather
39278 than arm_init_crc32_builtins.
39279 (arm_expand_acle_builtin): New.
39280 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
39281 * config/arm/arm_acle_builtins.def: New.
39283 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39285 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
39286 (arm_builtin_datum): ... this.
39287 (arm_init_neon_builtin): Rename to ...
39288 (arm_init_builtin): ... this. Add a new parameters PREFIX
39289 and USE_SIG_IN_NAME.
39290 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
39291 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
39292 'arm_builtin_datum'.
39293 (arm_init_vfp_builtins): Likewise.
39294 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
39295 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
39296 (arm_expand_neon_args): Rename to ...
39297 (arm_expand_builtin_args): ... this. Rename builtin_arg
39298 enum values and differentiate between ARG_BUILTIN_MEMORY
39299 and ARG_BUILTIN_NEON_MEMORY.
39300 (arm_expand_neon_builtin_1): Rename to ...
39301 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
39302 values, arm_expand_builtin_args and add bool parameter NEON.
39303 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
39304 (arm_expand_vfp_builtin): Likewise.
39305 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
39307 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39309 PR middle-end/77484
39310 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
39311 * predict.c (tree_estimate_probability_bb): Reverse direction of
39312 polymorphic call predictor.
39314 2017-01-06 David Malcolm <dmalcolm@redhat.com>
39316 * passes.c (execute_one_pass): Split out pass-skipping logic into...
39317 (determine_pass_name_match): ...this new function and...
39318 (should_skip_pass_p): ...this new function.
39320 2017-01-06 Nathan Sidwell <nathan@acm.org>
39322 * ipa-visibility.c (function_and_variable_visibility): Reformat
39323 comments and long lines. Remove extrneous if.
39324 * symtab.c (symtab_node::make_decl_local): Fix code format.
39325 (symtab_node::set_section_for_node): Fix comment typo.
39327 2017-01-06 Martin Liska <mliska@suse.cz>
39330 * lra-constraints.c: Rename invariant to lra_invariant.
39331 * predict.c (set_even_probabilities): Initialize e to NULL.
39333 2017-01-05 Martin Sebor <msebor@redhat.com>
39335 PR tree-optimization/78910
39336 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
39337 (format_integer): Correct off-by-one error in the handling
39338 of precision with negative numbers in signed conversions..
39340 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
39342 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
39344 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39346 PR tree-optimization/71016
39347 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
39348 factor_out_conditional_conversion. Formatting fix.
39349 (factor_out_conditional_conversion): Add cond_stmt argument.
39350 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
39351 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
39354 2017-01-05 David Malcolm <dmalcolm@redhat.com>
39356 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
39357 read-rtl-function.o, and selftest-rtl.o.
39358 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
39359 (selftest::aarch64_test_loading_full_dump): New function.
39360 (selftest::aarch64_run_selftests): New function.
39361 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
39362 selftest::aarch64_run_selftests.
39363 * config/i386/i386.c
39364 (selftest::ix86_test_loading_dump_fragment_1): New function.
39365 (selftest::ix86_test_loading_call_insn): New function.
39366 (selftest::ix86_test_loading_full_dump): New function.
39367 (selftest::ix86_test_loading_unspec): New function.
39368 (selftest::ix86_run_selftests): Call the new functions.
39369 * emit-rtl.c (maybe_set_max_label_num): New function.
39370 * emit-rtl.h (maybe_set_max_label_num): New decl.
39371 * function.c (instantiate_decls): Guard call to
39372 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
39373 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
39375 * gensupport.c (gen_reader::gen_reader): Pass "false"
39376 for new "compact" param of rtx_reader.
39377 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
39378 rather than an empty string for NULL strings.
39379 * read-md.c: Potentially include config.h rather than bconfig.h.
39380 Wrap include of errors.h with #ifdef GENERATOR_FILE.
39381 (have_error): New global, copied from errors.c.
39382 (md_reader::read_name): Rename to...
39383 (md_reader::read_name_1): ...this, adding "out_loc" param,
39384 and converting "missing name or number" to returning false, rather
39386 (md_reader::read_name): Reimplement in terms of read_name_1.
39387 (md_reader::read_name_or_nil): New function.
39388 (md_reader::read_string): Handle "(nil)" by returning NULL.
39389 (md_reader::md_reader): Add new param "compact".
39390 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
39391 (md_reader::read_file): New method.
39392 * read-md.h (md_reader::md_reader): Add new param "compact".
39393 (md_reader::read_file): New method.
39394 (md_reader::is_compact): New accessor.
39395 (md_reader::read_name): Convert return type from void to file_location.
39396 (md_reader::read_name_or_nil): New decl.
39397 (md_reader::read_name_1): New decl.
39398 (md_reader::m_compact): New field.
39399 (noop_reader::noop_reader): Pass "false" for new "compact" param
39401 (rtx_reader::rtx_reader): Add new "compact" param.
39402 (rtx_reader::read_rtx_operand): Make virtual and convert return
39403 type from void to rtx.
39404 (rtx_reader::read_until): New decl.
39405 (rtx_reader::handle_any_trailing_information): New virtual function.
39406 (rtx_reader::postprocess): New virtual function.
39407 (rtx_reader::finalize_string): New virtual function.
39408 (rtx_reader::m_in_call_function_usage): New field.
39409 (rtx_reader::m_reuse_rtx_by_id): New field.
39410 * read-rtl-function.c: New file.
39411 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
39412 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
39413 (selftest::verify_three_block_rtl_cfg): New decl.
39414 * read-rtl-function.h: New file.
39415 * read-rtl.c: Potentially include config.h rather than bconfig.h.
39416 For host, include function.h, memmodel.h, and emit-rtl.h.
39417 (one_time_initialization): New function.
39418 (struct compact_insn_name): New struct.
39419 (compact_insn_names): New array.
39420 (find_code): Handle insn codes in compact dumps.
39421 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
39422 (bind_subst_iter_and_attr): Likewise.
39423 (add_condition_to_string): Likewise.
39424 (add_condition_to_rtx): Likewise.
39425 (apply_attribute_uses): Likewise.
39426 (add_current_iterators): Likewise.
39427 (apply_iterators): Likewise.
39428 (initialize_iterators): Guard usage of apply_subst_iterator with
39429 #ifdef GENERATOR_FILE.
39430 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
39431 (md_reader::read_mapping): Likewise.
39432 (add_define_attr_for_define_subst): Likewise.
39433 (add_define_subst_attr): Likewise.
39434 (read_subst_mapping): Likewise.
39435 (check_code_iterator): Likewise.
39436 (rtx_reader::read_rtx): Likewise. Move one-time initialization
39438 (one_time_initialization): New function.
39439 (rtx_reader::read_until): New method.
39440 (read_flags): New function.
39441 (parse_reg_note_name): New function.
39442 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
39443 Handle reuse_rtx ids.
39444 Wrap iterator lookup within #ifdef GENERATOR_FILE.
39445 Add parsing support for RTL dumps, mirroring the special-cases in
39446 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
39447 values, and calling handle_any_trailing_information.
39448 (rtx_reader::read_rtx_operand): Convert return type from void
39449 to rtx, returning return_rtx. Handle case 'e'. Call
39450 finalize_string on XSTR and XTMPL fields.
39451 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
39452 "(nil)" values were omitted. Call the postprocess vfunc on the
39454 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
39455 class ctor. Initialize m_in_call_function_usage. Call
39456 one_time_initialization.
39457 * rtl-tests.c (selftest::test_uncond_jump): Call
39458 set_new_first_and_last_insn.
39459 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
39460 * selftest-rtl.c: New file.
39461 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
39462 (selftest::get_insn_by_uid): New decl.
39463 * selftest-run-tests.c (selftest::run_tests): Call
39464 read_rtl_function_c_tests.
39465 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
39466 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
39469 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
39471 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
39472 operands in a special way. Assert that pos+len <= mode precision.
39474 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39476 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
39477 3 argument Alias with unlimited for the negative form.
39478 (fno-vect-cost-model): Removed.
39480 2017-01-05 Martin Liska <mliska@suse.cz>
39482 * hsa-gen.c (gen_hsa_divmod): New function.
39483 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
39485 2017-01-05 Martin Liska <mliska@suse.cz>
39488 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
39491 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39493 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
39494 small constant length operands.
39496 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39498 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
39499 between loop iterations.
39501 2017-01-05 Martin Liska <mliska@suse.cz>
39504 * gimplify.c (gimplify_decl_expr): Compare to
39505 asan_poisoned_variables instread of checking flags.
39506 (gimplify_target_expr): Likewise.
39507 (gimplify_expr): Likewise.
39508 (gimplify_function_tree): Conditionally initialize
39509 asan_poisoned_variables.
39511 2017-01-04 Jeff Law <law@redhat.com>
39513 PR tree-optimizatin/78812
39514 * rtl.h (contains_mem_rtx_p): Prototype.
39515 * ifcvt.c (containts_mem_rtx_p): Move from here to...
39516 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
39517 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
39518 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
39519 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
39521 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39523 * input.c (assert_char_at_range): Default-initialize actual_range.
39525 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39527 * df-scan.c (df_ref_create_structure): Make regno unsigned,
39528 to match the caller.
39530 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39532 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
39533 insns after final jump in test to emit dummy move.
39535 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39537 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
39538 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
39540 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39542 * multiple_target.c (create_dispatcher_calls): Init e_next.
39543 * tree-ssa-loop-split.c (split_loop): Init border.
39544 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
39547 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
39552 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
39553 (altivec_register_operand): Do not return true if the operand
39554 contains a SUBREG mixing SImode and SFmode.
39555 (vsx_register_operand): Likewise.
39556 (vsx_reg_sfsubreg_ok): New predicate.
39557 (vfloat_operand): Do not return true if the operand contains a
39558 SUBREG mixing SImode and SFmode.
39559 (vint_operand): Likewise.
39560 (vlogical_operand): Likewise.
39561 (gpc_reg_operand): Likewise.
39562 (int_reg_operand): Likewise.
39563 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
39564 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
39565 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
39567 (rs6000_emit_move_si_sf_subreg): New helper function.
39568 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
39569 fixup SUBREGs involving SImode and SFmode.
39570 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
39571 numbers for the new peephole2 optimization.
39572 (peephole2 for SFmode unions): New peephole2 to optimize cases in
39573 the GLIBC math library that do AND/IOR/XOR operations on single
39574 precision floating point.
39575 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
39576 target macros to say whether we need to avoid SUBREGs mixing
39578 (TARGET_ALLOW_SF_SUBREG): Likewise.
39579 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
39580 (UNSPEC_SI_FROM_SF): Likewise.
39581 (iorxor): Change spacing.
39582 (and_ior_xor): New iterator for AND, IOR, and XOR.
39583 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
39584 (movdi_from_sf_zero_ext): Likewise.
39585 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
39586 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
39587 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
39588 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
39589 (fms<mode>4): Likewise.
39590 (fnma<mode>4): Likewise.
39591 (fnms<mode>4): Likewise.
39592 (nfma<mode>4): Likewise.
39593 (nfms<mode>4): Likewise.
39595 2017-01-04 Marek Polacek <polacek@redhat.com>
39598 * doc/invoke.texi: Document -Wpointer-compare.
39600 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39602 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
39605 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
39606 descriptions for -gdwarf-5 and emit them as uleb128 instead of
39609 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
39612 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
39613 documentation of the powerpc_popcntb_ok attribute.
39614 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
39615 code to issue warning messages if a requested CPU configuration is
39616 not supported by the binary (assembler and loader) toolchain.
39617 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
39618 made to define a built-in function that has been disabled.
39619 (paired_init_builtins): Add assertion to prevent ICE if attempt is
39620 made to define a built-in function that has been disabled.
39621 (altivec_init_builtins): Add comment explaining why definition
39622 of the DST built-in functions is not preceded by an assertion
39623 check. Add assertions to prevent ICE if attempts are made to
39624 define an altivec predicate or an abs* built-in function that has
39626 (htm_init_builtins): Add comment explaining why definition of the
39627 htm built-in functions is not preceded by an assertion check.
39629 2017-01-04 Jeff Law <law@redhat.com>
39631 PR tree-optimizatin/67955
39632 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
39633 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
39634 the points-to solution does not include pt_null. Use DECL_PT_UID
39637 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
39639 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
39640 Use gen_int_mode instead of gen_lopwart for const_int operands.
39642 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39644 PR tree-optimization/71563
39645 * match.pd: Simplify X << Y into X if Y is known to be 0 or
39646 out of range value - has low bits known to be zero.
39648 2017-01-04 Alan Modra <amodra@gmail.com>
39650 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
39651 * configure: Regenerate.
39652 * config.in: Regenerate.
39654 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39657 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
39658 a substring of the message, but strcmp with the whole message. Ifdef
39659 ENABLE_NLS, translate the message first using dgettext.
39661 2017-01-03 Jeff Law <law@redhat.com>
39663 PR tree-optimizatin/78856
39664 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
39665 (mark_threaded_blocks): Remove code to truncate thread paths that
39666 cross multiple loop headers. Instead invalidate the cached loop
39667 iteration information and handle case of a thread path walking
39668 into an irreducible region.
39670 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
39673 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
39674 assertions. Add support for doing the signbit if the IEEE 128-bit
39675 floating point value is in a GPR.
39676 * config/rs6000/rs6000.md (Fsignbit): Delete.
39677 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
39678 Update the length attribute if the value is in a GPR.
39679 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
39680 the sign or zero extension instruction, since the value is always 0/1.
39681 (signbit<mode>2_dm2): Delete using <Fsignbit>.
39684 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
39685 extracting SImode to a GPR register so that we can generate a
39686 store, limit the vector to be in a traditional Altivec register
39687 for the vextuwrx instruction.
39689 2017-01-03 Ian Lance Taylor <iant@google.com>
39691 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
39693 2017-01-03 Martin Sebor <msebor@redhat.com>
39695 PR tree-optimization/78696
39696 * gimple-ssa-sprintf.c (format_floating): Correct handling of
39697 precision. Use MPFR for %f for greater fidelity. Correct handling
39699 (pass_sprintf_length::compute_format_length): Set width and precision
39700 specified by asrerisk to void_node for vararg functions.
39701 (try_substitute_return_value): Adjust dump output.
39703 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
39705 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
39707 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
39709 * doc/invoke.texi (SPARC options): Document -mlra as the default.
39710 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
39711 -mlra/-mno-lra was passed to the compiler.
39713 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
39715 PR rtl-optimization/65618
39716 * emit-rtl.c (try_split): Move initialization of "before" and
39717 "after" to just before the call to emit_insn_after_setloc.
39719 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
39721 * doc/md.texi (Standard Names): Remove reference to Java frontend.
39723 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
39725 * dwarf2out.c (gen_enumeration_type_die): When
39726 -gno-strict-dwarf, add a DW_AT_encoding attribute.
39728 2017-01-03 Jakub Jelinek <jakub@redhat.com>
39730 PR tree-optimization/78965
39731 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
39732 Change first argument from const call_info & to call_info &. For %n
39733 set info.nowrite to false.
39735 PR middle-end/78901
39736 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
39737 possibly throwing calls.
39739 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
39740 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
39741 and fns handling, rather than in a separate case SSA_NAME.
39743 2017-01-02 Jeff Law <law@redhat.com>
39745 * config/darwin-driver.c (darwin_driver_init): Const-correctness
39746 fixes for first_period and second_period variables.
39748 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
39751 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
39752 (*insvqi_1): New insn pattern.
39753 (*insvqi_1_mem_rex64): Ditto.
39754 (*insvqi_2): Ditto.
39755 (*insvqi_3): Rename from *insvqi.
39757 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
39759 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
39761 * doc/cfg.texi (Edges): Remove reference to Java.
39762 (Maintaining the CFG): Ditto.
39764 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39766 PR middle-end/77674
39767 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
39768 transparent aliases.
39770 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39772 PR middle-end/77484
39773 * predict.def (PRED_CALL): Update hitrate.
39774 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
39775 * predict.c (tree_estimate_probability_bb): Split CALL predictor
39776 into direct/indirect/polymorphic variants.
39778 2017-01-01 Jakub Jelinek <jakub@redhat.com>
39780 Update copyright years.
39782 * gcc.c (process_command): Update copyright notice dates.
39783 * gcov-dump.c (print_version): Ditto.
39784 * gcov.c (print_version): Ditto.
39785 * gcov-tool.c (print_version): Ditto.
39786 * gengtype.c (create_file): Ditto.
39787 * doc/cpp.texi: Bump @copying's copyright year.
39788 * doc/cppinternals.texi: Ditto.
39789 * doc/gcc.texi: Ditto.
39790 * doc/gccint.texi: Ditto.
39791 * doc/gcov.texi: Ditto.
39792 * doc/install.texi: Ditto.
39793 * doc/invoke.texi: Ditto.
39795 Copyright (C) 2017 Free Software Foundation, Inc.
39797 Copying and distribution of this file, with or without modification,
39798 are permitted in any medium without royalty provided the copyright
39799 notice and this notice are preserved.