1 2017-12-18 Jeff Law <law@redhat.com>
3 PR tree-optimization/83477
4 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
5 a non-virtual PHI, always push a new range.
7 2017-12-19 Martin Sebor <msebor@redhat.com>
10 * builtins.c (compute_objsize): Handle non-constant offsets.
12 2017-12-19 Jakub Jelinek <jakub@redhat.com>
14 PR tree-optimization/83444
15 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
16 character load case, if get_stridx on MEM_REF's operand doesn't
17 look usable, retry with get_addr_stridx.
19 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
22 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
23 (variable_tracking_main_1): Keep markers even when VTA fails.
26 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
27 even if there are markers before it.
28 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
30 2017-12-19 Jakub Jelinek <jakub@redhat.com>
32 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
33 typical order conditions.
34 * sel-sched.c (extract_new_fences_from): Likewise.
35 * config/visium/constraints.md (J, K, L): Likewise.
36 * config/visium/predicates.md (const_shift_operand): Likewise.
37 * config/visium/visium.c (visium_legitimize_address,
38 visium_legitimize_reload_address): Likewise.
39 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
40 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
41 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
42 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
43 SET_DUMP_DETAIL): Likewise.
44 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
45 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
46 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
47 avr_set_core_architecture, avr_set_current_function,
48 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
49 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
50 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
51 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
52 avr_map_decompose, avr_fold_builtin): Likewise.
53 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
54 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
55 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
56 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
57 * config/m32c/m32c.c (m32c_conditional_register_usage,
58 m32c_address_cost): Likewise.
59 * config/m32c/predicates.md (shiftcount_operand,
60 longshiftcount_operand): Likewise.
61 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
62 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
63 can_use_cdx_ldstw): Likewise.
64 * config/nios2/nios2.h (CDX_REG_P): Likewise.
65 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
67 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
68 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
69 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
70 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
71 * config/vax/vax.c (adjacent_operands_p): Likewise.
72 * config/ft32/constraints.md (L, b, KA): Likewise.
73 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
75 * cfgexpand.c (expand_stack_alignment): Likewise.
76 * gcse.c (insert_expr_in_table): Likewise.
77 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
78 * cgraphunit.c (cgraph_node::expand): Likewise.
79 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
80 * emit-rtl.c (add_insn): Likewise.
81 * input.c (dump_location_info): Likewise.
82 * passes.c (NEXT_PASS): Likewise.
83 * read-rtl-function.c (parse_note_insn_name,
84 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
86 * sched-rgn.c (sched_rgn_init): Likewise.
87 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
88 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
89 if_then_else_cond, simplify_shift_const_1, simplify_comparison): Likewise.
90 * explow.c (eliminate_constant_term): Likewise.
91 * final.c (leaf_renumber_regs_insn): Likewise.
92 * cfgrtl.c (print_rtl_with_bb): Likewise.
93 * genhooks.c (emit_init_macros): Likewise.
94 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
95 * tree-data-ref.c (conflict_fn): Likewise.
96 * selftest.c (assert_streq): Likewise.
97 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
98 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
99 fold_binary_loc, multiple_of_p): Likewise.
100 * reload.c (push_reload, find_equiv_reg): Likewise.
101 * et-forest.c (et_nca, et_below): Likewise.
102 * dbxout.c (dbxout_symbol_location): Likewise.
103 * reorg.c (relax_delay_slots): Likewise.
104 * dojump.c (do_compare_rtx_and_jump): Likewise.
105 * gengtype-parse.c (type): Likewise.
106 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
107 simplify_const_relational_operation): Likewise.
108 * reload1.c (do_output_reload): Likewise.
109 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
110 * gengtype.c (type_for_name): Likewise.
111 * gimple-ssa-sprintf.c (format_directive): Likewise.
113 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
116 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
117 accessing it. Adjust comment.
119 2017-12-19 Jakub Jelinek <jakub@redhat.com>
122 * predict.c (zero_one_minusone): New function.
123 (apply_return_prediction): Avoid return prediction for functions
124 returning only -1, 0 and 1 values, unless they only return -1 and 0
127 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
129 * config/arc/arc.c (legitimate_scaled_address_p): Clean
130 fall-through warning.
131 (arc_compute_frame_size): Remove unused variables.
132 (arc_print_operand): Fix fprintif format.
133 (arc_can_follow_jump): Clean fall-through warning.
135 2017-12-19 Marek Polacek <polacek@redhat.com>
138 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
141 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
143 * config/arc/arc.c (overriderregs): New variable.
144 (arc_override_options): Track fixed/call saved/call options.
145 (arc_conditional_register_usage): Check against overrideregs
146 variable whenever we change register properties.
148 2017-12-19 Nathan Sidwell <nathan@acm.org>
150 * opts.c (finish_options): Don't prefix dump_base_name if it
151 already contains directories.
153 2017-12-19 Martin Liska <mliska@suse.cz>
155 PR rtl-optimization/82675
156 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
157 more element in sbitmap.
159 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
161 * gimplify.c (gimplify_expr): Use error_operand_p.
163 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
166 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
169 2017-12-19 Daniel Cederman <cederman@gaisler.com>
171 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
172 the jump is to a label.
174 2017-12-19 Jakub Jelinek <jakub@redhat.com>
176 PR tree-optimization/83444
177 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
182 * ipa-inline.c (flatten_remove_node_hook): New function.
183 (ipa_inline): Keep only nodes with flatten attribute at the end of
184 the array in the order from ipa_reverse_postorder, only walk that
185 portion of array for flattening, if there is more than one such
186 node, temporarily register a removal hook and ignore removed nodes.
188 PR tree-optimization/80631
189 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
190 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
191 IFN_REDUC_MAX or IFN_REDUC_MIN.
193 2017-12-18 Jeff Law <law@redhat.com>
195 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
196 record symbolic equivalences from backedges in the CFG.
199 2017-11-19 Jeff Law <law@redhat.com>
201 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
202 of degenerates resulting from ignoring an edge.
204 2017-12-18 Martin Sebor <msebor@redhat.com>
207 PR tree-optimization/78450
208 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
209 (handle_builtin_strlen): Call it.
211 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
213 PR rtl-optimization/83424
214 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
216 2017-12-18 Marek Polacek <polacek@redhat.com>
219 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
220 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
223 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
225 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
226 (umaddsidi4, umaddsidi_split): Likewise.
228 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
230 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
233 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
236 * config/s390/s390.c (s390_option_override): Avoid strncpy.
238 2017-12-18 Richard Biener <rguenther@suse.de>
240 PR tree-optimization/81877
241 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
242 (outermost_indep_loop): Adjust.
243 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
244 (can_sm_ref_p): Adjust.
246 2017-12-18 Richard Biener <rguenther@suse.de>
249 * tree.c (array_at_struct_end_p): Return true if the underlying
250 object has space for at least one element in excess of what
251 the array domain specifies.
253 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
255 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
256 example to avoid overfull hbox.
257 * doc/invoke.texi (Option Summary): Add missing @gol.
258 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
261 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
262 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
264 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
265 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
266 (AArch64 Options): Document -mverbose-cost-dump.
267 (ARM Options): Likewise, plus -mflip-thumb.
269 2017-12-17 Martin Sebor <msebor@redhat.com>
272 * gimple-ssa-warn-restrict.c
273 (builtin_memref::offset_out_of_bounds): Correct the handling of
276 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
278 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
280 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
282 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
284 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
286 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
289 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
291 * doc/invoke.texi: Fix some typos.
293 2017-12-16 Martin Sebor <msebor@redhat.com>
295 PR tree-optimization/78918
296 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
297 * builtins.c (check_sizes): Rename...
298 (check_access): ...to this. Rename function arguments for clarity.
299 (check_memop_sizes): Adjust names.
300 (expand_builtin_memchr, expand_builtin_memcpy): Same.
301 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
302 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
303 (check_strncat_sizes, expand_builtin_strncat): Same.
304 (expand_builtin_strncpy, expand_builtin_memset): Same.
305 (expand_builtin_bzero, expand_builtin_memcmp): Same.
306 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
307 (maybe_emit_sprintf_chk_warning): Same.
308 (expand_builtin_strcpy): Adjust.
309 (expand_builtin_stpcpy): Same.
310 (expand_builtin_with_bounds): Detect out-of-bounds accesses
311 in pointer-checking forms of memcpy, memmove, and mempcpy.
312 (gcall_to_tree_minimal, max_object_size): Define new functions.
313 * builtins.h (max_object_size): Declare.
314 * calls.c (alloc_max_size): Call max_object_size instead of
315 hardcoding ssizetype limit.
316 (get_size_range): Handle new argument.
317 * calls.h (get_size_range): Add a new argument.
318 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
319 * doc/invoke.texi (-Wrestrict): Adjust, add example.
320 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
322 (gimple_fold_builtin_memory_chk): Same.
323 (gimple_fold_builtin_stxcpy_chk): New function.
324 * gimple-ssa-warn-restrict.c: New source.
325 * gimple-ssa-warn-restrict.h: New header.
326 * gimple.c (gimple_build_call_from_tree): Propagate location.
327 * passes.def (pass_warn_restrict): Add new pass.
328 * tree-pass.h (make_pass_warn_restrict): Declare.
329 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
331 (handle_builtin_strcat): Same.
332 (strlen_optimize_stmt): Rename...
333 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
334 stpncpy, strncpy, and their checking forms.
336 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
338 PR rtl-optimization/82849
339 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
340 and get_max_loop_iterations_int.
342 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
343 Alan Hayward <alan.hayward@arm.com>
344 David Sherwood <david.sherwood@arm.com>
346 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
347 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
348 size as a poly_uint64.
349 (mode_for_vector, mode_for_int_vector): Take the number of vector
350 elements as a poly_uint64.
351 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
352 size as a poly_uint64.
353 (mode_for_vector, mode_for_int_vector): Take the number of vector
354 elements as a poly_uint64.
356 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
357 Alan Hayward <alan.hayward@arm.com>
358 David Sherwood <david.sherwood@arm.com>
360 * machmode.h (MACRO_MODE): New macro.
361 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
362 * targhooks.c (default_libcall_value, default_secondary_reload)
363 (default_memory_move_cost, default_register_move_cost)
364 (default_class_max_nregs): Likewise.
366 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
367 Alan Hayward <alan.hayward@arm.com>
368 David Sherwood <david.sherwood@arm.com>
370 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
371 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
373 * genautomata.c (main): Likewise.
374 * genconditions.c (write_header): Likewise.
375 * genemit.c (main): Likewise.
376 * genextract.c (print_header): Likewise.
377 * genopinit.c (main): Likewise.
378 * genoutput.c (output_prologue): Likewise.
379 * genpeep.c (main): Likewise.
380 * genpreds.c (write_insn_preds_c): Likewise.
381 * genrecog.c (writer_header): Likewise.
382 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
383 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
384 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
385 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
386 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
387 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
388 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
389 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
390 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
391 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
392 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
393 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
394 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
395 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
396 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
397 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
398 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
399 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
400 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
401 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
402 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
403 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
404 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
405 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
406 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
407 * config/darwin.c (IN_TARGET_CODE): Likewise.
408 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
409 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
410 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
411 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
412 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
413 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
414 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
415 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
416 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
417 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
418 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
419 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
420 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
421 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
422 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
423 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
424 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
425 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
426 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
427 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
428 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
429 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
430 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
431 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
432 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
433 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
434 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
435 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
436 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
437 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
438 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
439 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
440 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
441 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
442 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
443 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
444 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
445 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
446 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
447 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
448 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
449 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
450 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
451 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
452 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
453 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
454 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
455 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
456 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
457 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
458 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
459 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
460 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
461 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
462 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
463 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
464 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
465 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
466 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
467 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
468 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
469 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
470 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
471 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
472 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
473 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
474 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
475 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
476 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
477 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
478 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
479 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
480 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
481 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
482 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
483 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
484 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
485 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
486 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
487 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
488 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
489 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
490 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
491 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
492 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
493 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
494 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
495 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
496 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
497 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
498 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
499 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
500 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
501 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
502 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
503 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
504 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
505 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
506 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
507 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
508 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
509 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
510 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
511 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
512 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
513 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
515 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
516 Alan Hayward <alan.hayward@arm.com>
517 David Sherwood <david.sherwood@arm.com>
519 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
522 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
523 Alan Hayward <alan.hayward@arm.com>
524 David Sherwood <david.sherwood@arm.com>
526 * doc/generic.texi (VEC_SERIES_EXPR): Document.
527 * doc/md.texi (vec_series@var{m}): Document.
528 * tree.def (VEC_SERIES_EXPR): New tree code.
529 * tree.h (build_vec_series): Declare.
530 * tree.c (build_vec_series): New function.
531 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
532 * tree-pretty-print.c (dump_generic_node): Likewise.
533 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
534 * tree-inline.c (estimate_operator_cost): Likewise.
535 * expr.c (expand_expr_real_2): Likewise.
536 * optabs-tree.c (optab_for_tree_code): Likewise.
537 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
538 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
539 * expmed.c (make_tree): Handle VEC_SERIES.
540 * optabs.def (vec_series_optab): New optab.
541 * optabs.h (expand_vec_series_expr): Declare.
542 * optabs.c (expand_vec_series_expr): New function.
543 * tree-vect-generic.c (expand_vector_operations_1): Check that
544 the operands also have vector type.
546 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
547 Alan Hayward <alan.hawyard@arm.com>
548 David Sherwood <david.sherwood@arm.com>
550 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
551 (VEC_COND_EXPR): Add missing @tindex.
552 * doc/md.texi (vec_duplicate@var{m}): Document.
553 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
554 * tree.c (build_vector_from_val): Add stubbed-out handling of
555 variable-length vectors, using VEC_DUPLICATE_EXPR.
556 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
557 * cfgexpand.c (expand_debug_expr): Likewise.
558 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
559 * tree-inline.c (estimate_operator_cost): Likewise.
560 * tree-pretty-print.c (dump_generic_node): Likewise.
561 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
562 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
563 (test_vec_duplicate_folding): New function.
564 (fold_const_c_tests): Call it.
565 * optabs.def (vec_duplicate_optab): New optab.
566 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
567 * optabs.h (expand_vector_broadcast): Declare.
568 * optabs.c (expand_vector_broadcast): Make non-static. Try using
570 * expr.c (store_constructor): Try using vec_duplicate_optab for
572 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
574 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
577 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
578 div/mod latencies a bit.
580 2017-12-15 Jeff Law <law@redhat.com>
582 PR tree-optimization/36550
583 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
584 (mark_threaded_blocks): Rewrite code to avoid block copying when
585 optimizing for size. Don't pessimize blocks which will be
586 copied, but all the statements will be dead.
588 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
590 PR tree-optimization/81165
591 * tree-ssa-threadupdate.c (uses_in_bb): New.
592 (estimate_threading_killed_stmts): New.
593 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
594 * tree-ssa-threadedge.c
595 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
598 2017-12-15 Jeff Law <law@redhat.com>
600 PR tree-optimization/83410
601 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
602 threads when parallelizing loops.
604 2017-12-15 Jakub Jelinek <jakub@redhat.com>
606 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
608 * config/alpha/alpha.c (vms_attribute_table): Swap
609 affects_type_identity and handler fields, adjust comments.
610 * config/mips/mips.c (mips_attribute_table): Likewise.
611 * config/visium/visium.c (visium_attribute_table): Likewise.
612 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
613 * config/microblaze/microblaze.c (microblaze_attribute_table):
615 * config/spu/spu.c (spu_attribute_table): Likewise.
616 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
617 * config/arc/arc.c (arc_attribute_table): Likewise.
618 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
619 * config/v850/v850.c (v850_handle_interrupt_attribute,
620 v850_handle_data_area_attribute): Formatting fixes.
621 (v850_attribute_table): Swap affects_type_identity and handler
622 fields, adjust comments.
623 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
624 * config/arm/arm.c (arm_attribute_table): Likewise.
625 * config/avr/avr.c (avr_attribute_table): Likewise.
626 * config/s390/s390.c (s390_attribute_table): Likewise.
627 * config/sh/sh.c (sh_attribute_table): Likewise.
628 * config/i386/i386.c (ix86_handle_cconv_attribute,
629 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
630 (ix86_attribute_table): Swap affects_type_identity and handler
631 fields, adjust comments.
632 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
633 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
634 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
635 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
636 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
637 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
638 * config/rx/rx.c (rx_attribute_table): Likewise.
639 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
640 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
641 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
642 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
643 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
644 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
645 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
646 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
647 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
648 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
649 * doc/plugins.texi (user_attr): Likewise. Add NULL for
651 * attribs.c (empty_attribute_table): Swap affects_type_identity and
653 (register_scoped_attributes, decl_attributes): Formatting fixes.
655 PR tree-optimization/83269
656 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
657 subtraction in arg0's type if type is signed and arg0 is unsigned.
661 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
662 convert? on @0 instead of convert. Check type of @1, not @0.
663 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
664 convert? on @0 instead of convert. Check type of @1, not @0.
665 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
666 only optimize if either both @1 and @2 types are narrower
667 precision, or both are wider or equal precision, and in the former
668 case only if both have undefined overflow.
670 2017-12-15 Richard Biener <rguenther@suse.de>
673 * internal-fn.def (IFN_NOP): Add.
674 * internal-fn.c (expand_NOP): Do nothing.
675 * lto-streamer-in.c (input_function): Instead of removing
676 sanitizer calls replace them with IFN_NOP calls.
678 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
679 Alan Hayward <alan.hayward@arm.com>
680 David Sherwood <david.sherwood@arm.com>
682 * dse.c (store_info, read_info_type): Replace begin and end with
684 (print_range): New function.
685 (set_all_positions_unneeded, any_positions_needed_p)
686 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
688 (record_store): Likewise. Optimize the case in which all positions
690 (get_stored_val): Replace read_begin and read_end with read_offset
692 (replace_read): Update call accordingly.
694 2017-12-15 Bin Cheng <bin.cheng@arm.com>
696 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
697 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
698 (loop_cand::loop_cand): Initialize above members.
699 (loop_cand::supported_operations): Delete.
700 (loop_cand::can_interchange_p): Inline above function.
701 (loop_cand::classify_simple_reduction): Record number of constant
702 initialized simple reductions.
703 (should_interchange_loops): New parameters. Check stmt cost of loops
705 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
706 Update call to should_interchange_loops.
707 (should_interchange_loop_nest): Update call to
708 should_interchange_loops.
710 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
713 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
714 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
715 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
717 2017-12-15 Julia Koval <julia.koval@intel.com>
719 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
720 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi): New.
721 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
722 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
723 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
725 2017-12-15 Julia Koval <julia.koval@intel.com>
727 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
728 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
729 * config/i386/sse.md (vaesenc_<mode>): New pattern.
730 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
731 _mm_aesenc_epi128): New intrinsics.
733 2017-12-15 Julia Koval <julia.koval@intel.com>
735 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
736 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi): New.
737 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
738 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
739 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
741 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
743 * gimple-ssa-strength-reduction.c (analyze_increments):
744 Distinguish replacement costs for constant strides from those for
747 2017-12-14 Jakub Jelinek <jakub@redhat.com>
749 * var-tracking.c (variable_tracking_main_1): Formatting fix.
751 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
753 * doc/invoke.texi: Document -Wcast-function-type.
754 * recog.h (stored_funcptr): Change signature.
755 * tree-dump.c (dump_node): Avoid warning.
756 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
758 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
761 * gimple-fold.c (get_range_strlen): Add the handling of non-member
764 2017-12-14 David Malcolm <dmalcolm@redhat.com>
766 PR tree-optimization/83312
767 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
768 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
769 "val" param, and to cope with arbitrary basic blocks.
770 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
771 handle NULL_TREE for "val", dropping "bb" param.
772 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
773 drop "bb" param. Handle NULL_TREE for "val".
774 (find_case_label_for_value): Make "switch_stmt" param const.
775 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
777 (vrp_prop::check_all_array_refs): Reimplement as...
778 (check_array_bounds_dom_walker::before_dom_children): ...this new
779 vfunc. Replace linear search through BB block list, excluding
780 those with non-executable in-edges via dominator walk.
782 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
784 * config/arm/arm.opt (mverbose-cost-dump): New option.
785 * config/arm/arm.c (arm_rtx_costs): Use it.
787 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
790 * reload1.c (emit_input_reload_insns): Skip debug markers.
792 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
794 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
798 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
799 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
800 * config/sh/sh.c: Skip debug insns besides notes.
801 * config/sh/sh.md: Likewise.
802 * config/sh/sh_treg_combine.cc: Likewise.
803 * config/sh/sync.md: Likewise.
805 2017-12-14 Tom de Vries <tom@codesourcery.com>
807 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
808 Add item for weak_undefined.
810 2017-12-14 Richard Biener <rguenther@suse.de>
812 PR tree-optimization/67842
813 * sese.h (bb_in_region): Remove #if 0'ed code.
815 2017-12-14 Richard Biener <rguenther@suse.de>
817 PR tree-optimization/83326
818 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
819 may_be_zero parameter and handle it by not marking the first
820 peeled copy as not exiting the loop.
821 (try_peel_loop): Likewise.
822 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
823 to handle the case of constant or zero iterations and perform
824 loop header copying on-the-fly.
826 2017-12-14 Richard Biener <rguenther@suse.de>
828 PR tree-optimization/83418
829 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
830 Instead of asserting we don't get unfolded comparisons deal with
833 2017-12-14 Jakub Jelinek <jakub@redhat.com>
836 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
837 basic blocks. Assert debug bind insns don't appear outside of bbs,
838 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
841 PR tree-optimization/83198
842 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
843 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
844 value if arg is a REAL_CST with incompatible type.
846 2017-12-14 Sudakshina Das <sudi.das@arm.com>
847 Bin Cheng <bin.cheng@arm.com>
850 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
852 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
855 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
857 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
858 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
860 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
861 Alan Hayward <alan.hayward@arm.com>
862 David Sherwood <david.sherwood@arm.com>
864 * poly-int.h: New file.
865 * poly-int-types.h: Likewise.
866 * coretypes.h: Include them.
867 (POLY_INT_CONVERSION): Define.
868 * target.def (estimated_poly_value): New hook.
869 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
870 * doc/tm.texi: Regenerate.
871 * doc/poly-int.texi: New file.
872 * doc/gccint.texi: Include it.
873 * doc/rtl.texi: Describe restrictions on subreg modes.
874 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
875 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
876 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
877 * targhooks.h (default_estimated_poly_value): Declare.
878 * targhooks.c (default_estimated_poly_value): New function.
879 * target.h (estimated_poly_value): Likewise.
880 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
881 (wi::unary_traits): Delete.
882 (wi::binary_traits::signed_shift_result_type): Define for
883 offset_int << HOST_WIDE_INT, etc.
884 (generic_wide_int::operator <<=): Define for all types and use
885 wi::lshift instead of <<.
886 (wi::hwi_with_prec): Add a default constructor.
887 (wi::ints_for): New class.
888 (operator <<): Define for all wide-int types.
889 (operator /): New function.
890 (operator %): Likewise.
891 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
892 (ASSERT_MAYBE_NE_AT): New macros.
894 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
895 Dominik Vogt <vogt@linux.vnet.ibm.com>
898 * emit-rtl.c (init_emit): Remove ??? comment.
899 * explow.c (get_dynamic_stack_size): Take known alignment of stack
900 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
901 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
902 alignment of 3 virtual registers to BITS_PER_WORD.
904 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
906 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
908 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
909 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
911 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
914 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
916 2017-12-13 Jakub Jelinek <jakub@redhat.com>
918 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
919 with the exception of debug begin stmt markers appear before
923 * final.c (rest_of_handle_final): Call variable_tracking_main only
924 if !flag_var_tracking.
926 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
927 Jakub Jelinek <jakub@redhat.com>
931 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
932 labels debug stmts that can only appear after labels.
934 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
936 PR rtl-optimization/82398
937 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
938 EXPR_USEFULNESS in priority comparison.
940 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
942 PR rtl-optimization/83393
943 * combine.c (move_deaths): If reg_stat points to a too new insn in
944 last_death, do not use it: find the proper insn instead.
946 2017-12-12 Jeff Law <law@redhat.com>
948 PR tree-optimization/83298
949 PR tree-optimization/83362
950 PR tree-optimization/83383
951 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
952 push_value_range a public interface. Add new argument to
953 record_ranges_from_stmt.
954 * gimple-ssa-evrp-analyze.c
955 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
956 Update comments. Handle recording temporary equivalences.
957 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
958 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
959 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
960 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
961 gimple-ssa-evrp-analyze.h.
962 (record_temporary_equivalences_from_phis): Add new argument. When
963 the PHI arg is an SSA_NAME, set the result's range to the range
965 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
967 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
968 Pass it down to children as needed.
969 (thread_outgoing_edges): Likewise.
970 (thread_across_edge): Likewise. Push/pop range state as needed.
971 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
973 2017-12-12 Julia Koval <julia.koval@intel.com>
975 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
976 (PTA_CANNONLAKE): Remove PTA_CLWB.
978 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
981 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
982 (vcondv2div2df): Likewise.
983 (vconduv2dfv2di): Likewise.
985 2017-12-12 Jakub Jelinek <jakub@redhat.com>
987 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
988 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
989 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
990 ATTR_MATHFN_FPROUNDING_ERRNO.
992 2017-12-12 Richard Biener <rguenther@suse.de>
994 PR tree-optimization/83385
995 * graphite-scop-detection.c (get_order, order): Remove.
996 (bb_to_rpo): New global.
998 (build_scops): Sort pbbs in RPO order.
1000 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
1002 * combine.c (simplify_set): Do not transform subregs to zero_extends
1003 if the destination mode is a vector mode.
1005 2017-12-12 Jakub Jelinek <jakub@redhat.com>
1007 PR tree-optimization/83359
1008 * tree-cfg.h (fold_loop_internal_call): Declare.
1009 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
1010 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
1011 (find_loop_dist_alias): New function.
1012 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
1013 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
1016 PR tree-optimization/80631
1017 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
1018 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
1019 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
1020 hardcoding zero as the value if COND_EXPR is never true. For
1021 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
1022 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
1023 hardcoding MAX_EXPR as the reduction operation.
1024 (is_nonwrapping_integer_induction): Allow negative step.
1025 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
1026 vect_create_epilog_for_reduction, if no value is suitable, don't
1027 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
1029 2017-12-12 Richard Biener <rguenther@suse.de>
1031 PR tree-optimization/81889
1032 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
1033 range info from the non-wrapping IV instead of just the range
1036 2017-12-12 Julia Koval <julia.koval@intel.com>
1038 * config.gcc: Add vaesintrin.h.
1039 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1041 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1042 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1044 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1045 * config/i386/immintrin.h: Include vaesintrin.h.
1046 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1047 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1048 _mm_aesdec_epi128): New intrinsics.
1050 2017-12-12 Julia Koval <julia.koval@intel.com>
1052 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
1053 OPTION_MASK_ISA_VAES_UNSET): New.
1054 (ix86_handle_option): Handle -mvaes.
1055 * config/i386/cpuid.h: Define bit_VAES.
1056 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
1057 * config/i386/i386-c.c (__VAES__): New.
1058 * config/i386/i386.c (ix86_target_string): Add -mvaes.
1059 (ix86_valid_target_attribute_inner_p): Ditto.
1060 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
1061 * config/i386/i386.opt: Add -mvaes.
1062 * doc/invoke.texi: Ditto.
1064 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
1066 * debug.h (gcc_debug_hooks): Add inline_entry.
1067 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
1068 * debug.c (do_nothing_debug_hooks): Likewise.
1069 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
1070 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
1071 (dwarf2_lineno_debug_hooks): Likewise.
1073 * common.opt (gstatement-frontiers): New, setting
1074 debug_nonbind_markers_p.
1075 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
1076 * toplev.c (process_options): Autodetect value for debug statement
1078 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
1079 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
1081 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
1082 markers. Integrate source bind into debug stmt expand loop.
1083 (pass_expand::execute): Check debug marker limit. Avoid deep
1084 TER and expand debug locations for debug bind insns only.
1085 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
1087 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
1088 * final.c (reemit_insn_block_notes): Take current block from
1089 nonbind markers. Declare note where it's first set.
1090 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
1091 begin stmt markers if enabled.
1092 (notice_source_line): Handle nonbind markers. Fail if their
1093 location is unknown or that of builtins.
1094 (rest_of_handle_final): Convert begin stmt markers to notes if
1095 var-tracking didn't run.
1096 (rest_of_clean_state): Skip begin stmt markers.
1097 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
1099 * function.c (allocate_struct_function): Set begin_stmt_markers.
1100 * function.h (struct function): Add debug_marker_count counter
1101 and debug_nonbind_markers flag.
1102 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
1103 * gimple-low.c (lower_function_body): Adjust
1104 debug_nonbind_markers.
1105 (lower_stmt): Drop or skip gimple debug stmts.
1106 (lower_try_catch): Skip debug stmts.
1107 * gimple.c (gimple_build_debug_begin_stmt): New.
1108 (gimple_copy): Increment debug_marker_count if copying one.
1109 * gimple.h (gimple_build_debug_begin_stmt): Declare.
1110 * gimplify.c (rexpr_location): New.
1111 (rexpr_has_location): New.
1112 (warn_switch_unreachable_r): Handle gimple debug stmts.
1113 (shortcut_cond_r): Call expr_location.
1115 (find_goto_label): New.
1116 (shortcut_cond_expr): Call expr_has_location, expr_location, and
1118 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
1120 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
1121 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
1122 (LANG_HOOKS_INITIALIZER): ... this.
1123 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
1124 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
1126 (update_ebb_live_info): Skip debug insn markers.
1127 * lra.c (debug_insn_static_data): Rename to...
1128 (debug_bind_static_data): ... this.
1129 (debug_marker_static_data): New.
1130 (lra_set_insn_recog_data): Select one of the above depending
1132 (lra_update_isn_regno_info): Don't assume debug insns have
1134 (push_insns): Skip debug insns.
1135 * lto-streamer-in.c (input_function): Drop debug stmts
1136 depending on active options. Adjust debug_nonbind_markers.
1137 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
1138 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
1139 begin stmt marker notes.
1140 (print_insn): Likewise.
1141 * recog.c (extract_insn): Recognize rtl for debug markers.
1142 * rtl.def (DEBUG_MARKER): New.
1143 * tree-inline.c: Include params.h.
1144 (remap_gimple_stmt): Handle nonbind markers.
1145 (maybe_move_debug_stmts_to_successors): Likewise.
1146 (copy_debug_stmt): Likewise.
1147 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
1148 markers regardless of no side effects.
1149 (tsi_link_before): Don't update container's side effects when adding
1150 a begin stmt marker.
1151 (tsi_link_after): Likewise.
1152 (expr_first): Skip begin stmt markers.
1153 (expr_last): Likewise.
1154 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
1155 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1156 Disregard nonbind markers.
1157 * tree.c (make_node_stat): Don't set side effects for begin stmt
1159 (build1_stat): Likewise.
1160 * tree.def (DEBUG_BEGIN_STMT): New.
1161 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
1162 * var-tracking.c (delete_debug_insns): Renamed to...
1163 (delete_vta_debug_insns): ... this.
1164 (reemit_marker_as_note): New.
1165 (vt_initialize): Reemit markers.
1166 (delete_vta_debug_insns): Likewise.
1167 (vt_debug_insns_local): Reemit or delete markers.
1168 (variable_tracking_main_1): Likewise.
1169 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
1170 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
1171 (gimple_debug_nonbind_marker_p): New.
1172 (gimple_build_debug_bind): Adjust.
1173 (gimple_build_debug_begin_stmt): New.
1174 * doc/invoke.texi (max-debug-marker-count): New param.
1175 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
1176 (debug_parameter_ref, debug_marker): New.
1177 (NOTE_INSN_BEGIN_STMT): New.
1178 (DEBUG_INSN): Describe begin stmt markers.
1180 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
1181 without debug insns, we wouldn't, but clean up debug insns
1182 after a control flow insn nevertheless.
1184 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
1185 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
1186 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
1187 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
1188 (rtl_tidy_fallthru_edge): Likewise.
1189 (rtl_verify_fallthru): Likewise.
1190 (rtl_verify_bb_layout): Likewise.
1191 (skip_insns_after_block): Likewise.
1192 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
1193 * dwarf2out.c: Include print-rtl.h.
1194 (dwarf2out_next_real_insn): New.
1195 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
1196 Dump debug binds in asm comments.
1197 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
1198 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
1199 callers to use gsi_start_nondebug_bb instead.
1200 (gsi_after_labels): Skip gimple debug stmts.
1201 (gsi_start_nondebug): New.
1202 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
1203 (proper_loop_form_for_interchange): Adjust.
1204 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
1205 * gimple.h (gimple_seq_last_nondebug_stmt): New.
1206 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
1207 (collect_fallthrough_labels): Likewise.
1208 (should_warn_for_implicit_fallthrough): Likewise.
1209 (warn_implicit_fallthrough_r): Likewise.
1210 (expand_FALLTHROUGH_r): Likewise.
1211 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
1212 (graphite_copy_stmts_from_block): Skip nonbind markers.
1213 * haifa-sched.c (sched_extend_bb): Skip debug insns.
1214 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
1215 * jump.c (clean_barriers): Skip debug insns.
1216 * omp-expand.c (expand_parallel_call): Skip debug insns.
1217 (expand_task_call): Likewise.
1218 (remove_exit_barrier): Likewise.
1219 (expand_omp_taskreg): Likewise.
1220 (expand_omp_for_init_counts): Likewise.
1221 (expand_omp_for_generic): Likewise.
1222 (expand_omp_for_static_nochunk): Likewise.
1223 (expand_omp_for_static_chunk): Likewise.
1224 (expand_omp_simd): Likewise.
1225 (expand_omp_taskloop_for_outer): Likewise.
1226 (expand_omp_taskloop_for_inner): Likewise.
1227 (expand_oacc_for): Likewise.
1228 (expand_omp_sections): Likewise.
1229 (expand_omp_single): Likewise.
1230 (expand_omp_synch): Likewise.
1231 (expand_omp_atomic_load): Likewise.
1232 (expand_omp_atomic_store): Likewise.
1233 (expand_omp_atomic_fetch_op): Likewise.
1234 (expand_omp_atomic_pipeline): Likewise.
1235 (expand_omp_atomic_mutex): Likewise.
1236 (expand_omp_target): Likewise.
1237 (grid_expand_omp_for_loop): Likewise.
1238 (grid_expand_target_grid_body): Likewise.
1239 (build_omp_regions_1): Likewise.
1240 * omp-low.c (check_combined_parallel): Skip debug stmts.
1241 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
1242 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
1243 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
1245 * tree-cfg.c (make_blobs_1): Skip debug stmts.
1246 (make_edges): Likewise.
1247 (cleanup_dead_labels): Likewise.
1248 (gimple_can_merge_blocks_p): Likewise.
1249 (stmt_starts_bb_p): Likewise.
1250 (gimple_block_label): Likewise.
1251 (gimple_redirect_edge_and_branch): Likewise.
1252 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
1254 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
1256 * tree-pretty-print (print_declaration): Omit initializer in slim
1258 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
1260 (eliminate_unnecessary_stmts): Stabilize block removal order.
1261 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
1262 * var-tracking.c (get_first_insn): New.
1263 (vt_emit_notes): Call it.
1264 (vt_initialize): Walk any insns before the first BB.
1265 (delete_debug_insns): Likewise.
1267 * gimple.h (enum gimple_debug_subcode): Add
1268 GIMPLE_DEBUG_BEGIN_STMT.
1269 (gimple_debug_begin_stmt_p): New.
1270 (gimple_debug_nonbind_marker_p): New.
1271 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
1272 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
1273 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
1274 * insn-notes.def (BEGIN_STMT): New.
1275 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
1276 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
1277 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
1278 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
1279 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
1280 (INSN_DEBUG_MARKER_KIND): New.
1281 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
1282 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
1283 (INSN_VAR_LOCATION_PTR): New.
1284 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
1286 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
1287 deps only if debug bind insns are enabled.
1288 (pass_expand::execute): Avoid deep TER and expand
1289 debug locations for debug bind insns only.
1290 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
1291 debug stmts special handling down to debug bind stmts.
1292 * combine.c (try_combine): Narrow debug insns special handling
1293 down to debug bind insns.
1294 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
1295 Narrow debug insns preexisting special handling down to debug
1297 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
1298 handling down to debug bind insns.
1299 * function.c (instantiate_virtual_regs): Skip debug markers,
1300 adjust handling of debug binds.
1301 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
1302 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
1303 * haifa-sched.c (schedule_insn): Narrow special handling of debug
1304 insns to debug bind insns.
1305 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
1306 special handling of debug stmts to debug bind stmts.
1307 * ipa-split.c (split_function): Likewise.
1308 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
1309 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
1311 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
1312 * regrename.c (build_def_use): Likewise.
1313 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
1314 (pass_cprop_hardreg): Narrow special casing of debug insns to
1316 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
1317 * reload1.c (reload): Likewise.
1318 * sese.c (sese_insert_phis_for_liveouts): Narrow special
1319 casing of debug stmts to debug bind stmts.
1320 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
1321 * ssa-iterators.h (num_imm_uses): Likewise.
1322 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
1323 debug stmts to debug bind stmts.
1324 * tree-inline.c (tree_function_versioning): Narrow special casing
1325 of debug stmts to debug bind stmts.
1326 * tree-loop-distribution.c (generate_loops_for_partition):
1327 Narrow special casing of debug stmts to debug bind stmts.
1328 * tree-sra.c (analyze_access_subtree): Narrow special casing
1329 of debug stmts to debug bind stmts.
1330 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
1331 stmts to debug bind stmts.
1332 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
1333 casing of debug stmts to debug bind stmts.
1334 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
1335 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
1336 casing of debug stmts to debug bind stmts.
1337 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
1339 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
1340 debug stmts to debug bind stmts.
1341 (gimple_replace_ssa_lhs): Likewise.
1342 (insert_debug_temp_for_var_def): Likewise.
1343 (insert_debug_temps_for_defs): Likewise.
1344 (reset_debug_uses): Likewise.
1345 * tree-ssanames.c (release_ssa_name_fn): Likewise.
1346 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
1347 (adjust_debug_stmts): Likewise.
1348 (adjust_phi_and_debug_stmts): Likewise.
1349 (vect_do_peeling): Likewise.
1350 * tree-vect-loop.c (vect_transform_loop): Likewise.
1351 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
1352 * var-tracking.c (adjust_mems): Narrow special casing of debug
1353 insns to debug bind insns.
1354 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
1355 (compute_bb_dataflow, vt_find_locations): Likewise.
1356 (vt_expand_loc, emit_notes_for_changes): Likewise.
1357 (vt_init_cfa_base): Likewise.
1358 (vt_emit_notes): Likewise.
1359 (vt_initialize): Likewise.
1360 (vt_finalize): Likewise.
1362 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
1363 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
1364 (next_nonnote_nondebug_insn_bb): New.
1365 (prev_nonnote_nondebug_insn_bb): New.
1366 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
1367 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
1368 (prev_nonnote_nondebug_insn_bb): Declare.
1369 (next_nonnote_nondebug_insn_bb): Declare.
1370 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
1371 * cfgrtl.c (get_last_bb_insn): Likewise.
1372 * lra.c (push_insns): Likewise.
1374 2017-12-11 David Malcolm <dmalcolm@redhat.com>
1377 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
1379 * selftest.c (selftest::test_runner::test_runner): New ctor.
1380 (selftest::test_runner::~test_runner): New dtor.
1381 * selftest.h (class selftest::test_runner): New class.
1383 2017-12-11 Carl Love <cel@us.ibm.com>
1385 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
1386 vec_extract_fp32_from_shortl]): Add #defines.
1387 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
1388 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
1389 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
1390 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
1391 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
1392 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
1393 * doc/extend.texi: Add documentation for the added builtins.
1395 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
1397 PR rtl-optimization/80693
1398 PR rtl-optimization/81019
1399 PR rtl-optimization/81020
1400 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
1401 are not mentioned in i3. Place the REG_UNUSED note on i2,
1402 possibly modified to REG_DEAD, if it did not originate in i3.
1404 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1406 * recog.c (store_data_bypass_p_1): New function.
1407 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
1408 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
1410 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
1412 PR rtl-optimization/83361
1413 * ifcvt.c (if_convert): Call fixup_partitions.
1415 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
1417 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1418 early folding of splat_u{8,16,32}.
1420 2017-12-11 Jakub Jelinek <jakub@redhat.com>
1422 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
1423 output_operand_lossage first argument with capital letter.
1424 (aarch64_override_options): Don't start error and sorry first argument
1425 with capital letter.
1427 2017-12-11 Andi Kleen <ak@linux.intel.com>
1429 PR gcov-profile/83355
1430 * auto-profile.c (string_table::get_index_by_decl): Don't
1431 recurse when abstract origin points to itself.
1433 2017-12-11 Bin Cheng <bin.cheng@arm.com>
1435 PR tree-optimization/83320
1436 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
1437 (prune_datarefs_not_in_loop): Ditto.
1439 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
1441 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
1443 2017-12-10 Jakub Jelinek <jakub@redhat.com>
1445 PR tree-optimization/83337
1446 * gimple-loop-interchange.cc (compute_access_stride): Handle
1447 bitfield DRs properly.
1449 2017-12-09 Jakub Jelinek <jakub@redhat.com>
1451 PR tree-optimization/83338
1452 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
1453 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
1454 vector of unsigned integers to vector of signed integers.
1456 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
1458 PR rtl-optimization/83317
1459 * lra-constraints.c (process_address_1): Add insn code check.
1461 2017-12-08 Michael Matz <matz@suse.de>
1463 Fix PR tree-optimization/83323
1464 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
1465 head-controlled loops and loop BBs.
1466 * common.opt (funroll-and-jam): Remove, instead ...
1467 (floop-unroll-and-jam): ... reuse this option.
1468 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
1469 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
1470 (-floop-unroll-and-jam): ... this option.
1472 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1474 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
1475 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
1476 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
1478 2017-12-08 Richard Biener <rguenther@suse.de>
1481 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
1482 handle accesses outside of zero-sized vars.
1484 2017-12-08 Martin Jambor <mjambor@suse.cz>
1486 PR tree-optimization/83141
1487 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
1488 test for MEM_REFs implicitely changing types with padding. Remove
1490 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
1492 2017-12-08 Jakub Jelinek <jakub@redhat.com>
1494 * config/arc/arc.c (arc_attribute_table): Add exclusions to
1496 * config/avr/avr.c (avr_attribute_table): Likewise.
1497 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
1498 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
1499 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
1500 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
1501 of struct attribute_spec.
1502 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
1504 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
1507 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
1509 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
1512 * profile-count.c (profile_count::from_gcov_type): Move from
1513 profile-count.h; handle overflow.
1514 * profile-count.h (profile_count::from_gcov_type): Move offline.
1516 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
1518 PR rtl-optimization/83304
1519 * combine.c (move_deaths): If we do not know where a register died,
1522 2017-12-08 Richard Biener <rguenther@suse.de>
1524 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
1525 Provide -fopt-info-loop feedback when we interchange in a nest.
1527 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1529 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
1530 for armv6 ARM CPU IDs.
1532 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1534 * common/config/arm/arm-common.c: Include <algorithm>.
1535 (INCLUDE_VECTOR): Define.
1536 (compare_opt_names): New function.
1537 (arm_rewrite_selected_arch): Only strip out extensions that can be
1538 expressed through -mfpu. Sort the remaining extensions
1541 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
1543 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
1544 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
1545 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
1546 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
1548 2017-06-08 Tristan Gingold <gindold@adacore.com>
1551 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
1552 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
1554 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1556 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
1557 of struct attribute_spec.
1559 2017-12-08 Julia Koval <julia.koval@intel.com>
1561 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
1562 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
1563 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
1564 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
1565 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
1566 _mm_maskz_dpwssds_epi32): Ditto.
1568 2017-12-08 Richard Biener <rguenther@suse.de>
1570 PR tree-optimization/81303
1571 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
1572 conditions try to create a comparison vector type matching
1573 the data vector type.
1574 (vectorizable_condition): Adjust.
1575 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
1576 Leave invariant conditions alone in case we can vectorize those.
1578 2017-12-08 Julia Koval <julia.koval@intel.com>
1580 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
1581 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
1582 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
1583 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
1584 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
1586 2017-12-08 Julia Koval <julia.koval@intel.com>
1588 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
1589 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
1590 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
1591 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
1592 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
1593 _mm_maskz_dpbusds_epi32): New intrinsics.
1595 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1597 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
1599 (ld<bh>io_signed): Likewise.
1600 (st<bhw>io): Likewise.
1601 * config/nios2/predicates.md (ldstio_memory_operand): Allow
1602 SMALL_INT12 constant integer operand.
1604 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
1606 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
1608 (nios2_large_constant_allowed): ...this. Adjust uses.
1609 (nios2_plus_symbolic_constant_p): Rename to...
1610 (nios2_plus_large_constant_p): ...this. Adjust uses.
1611 (nios2_legitimate_address_p): Correct CONST_INT handling.
1612 (nios2_symbolic_memory_operand_p): Rename to...
1613 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
1614 (nios2_large_constant_p): Check for large constant integers too.
1615 (nios2_split_large_constant): Handle constant integers.
1616 (nios2_split_symbolic_memory_operand): Rename to...
1617 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
1618 (nios2_legitimize_constant_address): Handle constant integers.
1619 (r0rel_constant_p): Handle small constant integers.
1620 (nios2_print_operand_address): Handle r0-relative integer addresses.
1621 * config/nios2/nios2-protos.h: Adjust for renamed functions.
1622 * config/nios2/nios2.md: Adjust for renamed functions.
1624 2017-12-07 Andrew Waterman <andrew@sifive.com>
1626 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
1627 (TARGET_HAVE_SRODATA_SECTION): New define.
1628 (riscv_select_section): New function.
1630 2017-12-08 Joseph Myers <joseph@codesourcery.com>
1631 Alexander Monakov <amonakov@ispras.ru>
1632 Jakub Jelinek <jakub@redhat.com>
1635 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
1637 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1639 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
1640 a bool success value. Don't call output_operand_lossage here.
1641 (aarch64_print_ldpstp_address): Return a bool success value.
1642 (aarch64_print_operand_address): Call output_addr_const if
1643 aarch64_print_address_internal fails.
1644 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
1645 'y'; call output_operand_lossage instead. Call output_operand_lossage
1646 if aarch64_print_ldpstp_address fails.
1648 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1650 * tree-vector-builder.h
1651 (tree_vector_builder::binary_encoded_nelts): Declare.
1652 * tree-vector-builder.c
1653 (tree_vector_builder::binary_encoded_nelts): New function.
1654 * fold-const.c (negate_expr_p): Likewise.
1655 (operand_equal_p, fold_checksum_tree): Likewise.
1656 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
1657 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
1658 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
1659 (uniform_vector_p): Likewise.
1660 * varasm.c (const_hash_1, compare_constant): Likewise.
1661 * tree-ssa-ccp.c: Include tree-vector-builder.h.
1662 (valid_lattice_transition): Operate directly on the VECTOR_CST
1664 * ipa-icf.c: Include tree-vector-builder.h.
1665 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
1666 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
1668 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1670 * tree.c (build_vector): Delete.
1671 * tree.h (build_vector): Make static and move into the self-testing
1674 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1676 * vector-builder.h (vector_builder::derived): New const overload.
1677 (vector_builder::elt): New function.
1678 * tree-vector-builder.h (tree_vector_builder::type): New function.
1679 (tree_vector_builder::apply_step): Declare.
1680 * tree-vector-builder.c (tree_vector_builder::apply_step): New
1682 * gimple-fold.h (tree_vector_builder): Declare.
1683 (gimple_build_vector): Take a tree_vector_builder instead of a
1684 type and vector of elements.
1685 * gimple-fold.c (gimple_build_vector): Likewise.
1686 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
1688 (get_initial_defs_for_reduction): Likewise.
1689 (vectorizable_induction): Likewise.
1691 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1693 * tree-vector-builder.h
1694 (tree_vector_builder::new_binary_operation): Declare.
1695 * tree-vector-builder.c
1696 (tree_vector_builder::new_binary_operation): New function.
1697 * fold-const.c (fold_relational_const): Use it.
1698 (const_binop): Likewise. Check that both input vectors have
1699 the same number of elements, thus excluding things like WIDEN_SUM.
1700 Check whether it is possible to operate directly on the encodings
1703 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1705 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
1706 new_unary_operation, operating only on the encoded elements.
1707 (const_unop): Likewise.
1708 (exact_inverse): Likewise.
1709 (distributes_over_addition_p): New function.
1710 (const_binop): Use tree_vector_builder and new_unary_operation
1711 for combinations of VECTOR_CST and INTEGER_CST. Operate only
1712 on the encoded elements unless the encoding is strided and the
1713 operation does not distribute over addition.
1714 (fold_convert_const): Use tree_vector_builder and
1715 new_unary_operation. Operate only on the encoded elements
1716 for truncating integer conversions, or for non-stepped encodings.
1718 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1720 * config/sparc/sparc.c: Include tree-vector-builder.h.
1721 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
1722 * expmed.c: Include tree-vector-builder.h.
1723 (make_tree): Use tree_vector_builder instead of build_vector.
1724 * fold-const.c: Include tree-vector-builder.h.
1725 (const_binop): Use tree_vector_builder instead of build_vector.
1726 (const_unop): Likewise.
1727 (native_interpret_vector): Likewise.
1728 (fold_vec_perm): Likewise.
1729 (fold_ternary_loc): Likewise.
1730 * gimple-fold.c: Include tree-vector-builder.h.
1731 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
1733 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
1734 (simplify_vector_constructor): Use tree_vector_builder instead
1736 * tree-vect-generic.c: Include tree-vector-builder.h.
1737 (add_rshift): Use tree_vector_builder instead of build_vector.
1738 (expand_vector_divmod): Likewise.
1739 (optimize_vector_constructor): Likewise.
1740 * tree-vect-loop.c: Include tree-vector-builder.h.
1741 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
1742 of build_vector. Explicitly use a stepped encoding for
1744 * tree-vect-slp.c: Include tree-vector-builder.h.
1745 (vect_get_constant_vectors): Use tree_vector_builder instead
1747 (vect_transform_slp_perm_load): Likewise.
1748 (vect_schedule_slp_instance): Likewise.
1749 * tree-vect-stmts.c: Include tree-vector-builder.h.
1750 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
1751 (vect_gen_perm_mask_any): Likewise.
1752 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
1753 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
1755 (build_vector_from_val): Likewise. Explicitly use a duplicate
1758 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
1760 * doc/generic.texi (VECTOR_CST): Describe new representation of
1762 * vector-builder.h: New file.
1763 * tree-vector-builder.h: Likewise.
1764 * tree-vector-builder.c: Likewise.
1765 * Makefile.in (OBJS): Add tree-vector-builder.o.
1766 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
1767 * tree-core.h (tree_base): Add a vector_cst field to the u union.
1768 (tree_vector): Change the number of elements to
1769 vector_cst_encoded_nelts.
1770 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
1771 (VECTOR_CST_ELTS): Delete.
1772 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
1773 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
1774 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
1775 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
1776 (VECTOR_CST_ENCODED_ELT): Likewise.
1777 (vector_cst_encoded_nelts): New function.
1778 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1779 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1780 (vector_cst_int_elt, vector_cst_elt): Declare.
1781 * tree.c: Include tree-vector-builder.h.
1782 (tree_code_size): Abort if passed VECTOR_CST.
1783 (tree_size): Update for new VECTOR_CST layout.
1784 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
1785 VECTOR_CST_NELTS_PER_PATTERN as arguments.
1786 (build_vector): Use tree_vector_builder.
1787 (vector_cst_int_elt, vector_cst_elt): New functions.
1788 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
1789 encoded elements and then create the vector in the canonical form.
1790 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
1791 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
1792 (tree_c_tests): Call test_vector_cst_patterns.
1793 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
1795 (hash_tree): Likewise.
1796 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
1797 (streamer_write_tree_header): Likewise.
1798 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
1799 (streamer_alloc_tree): Likewise. Update call to make_vector.
1800 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
1802 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
1804 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
1805 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
1806 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
1808 2017-12-07 Bin Cheng <bin.cheng@arm.com>
1809 Richard Biener <rguenther@suse.de>
1811 PR tree-optimization/81303
1812 * Makefile.in (gimple-loop-interchange.o): New object file.
1813 * common.opt (floop-interchange): Reuse the option from graphite.
1814 * doc/invoke.texi (-floop-interchange): Ditto. New document for
1815 -floop-interchange and mention it for -O3.
1816 * opts.c (default_options_table): Enable -floop-interchange at -O3.
1817 * gimple-loop-interchange.cc: New file.
1818 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
1819 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
1820 * passes.def (pass_linterchange): New pass.
1821 * timevar.def (TV_LINTERCHANGE): New time var.
1822 * tree-pass.h (make_pass_linterchange): New declaration.
1823 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
1824 interchange. Record IV before/after increment in new parameters.
1825 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
1826 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
1828 (check_reduction_path): ...New function here.
1829 * tree-vectorizer.h (check_reduction_path): New declaration.
1831 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
1834 PR rtl-optimization/80818
1835 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
1836 always early clobbered.
1837 * lra-lives.c (process_bb_lives): Check input hard regs for early
1838 clobbered non-operand hard reg.
1840 2017-12-07 Jakub Jelinek <jakub@redhat.com>
1843 * tree-cfg.c (verify_gimple_assign_binary): Don't require
1844 types_compatible_p, just that TYPE_MODE is the same.
1846 2017-12-07 Martin Sebor <msebor@redhat.com>
1849 * attribs.c (empty_attribute_table): Initialize new member of
1850 struct attribute_spec.
1851 (decl_attributes): Add argument. Handle mutually exclusive
1852 combinations of attributes.
1853 (selftests::test_attribute_exclusions): New function.
1854 (selftests::attribute_c_tests): Ditto.
1855 * attribs.h (decl_attributes): Add default argument.
1856 * selftest.h (attribute_c_tests): Declare.
1857 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
1858 * tree-core.h (attribute_spec::exclusions, exclude): New type and
1860 * doc/extend.texi (Common Function Attributes): Update const and pure.
1861 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
1862 of struct attribute_spec.
1863 * config/arc/arc.c (arc_attribute_table): Same.
1864 * config/arm/arm.c (arm_attribute_table): Same.
1865 * config/avr/avr.c ( avr_attribute_table): Same.
1866 * config/bfin/bfin.c (bfin_attribute_table): Same.
1867 * config/cr16/cr16.c (cr16_attribute_table): Same.
1868 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
1869 * config/h8300/h8300.c (h8300_attribute_table): Same.
1870 * config/i386/i386.c (ix86_attribute_table): Same.
1871 * config/ia64/ia64.c (ia64_attribute_table): Same.
1872 * config/m32c/m32c.c (m32c_attribute_table): Same.
1873 * config/m32r/m32r.c (m32r_attribute_table): Same.
1874 * config/m68k/m68k.c (m68k_attribute_table): Same.
1875 * config/mcore/mcore.c (mcore_attribute_table): Same.
1876 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
1877 * config/mips/mips.c (mips_attribute_table): Same.
1878 * config/msp430/msp430.c (msp430_attribute_table): Same.
1879 * config/nds32/nds32.c (nds32_attribute_table): Same.
1880 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
1881 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
1882 * config/rl78/rl78.c (rl78__attribute_table): Same.
1883 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
1884 * onfig/rx/rx.c (rx_attribute_table): Same.
1885 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
1886 * config/sh/sh.c (sh_attribute_table): Same.
1887 * config/sparc/sparc.c (sparc_attribute_table): Same.
1888 * config/spu/spu.c (spu_attribute_table): Same.
1889 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
1890 * config/v850/v850.c (v850_attribute_table): Same.
1891 * config/visium/visium.c (visium_attribute_table): Same.
1893 2017-12-07 Tamar Christina <tamar.christina@arm.com>
1896 * config/arm/arm.c (INCLUDE_STRING): Define.
1897 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
1898 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
1901 2017-12-07 Michael Matz <matz@suse.de>
1903 Add unroll and jam pass
1905 * gimple-loop-jam.c: New file.
1906 * Makefile.in (OBJS): Add gimple-loop-jam.o.
1907 * common.opt (funroll-and-jam): New option.
1908 * opts.c (default_options_table): Add unroll-and-jam at -O3.
1909 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
1910 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
1911 * passes.def: Add pass_loop_jam.
1912 * timevar.def (TV_LOOP_JAM): Add.
1913 * tree-pass.h (make_pass_loop_jam): Declare.
1914 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
1915 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
1916 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
1917 to flow_loop_tree_node_add.
1918 (duplicate_subloops, copy_loops_to): Append to sibling list.
1919 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
1920 * doc/invoke.texi (-funroll-and-jam): Document new option.
1921 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
1923 2017-12-07 Richard Biener <rguenther@suse.de>
1925 PR tree-optimization/83296
1926 PR tree-optimization/67769
1927 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
1928 flow sensitive info in an unrelated BB.
1929 (value_replacement): Use reset_flow_sensitive_info.
1930 (minmax_replacement): Reset flow sensitive info on the def
1931 we move. Do not reset flow sensitive info in the whole BB
1932 we move the stmt to.
1933 (abs_replacement): Likewise.
1935 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1938 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
1939 rs6000_cpu to the given -mcpu=, or to the default processor.
1941 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1943 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
1944 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
1945 instead of rs6000_cpu_attr.
1946 (rs6000_adjust_cost): Ditto.
1947 (is_microcoded_insn): Ditto.
1948 (rs6000_adjust_priority): Ditto.
1949 (rs6000_issue_rate): Ditto.
1950 (rs6000_use_sched_lookahead): Ditto.
1951 (rs6000_use_sched_lookahead_guard): Ditto.
1952 (rs6000_sched_reorder): Ditto.
1953 (force_new_group): Ditto.
1954 * config/rs6000/rs6000.md (cpu attribute): Ditto.
1955 (group_ending_nop): Ditto.
1957 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
1959 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
1960 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
1961 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
1962 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
1964 (rs6000_reassociation_width): Ditto.
1965 (rs6000_emit_epilogue): Ditto.
1966 (rs6000_adjust_cost): Ditto.
1967 (is_microcoded_insn): Ditto.
1968 (is_cracked_insn): Ditto.
1969 (rs6000_adjust_priority): Ditto.
1970 (rs6000_sched_reorder): Ditto.
1971 (rs6000_sched_reorder2): Ditto.
1972 (insn_must_be_first_in_group): Ditto.
1973 (insn_must_be_last_in_group): Ditto.
1974 (rs6000_register_move_cost): Ditto.
1975 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
1978 2017-12-07 Julia Koval <julia.koval@intel.com>
1980 * config.gcc: Add vaesintrin.h.
1981 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
1983 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
1984 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
1986 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
1987 * config/i386/immintrin.h: Include vaesintrin.h.
1988 * config/i386/sse.md (vaesdec_<mode>): New pattern.
1989 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
1990 _mm_aesdec_epi128): New intrinsics.
1992 2017-12-06 David Malcolm <dmalcolm@redhat.com>
1994 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
1995 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
1996 c-family/c-spellcheck.cc.
1997 (best_macro_match::best_macro_match): Likewise.
1999 (struct edit_distance_traits<cpp_hashnode *>): Move to
2000 c-family/c-spellcheck.h.
2001 (class best_macro_match): Likewise.
2003 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2005 PR tree-optimization/83293
2006 * gimple-ssa-strength-reduction.c (insert_initializers): Use
2007 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
2008 might insert into empty bb.
2011 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
2012 simplify for plus with :c added, and pointer_plus without that.
2013 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
2014 with undefined overflow and the conversion is not widening,
2015 perform negation in utype and only convert to type afterwards.
2016 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
2017 simplify for plus with :c added, and pointer_plus without that.
2018 If type is integral with undefined overflow and the conversion is
2019 not widening, perform minus in utype and only convert to type
2020 afterwards. Move the last pointer_diff_expr simplify into the
2023 2017-12-06 Martin Sebor <msebor@redhat.com>
2025 PR tree-optimization/82646
2026 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
2027 strncpy, not maxlen.
2029 2017-12-06 Martin Sebor <msebor@redhat.com>
2031 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
2034 PR tree-optimization/83075
2035 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
2036 strncat/strncpy don't change length of source string.
2038 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
2041 2017-11-29 Martin Aberg <maberg@gaisler.com>
2043 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2044 to prevent b2bst errata sequence.
2045 (sqrtdf2_fix): Likewise.
2047 2017-12-06 Jakub Jelinek <jakub@redhat.com>
2049 PR tree-optimization/81945
2050 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
2051 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
2052 to dest_cfun has orig_loop_num set, either remap it to the new
2053 loop number if the loop got moved too, or clear it.
2055 2017-12-05 Steve Ellcey <sellcey@cavium.com>
2057 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
2059 (thunderx2t99_nothing): New insn reservation.
2060 (thunderx2t99_mrs): New insn reservation.
2061 (thunderx2t99_multiple): New insn reservation.
2062 (thunderx2t99_alu_basi): Add bfx to reservation.
2063 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
2065 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2068 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
2070 2017-12-05 Bin Cheng <bin.cheng@arm.com>
2072 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
2073 tree-ssa-pre.c::remove_dead_inserted_code.
2074 * tree-ssa-dce.h: New file.
2075 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
2076 (remove_dead_inserted_code): Move and rename to function
2077 tree-ssa-dce.c::simple_dce_from_worklist.
2078 (pass_pre::execute): Update use.
2080 2017-12-05 Richard Biener <rguenther@suse.de>
2082 PR tree-optimization/83277
2083 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
2084 to code-gen liveout vars.
2086 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
2088 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
2089 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
2090 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
2093 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
2095 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
2097 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
2098 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
2101 2017-12-05 Richard Biener <rguenther@suse.de>
2103 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
2104 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
2105 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
2106 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
2108 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2110 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
2112 2017-12-05 Martin Liska <mliska@suse.cz>
2113 Jakub Jelinek <jakub@redhat.com>
2115 * doc/invoke.texi: Document the options.
2116 * flag-types.h (enum sanitize_code): Add
2117 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2118 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
2119 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
2120 * opts.c: Define new sanitizer options.
2121 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
2122 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
2124 2017-12-05 Julia Koval <julia.koval@intel.com>
2126 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
2127 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
2128 (ix86_handle_option): Handle -mavx512vnni.
2129 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
2130 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
2131 * config/i386/i386-c (__AVX512VNNI__): New.
2132 * config/i386/i386.c (ix86_target_string): Handle new option.
2133 (ix86_valid_target_attribute_inner_p): Handle new option.
2134 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
2135 * config/i386/i386.opt (mavx512vnni): New option.
2137 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2140 * athlon.md: Disable for generic.
2141 * haswell.md: Enable for generic.
2142 * i386.c (ix86_sched_init_global): Add core hooks for generic.
2143 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
2145 (ix86_adjust_cost): Move generic to haswell path.
2147 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
2149 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
2150 instead of MEM_P in a couple more places. Fix formatting issues.
2152 2017-12-04 Jim Wilson <jimw@sifive.com>
2154 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
2155 instead of GP_REG_LAST-1.
2156 (riscv_adjust_libcall_cfi_prologue): Likewise.
2157 (riscv_adjust_libcall_cri_epilogue): Likewise.
2158 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
2161 2017-12-04 Luis Machado <luis.machado@linaro.org>
2163 * ipa-pure-const.c (check_decl): Add missing newline.
2164 (state_from_flags): Likewise.
2166 2017-12-04 Jeff Law <law@redhat.com>
2168 PR tree-optimizatin/78496
2169 * gimple-ssa-evrp-analyze.h
2170 (evrp_range_analyzer::get_vr_values): Simplify.
2171 * gimple-ssa-evrp-analyze.c: Corresponding changes.
2172 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
2173 and gimple-ssa-evrp-analyze.h.
2174 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
2175 (simplify_stmt_for_jump_threading): Copy a blob of code from
2176 tree-vrp.c to use ranges to simplify statements.
2177 (dom_opt_dom_walker::before_dom_children): Call
2178 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
2179 (dom_opt_dom_walker::after_dom_children): Similarly for
2180 evrp_range_analyzer::leave.
2181 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
2184 * gimple-ssa-evrp-analyze.c
2185 (evrp_range_analyzer::extract_range_from_stmt): Always use
2186 vr_values::update_value_range so preexisting range info is
2187 medged with new range info, even if the new range is VR_VARYING.
2189 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2191 * combine.c: Adjust comment.
2192 (use_crosses_set_p): Delete.
2193 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
2194 (try_combine): Ditto.
2196 2017-12-04 Richard Biener <rguenther@suse.de>
2198 PR tree-optimization/83255
2199 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
2200 Re-add zero-iteration check.
2202 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2204 PR rtl-optimization/83245
2205 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
2206 hard registers as earlyclobber, also if not in an asm.
2208 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
2212 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2215 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2216 rs6000_cpu based on cpu_index, not tune_index.
2218 2017-12-04 Richard Biener <rguenther@suse.de>
2220 PR tree-optimization/83238
2221 * graphite-scop-detection.c (scop_detection::merge_sese): Make
2222 code match comment, rejecting invalid SESE regions.
2224 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
2226 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
2227 require base operand is a REG_POINTER prior to reload on targets
2228 with non-equivalent space registers.
2230 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2232 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
2233 (update_specialized_profile): Fix updating of counts.
2234 (perhaps_add_new_callers): Likewise.
2236 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
2239 * x86-tune.def: Remove obsolette FIXMEs.
2240 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
2241 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
2242 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
2244 (X86_TUNE_PAD_RETURNS): Disable for generic.
2246 2017-12-02 Jakub Jelinek <jakub@redhat.com>
2248 PR tree-optimization/83170
2249 PR tree-optimization/83241
2250 * gimple-ssa-store-merging.c
2251 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
2252 gimple_vuse (ins_stmt) in case it has changed.
2253 (imm_store_chain_info::output_merged_store): Likewise.
2255 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
2259 * tree-cfg.c (pass_warn_function_return::execute): Handle
2260 __builtin_ubsan_handle_missing_return like __builtin_unreachable
2261 with BUILTINS_LOCATION.
2265 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
2266 is BLKmode for vector field with vector raw mode, use TYPE_MODE
2267 instead of DECL_MODE.
2269 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
2270 last argument to rtx pointer.
2271 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
2272 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
2273 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
2274 depending on the chosen ISAs.
2275 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
2276 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
2277 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
2279 * config/i386/sse.md (mov<mode>_internal): Likewise.
2280 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
2282 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2284 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
2286 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2288 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
2289 parameters from prototype.
2290 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
2291 parameters. Don't print a comment.
2292 (emit_fusion_gpr_load): Adjust.
2293 (emit_fusion_load_store): Adjust.
2294 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
2295 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
2296 comment on the second line.
2298 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2301 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
2302 rs6000_cpu based on cpu_index, not tune_index.
2304 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
2306 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
2307 which_alternative instead of which_alternative + 1.
2308 (output_asm_insn): Print an extra tab if the template is short.
2310 2017-12-01 Jim Wilson <jimw@sifive.com>
2312 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
2314 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
2315 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
2317 * doc/tm.texi: Regenerate.
2319 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2322 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
2323 whether we can allocate pseudos before trying to fix an address.
2324 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
2325 memory address is indexed or indirect.
2326 (floatuns_<mode>si2_hw2): Likewise.
2328 2017-12-01 Jason Merrill <jason@redhat.com>
2330 * Makefile.in (TAGS): Add c-family/*.cc.
2332 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
2334 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
2335 (call_value_insn): Likewise.
2336 (sibcall_insn): Likewise.
2337 (sibcall_value_insn): Likewise.
2338 (movsi_aarch64): Likewise.
2339 (movdi_aarch64): Likewise.
2340 (add_losym_): Likewise.
2341 (ldr_got_small_): Likewise.
2342 (ldr_got_small_sidi): Likewise.
2343 (ldr_got_small_28k_): Likewise.
2344 (ldr_got_small_28k_sidi): Likewise.
2345 * config/aarch64/aarch64.c (aarch64_print_address_internal):
2346 Move output_addr_const to symbolic case. Add error check.
2348 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2350 * config/s390/predicates.md (plus16_Q_operand): New predicate.
2351 * config/s390/s390.md: Disable MVC merging peephole if it would
2352 disable operand forwarding.
2353 (new peephole2): Split MVCs if it would turn them into up to 2
2356 2017-12-01 Richard Biener <rguenther@suse.de>
2358 PR tree-optimization/83232
2359 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
2360 detection of same access. Instead of breaking the group here
2361 do not consider the duplicate. Add comment explaining real fix.
2363 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
2365 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
2367 2017-12-01 Sudakshina Das <sudi.das@arm.com>
2369 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
2371 2017-12-01 Jakub Jelinek <jakub@redhat.com>
2373 * function.h (struct function): Remove cilk_frame_decl,
2374 is_cilk_function and calls_cilk_spawn fields.
2375 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
2377 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
2378 cilk_elemental field.
2379 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
2380 * target.def: Adjust comment.
2381 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
2382 Don't test cilk_elemental.
2384 PR tree-optimization/83233
2385 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
2386 bswap_stat name for the struct.
2389 * tree.h (SWITCH_BREAK_LABEL_P): Define.
2390 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
2391 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
2392 SWITCH_BREAK_LABEL_P set on the label.
2393 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
2394 added for default case if it was missing and not all cases covered.
2395 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
2396 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
2398 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
2399 to avoid -Wimplicit-fallthrough warning.
2400 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
2401 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
2403 PR tree-optimization/83221
2404 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
2406 (init_reassoc): Formatting fix.
2409 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
2410 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
2414 * tree-cfg.c: Include asan.h.
2415 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
2417 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2419 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
2421 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2423 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
2424 vpcmpeqd instruction.
2426 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
2428 * config/i386/i386.c (standard_sse_constant_opcode): Fix
2429 registers type for 128bit mode.
2431 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2433 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
2435 * gengtype-state.c (read_state_pair): Likewise.
2436 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
2437 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
2438 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
2439 * ifcvt.c (noce_try_store_flag_constants): Likewise.
2440 * tree-ssa-ccp.c (ccp_finalize): Likewise.
2441 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
2442 * builtins.c (fold_builtin_3): Likewise.
2443 * graphite-scop-detection.c
2444 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
2445 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
2448 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
2451 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
2452 parameter. Return true if ignore_decl_rtl_set_p is true and other
2453 conditions are satisfied.
2454 * asan.h (asan_protect_global): Add new parameter.
2455 * varasm.c (categorize_decl_for_section): Pass true as second parameter
2456 to asan_protect_global calls.
2458 2017-11-30 Jim Wilson <jimw@sifive.com>
2460 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
2461 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
2462 -mstrict-align, add info on default value. Delete redundant lines for
2463 -mabi. Add missing -mexplicit-relocs docs.
2465 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2467 * config/arc/arc.md (trap): New pattern.
2469 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2471 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
2472 instruction to end into a delay slot.
2473 * config/arc/arc.md (cond_delay_insn): Check if the instruction
2474 can be placed into a delay slot against reg_note.
2476 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2478 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
2479 labels number of usages.
2481 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
2483 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
2485 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
2487 2017-11-30 Jakub Jelinek <jakub@redhat.com>
2490 * internal-fn.c (expand_mul_overflow): Optimize unsigned
2491 multiplication by power of 2 constant into two shifts + comparison.
2493 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
2496 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
2498 2017-11-30 Richard Biener <rguenther@suse.de>
2500 PR tree-optimization/83202
2501 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
2502 allow_peel argument and guard peeling.
2503 (canonicalize_loop_induction_variables): Likewise.
2504 (canonicalize_induction_variables): Pass false.
2505 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
2506 peeling from cunrolli.
2508 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2510 * combine.c (try_combine): Print a message to dump file whenever
2511 I0, I1, or I2 cannot be combined into I3.
2513 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2515 PR rtl-optimization/83156
2516 PR rtl-optimization/82621
2517 * combine.c (try_combine): Don't split an I2 if one of the dests is
2518 set again before I3. Allow unused dests.
2520 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
2522 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
2524 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
2526 PR rtl-optimization/80818
2527 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
2528 recursively. Use insn code for clobber.
2529 (lra_set_insn_recog_data): Pass the new arg to
2530 collect_non_operand_hard_regs.
2531 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
2533 (lra_update_insn_regno_info): Pass insn to
2534 add_regs_to_insn_regno_info.
2536 2017-11-29 Jim Wilson <jimw@sifive.com>
2537 Andrew Waterman <andrew@sifive.com>
2539 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
2540 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
2541 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
2542 use SINGLE_SHIFT_COST.
2543 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
2544 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
2546 2017-11-29 Julia Koval <julia.koval@intel.com>
2548 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
2549 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
2550 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
2551 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
2552 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
2553 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
2554 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
2555 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
2556 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
2557 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
2558 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
2559 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
2560 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
2561 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
2562 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
2563 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
2564 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
2565 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
2566 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
2567 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
2568 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
2569 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
2570 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
2571 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
2572 __builtin_ia32_vpshldv_v2di_mask,
2573 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
2574 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
2575 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
2577 2017-11-29 Julia Koval <julia.koval@intel.com>
2579 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
2580 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
2581 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
2582 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
2583 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
2584 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
2585 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
2586 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
2587 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
2588 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
2589 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
2590 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
2591 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
2592 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
2593 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
2594 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
2595 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
2596 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
2597 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2598 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
2599 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
2601 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2603 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
2604 movsi_pic_gotdata_op instruction as a load for the UT699 errata
2607 2017-11-29 Martin Aberg <maberg@gaisler.com>
2609 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
2610 to prevent b2bst errata sequence.
2611 (sqrtdf2_fix): Likewise.
2613 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2615 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
2616 (div_sqrt_insn_p): New function.
2617 (sparc_do_work_around_errata): Insert NOP instructions to
2618 prevent sequences that could trigger the TN-0013 errata for
2619 certain LEON3 processors.
2620 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
2621 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
2622 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
2623 (in_branch_delay): Prevent div and sqrt in delay slot if
2625 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
2627 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2629 * config/sparc/sparc.c (atomic_insn_p): New function.
2630 (sparc_do_work_around_errata): Insert NOP instructions to
2631 prevent sequences that could trigger the TN-0010 errata for
2633 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
2634 instruction referable in atomic_insns_p.
2636 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2638 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
2639 (atomic_compare_and_swap_leon3_1): Likewise.
2642 2017-11-29 Daniel Cederman <cederman@gaisler.com>
2644 * config/sparc/sparc.c (fpop_insn_p): New function.
2645 (sparc_do_work_around_errata): Insert NOP instructions to
2646 prevent sequences that could trigger the TN-0012 errata for
2648 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
2649 * config/sparc/sparc.md (fix_gr712rc): New attribute.
2650 (in_branch_annul_delay): Prevent floating-point instructions
2651 in delay slot of annulled integer branch.
2653 2017-11-29 Richard Biener <rguenther@suse.de>
2655 PR tree-optimization/83202
2656 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
2658 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
2659 nodes vectorized to the same stmts multiple times.
2660 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
2661 (vect_analyze_slp_instance): Adjust.
2662 (scalar_stmts_to_slp_tree_map_t): New typedef.
2663 (vect_schedule_slp_instance): Add a map recording the SLP node
2664 representing the vectorized stmts for a set of scalar stmts.
2665 Avoid code-generating redundancies.
2666 (vect_schedule_slp): Allocate map and pass it down.
2668 2017-11-29 Nathan Sidwell <nathan@acm.org>
2671 * tree.c (build_complex_type): Fix canonicalization. Only fill in
2674 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
2676 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
2677 cases for printing LDP/STP memory addresses.
2678 (aarch64_print_address_internal): Renamed from
2679 aarch64_print_operand_address, added parameter, add Pmode check.
2680 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
2681 (aarch64_print_operand_address): Indirect to
2682 aarch64_print_address_internal.
2683 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
2684 'y' operand output specifier.
2686 2017-11-29 Jakub Jelinek <jakub@redhat.com>
2689 * tree.c (build_simple_mem_ref_loc): Handle
2690 get_addr_base_and_unit_offset returning a MEM_REF.
2693 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
2696 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
2699 2017-11-28 David Malcolm <dmalcolm@redhat.com>
2701 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
2703 (layout::move_to_column): Likewise.
2705 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2708 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
2709 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
2711 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2712 Martin Jambor <mjambor@suse.cz>
2715 * tree.h (expr_type_first_operand_type_p): Declare
2716 * tree.c (expr_type_first_operand_type_p): New function.
2717 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
2718 (ipa_value_from_jfunc): Adjust declaration.
2719 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
2720 Use it as result type for arithmetics, unless it is NULL in which case
2721 be more conservative.
2722 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
2723 ipa_get_jf_pass_through_result.
2724 (propagate_vals_across_pass_through): Likewise.
2725 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
2726 is to propagate_vals_across_pass_through.
2727 (propagate_constants_across_call): Pass PARM_TYPE to
2728 propagate_scalar_across_jump_function.
2729 (find_more_scalar_values_for_callers_subset): Pass parameter type to
2730 ipa_value_from_jfunc.
2731 (cgraph_edge_brings_all_scalars_for_node): Likewise.
2732 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
2733 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
2734 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
2735 target_type, pass it to ipa_value_from_jfunc.
2736 (update_indirect_edges_after_inlining): Pass parameter type to
2737 try_make_edge_direct_simple_call.
2739 2017-11-28 Jeff Law <law@redhat.com>
2741 * gimple-ssa-evrp-analyze.c
2742 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
2743 refine ranges if scev_initialized_p returns true.
2744 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
2746 2017-11-28 Julia Koval <julia.koval@intel.com>
2748 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
2749 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
2750 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
2751 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
2752 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
2753 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
2754 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
2755 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
2756 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
2757 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
2758 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
2759 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
2760 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
2761 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
2762 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
2763 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
2764 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
2765 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
2766 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
2767 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
2768 __builtin_ia32_vpshrd_v2di_mask): New builtins.
2769 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
2771 2017-11-28 Julia Koval <julia.koval@intel.com>
2773 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
2774 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
2775 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
2776 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
2777 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
2778 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
2779 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
2780 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
2781 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
2782 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
2783 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
2784 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
2785 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
2786 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
2787 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
2788 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
2789 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
2790 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
2791 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
2792 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
2793 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
2794 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
2795 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
2796 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
2797 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
2798 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
2799 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
2800 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
2801 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
2802 __builtin_ia32_vpshld_v2di_mask): New builtins.
2803 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
2804 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
2806 2017-11-28 Richard Biener <rguenther@suse.de>
2808 PR tree-optimization/80776
2809 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
2811 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
2813 (evrp_range_analyzer::record_ranges_from_incoming_edges):
2814 If the incoming edge is an effective fallthru because the other
2815 edge only reaches a __builtin_unreachable () then record ranges
2816 derived from the controlling condition in SSA info.
2817 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
2818 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
2820 2017-11-28 Olivier Hainque <hainque@adacore.com>
2822 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
2823 on mingw build hosts.
2825 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2828 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
2830 2017-11-28 Richard Biener <rguenther@suse.de>
2833 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
2834 copies generated from memcpy use a character array as reference
2837 2017-11-28 Julia Koval <julia.koval@intel.com>
2838 Sebastian Peryt <sebastian.peryt@intel.com>
2840 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
2841 c-family/c-cilkplus.o, c-family/array-notation-common.o,
2842 cilk-common.o, cilk.h, cilk-common.c): Remove.
2844 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
2845 * builtins.c (is_builtin_name): Remove cilkplus condition.
2846 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
2847 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
2848 cilk-builtins.def, cilkplus.def): Remove.
2849 * cif-code.def (CILK_SPAWN): Remove.
2850 * cilk-builtins.def: Delete.
2851 * cilk-common.c: Ditto.
2853 * cilkplus.def: Ditto.
2854 * config/darwin.h (fcilkplus): Delete.
2855 * cppbuiltin.c: Ditto.
2856 * doc/extend.texi: Remove cilkplus doc.
2857 * doc/generic.texi: Ditto.
2858 * doc/invoke.texi: Ditto.
2859 * doc/passes.texi: Ditto.
2860 * gcc.c (fcilkplus): Remove.
2861 * gengtype.c (cilk.h): Remove.
2862 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
2864 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
2866 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
2868 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
2869 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
2870 cilkplus conditions.
2871 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
2872 inline_read_section): Ditto.
2873 * ipa-inline-analysis.c (cilk.h): Remove.
2874 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
2875 * lto-wrapper.c (merge_and_complain, append_compiler_options,
2876 append_linker_options): Remove condition for fcilkplus.
2877 * lto/lto-lang.c (cilk.h): Remove.
2878 (lto_init): Remove condition for fcilkplus.
2879 * omp-expand.c (expand_cilk_for_call): Delete.
2880 (expand_omp_taskreg, expand_omp_for_static_chunk,
2881 expand_omp_for): Remove cilkplus
2883 (expand_cilk_for): Delete.
2884 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
2885 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
2886 execute_lower_omp, diagnose_sb_0): Ditto.
2887 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
2888 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2889 * tree-nested.c: Ditto.
2890 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
2891 (dump_generic_node): Ditto.
2892 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
2893 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
2895 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
2897 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2899 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
2900 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
2901 for vector mode and !TARGET_SIMD.
2903 2017-11-28 Jakub Jelinek <jakub@redhat.com>
2905 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
2907 * tree.h (SWITCH_LABELS): Remove.
2908 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
2909 assert SWITCH_BODY is non-NULL.
2910 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
2912 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
2914 PR tree-optimization/80788
2915 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
2916 has TREE_OVERFLOW set, call drop_tree_overflow.
2918 2017-11-28 Richard Biener <rguenther@suse.de>
2920 PR tree-optimization/83158
2921 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
2923 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
2926 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
2927 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
2929 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
2931 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
2932 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
2934 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2937 * varasm.c (compare_constant): Take the mode of the constants into
2938 account when comparing floating point constants.
2940 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
2942 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
2943 from explicit instantiation of debug_helper.
2944 * vec.h (DEFINE_DEBUG_VEC): Ditto.
2946 2017-11-27 Richard Biener <rguenther@suse.de>
2948 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
2951 2017-11-27 Richard Biener <rguenther@suse.de>
2953 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
2954 (type_hash_canon): Also clear int_cst_hash_table entry for
2956 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
2958 2017-11-27 Tamar Christina <tamar.christina@arm.com>
2960 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2961 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2962 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
2964 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
2966 * hash-map.h (gt_cleare_cache): Avoid UB.
2968 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
2970 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
2971 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
2972 (decide_unroll_runtime_iterations): Likewise.
2973 (decide_unroll_stupid): Likewise.
2975 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2978 * config/i386/i386.md: Add a loop with incssp.
2980 2017-11-27 Martin Jambor <mjambor@suse.cz>
2982 PR tree-optimization/81248
2983 * tree-sra.c (splice_param_accesses): Remove size check.
2984 (decide_one_param_reduction): Fix size check.
2985 * gimple-pretty-print.c (dump_profile): Silence warning.
2986 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
2988 2017-11-27 Jakub Jelinek <jakub@redhat.com>
2991 * dbxout.c (lastlineno): New variable.
2992 (dbx_debug_hooks): Use dbxout_switch_text_section as
2993 switch_text_section debug hook.
2994 (dbxout_function_end): Switch to current_function_section
2995 rather than function_section. If crtl->has_bb_partition,
2996 output just one N_FUN, depending on in_cold_section_p.
2997 (dbxout_source_line): Remember last lineno in lastlineno.
2998 (dbxout_switch_text_section): New function.
2999 (dbxout_function_decl): Adjust dbxout_block caller.
3000 (dbx_block_with_cold_children): New function.
3001 (dbxout_block): Return true if any LBRAC/RBRAC have been
3002 emitted. Use dbx_block_with_cold_children at depth == 0
3003 in second partition. Add PARENT_BLOCKNUM argument, pass
3004 it optionally adjusted to children. Output LBRAC/RBRAC
3005 around recursive call only if the block is in the current
3006 partition, if not and anything was output, emit empty
3008 * final.c (final_scan_insn): Compute cold_function_name
3009 before calling switch_text_section debug hook. Call
3010 that hook even if dwarf2out_do_frame if not emitting
3014 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
3015 TREE_READONLY decls.
3017 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
3019 PR rtl-optimization/82488
3020 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
3022 2017-11-26 Julia Koval <julia.koval@intel.com>
3024 * config/i386/i386.c (processor_target_table): Add skylake_cost for
3026 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
3029 2017-11-26 Julia Koval <julia.koval@intel.com>
3031 * config/i386/driver-i386.c (host_detect_local_cpu):
3032 Detect skylake-avx512.
3034 2017-11-26 Julia Koval <julia.koval@intel.com>
3036 * config.gcc: Add -march=cannonlake.
3037 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
3038 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
3039 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
3040 (PTA_CANNONLAKE): New.
3041 (processor_target_table): Add cannonlake.
3042 (ix86_option_override_internal): Ditto.
3043 (fold_builtin_cpu): Ditto.
3044 (get_builtin_code_for_version): Handle cannonlake.
3045 (M_INTEL_COREI7_CANNONLAKE): New.
3046 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
3047 * doc/invoke.texi: Add -march=cannonlake.
3049 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
3051 * plugin.c (add_new_plugin): Use platform-specific library extensions.
3052 (try_init_one_plugin): Alternative implementation for MinGW.
3053 * Makefile.in (plugin_implib): New.
3054 (gengtype-lex.c): Fix broken AIX workaround.
3055 * configure: Regenerate.
3056 * doc/plugins.texi: Document support for MinGW.
3058 2017-11-25 Jakub Jelinek <jakub@redhat.com>
3060 PR rtl-optimization/81553
3061 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
3062 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
3063 is a shift where C1 has different mode than the whole shift, use C1's
3064 mode for MULT rather than the shift's mode.
3067 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
3068 builtins not enabled in the currently selected ISA.
3070 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
3072 PR tree-optimization/71026
3073 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
3074 (insert_reciprocals): Change to insert reciprocals before a division
3075 by a square and to insert the square of a reciprocal.
3076 (execute_cse_reciprocals_1): Change to consider division by a square.
3077 (register_division_in): Add importance parameter.
3079 2017-11-24 Richard Biener <rguenther@suse.de>
3081 PR tree-optimization/82402
3082 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
3083 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
3085 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
3087 * match.pd (0-ptr): New transformation.
3089 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
3092 * ipa-inline.c (inline_small_functions): Set current badnes correctly
3093 when skipping checking.
3095 2017-11-24 Richard Biener <rguenther@suse.de>
3097 PR tree-optimization/83128
3098 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
3099 (vn_reference_lookup_3): Likewise.
3101 2017-11-24 Jakub Jelinek <jakub@redhat.com>
3104 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
3105 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
3108 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
3109 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
3110 check for the latter with gimple_call_builtin_p. Do not handle
3111 BUILT_IN_STPNCPY_CHK which is not a pass through call.
3113 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
3115 * config/arm/arm_neon.h: Fix pragma GCC push_options before
3118 2017-11-23 Julia Koval <julia.koval@intel.com>
3120 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
3121 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
3122 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
3123 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
3124 _mm512_maskz_expandloadu_epi16): New intrinsics.
3125 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
3126 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
3127 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
3128 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
3129 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
3130 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
3131 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
3132 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
3133 _mm256_maskz_expandloadu_epi8): New intrinsics.
3134 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
3135 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
3136 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
3137 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
3138 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
3139 * config/i386/sse.md (VI248_VLBW): New iterator.
3140 (expand<mode>_mask, expand<mode>_maskz): New patterns.
3142 2017-11-23 Julia Koval <julia.koval@intel.com>
3144 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
3145 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
3146 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
3147 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
3148 _mm512_mask_compressstoreu_epi16): New.
3149 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
3150 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
3151 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
3152 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
3153 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
3154 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
3155 _mm256_mask_compressstoreu_epi8): New.
3156 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
3157 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
3158 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
3159 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
3160 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
3161 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
3162 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
3163 __builtin_ia32_compresshi128_mask,
3164 __builtin_ia32_compressstoreuqi512_mask,
3165 __builtin_ia32_compressstoreuhi512_mask,
3166 __builtin_ia32_compressstoreuqi256_mask,
3167 __builtin_ia32_compressstoreuqi128_mask,
3168 __builtin_ia32_compressstoreuhi256_mask,
3169 __builtin_ia32_compressstoreuhi128_mask): New builtins.
3170 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
3172 (ix86_expand_special_args_builtin): Handle new types.
3173 (s4fma_expand): Handle new builtin array.
3174 * config/i386/immintrin.h: Include new headers.
3175 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
3176 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
3178 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3180 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
3181 fixes. Declare temp and g variables at the top in order to avoid
3182 {} in most of the cases.
3184 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
3186 * match.pd (ptr-0): New transformation.
3188 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3190 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
3191 (struct addr_mode_cost_table): New.
3192 (struct tune_params): Add field addr_mode_costs.
3193 * config/arm/arm.c (generic_addr_mode_costs): New.
3194 (arm_slowmul_tune): Initialise addr_mode_costs field.
3195 (arm_fastmul_tune): Likewise.
3196 (arm_strongarm_tune): Likewise.
3197 (arm_xscale_tune): Likewise.
3198 (arm_9e_tune): Likewise.
3199 (arm_marvell_pj4_tune): Likewise.
3200 (arm_v6t2_tune): Likewise.
3201 (arm_cortex_tune): Likewise.
3202 (arm_cortex_a8_tune): Likewise.
3203 (arm_cortex_a7_tune): Likewise.
3204 (arm_cortex_a15_tune): Likewise.
3205 (arm_cortex_a35_tune): Likewise.
3206 (arm_cortex_a53_tune): Likewise.
3207 (arm_cortex_a57_tune): Likewise.
3208 (arm_exynosm1_tune): Likewise.
3209 (arm_xgene1_tune): Likewise.
3210 (arm_cortex_a5_tune): Likewise.
3211 (arm_cortex_a9_tune): Likewise.
3212 (arm_cortex_a12_tune): Likewise.
3213 (arm_cortex_a73_tune): Likewise.
3214 (arm_v7m_tune): Likewise.
3215 (arm_cortex_m7_tune): Likewise.
3216 (arm_v6m_tune): Likewise.
3217 (arm_fa726te_tune): Likewise.
3218 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
3220 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
3222 * config/arm/arm.c (arm_mem_costs): New function.
3223 (arm_rtx_costs_internal): Use arm_mem_costs.
3225 2017-11-23 Mark Wielaard <mark@klomp.org>
3227 * dwarf2out.c (init_sections_and_labels): Use generation to create
3228 unique ranges_section_label and ranges_base_label. Return generation.
3229 (output_rnglists): Add generation argument. Use generation to create
3230 unique ranges labels.
3231 (dwarf2out_finish): Get generation from init_sections_and_labels
3232 and pass generation to output_rnglists.
3234 2017-11-23 Mike Stump <mikestump@comcast.net>
3235 Eric Botcazou <ebotcazou@adacore.com>
3237 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
3238 * cfgloop.h (struct loop): Add unroll field.
3239 * function.h (struct function): Add has_unroll bitfield.
3240 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
3241 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
3242 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
3243 (pass_rtl_unroll_loops::gate): Likewise.
3244 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
3245 for which loop->unroll==1.
3246 (decide_unroll_constant_iterations): Use note for consistency and
3247 take loop->unroll into account. Return early if loop->unroll is set.
3248 Fix thinko in existing test.
3249 (decide_unroll_runtime_iterations): Use note for consistency and
3250 take loop->unroll into account.
3251 (decide_unroll_stupid): Likewise.
3252 * lto-streamer-in.c (input_cfg): Read loop->unroll.
3253 * lto-streamer-out.c (output_cfg): Write loop->unroll.
3254 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
3256 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
3257 (print_loop): Print loop->unroll if set.
3258 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
3259 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
3260 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
3262 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
3263 loop->unroll is set and smaller than the trip count. Otherwise bypass
3264 entirely the heuristics if loop->unroll is set. Remove dead note.
3265 Fix off-by-one bug in other note.
3266 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
3267 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
3269 (tree_unroll_loops_completely): Make static.
3270 (pass_complete_unroll::execute): Use correct type for variable.
3271 (pass_complete_unrolli::execute): Fix formatting.
3272 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
3274 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
3276 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
3277 enable when TARGET_PREFER_AVX128 is set.
3279 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3281 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
3283 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
3284 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
3286 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3288 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
3289 frequencies. Fix estimation of aggregate parameters.
3291 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
3293 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
3296 2017-11-23 Tom de Vries <tom@codesourcery.com>
3298 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
3300 2017-11-23 Tom de Vries <tom@codesourcery.com>
3302 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
3305 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
3308 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
3309 sibcall_value_pcrel_fdpic): Use local variable instead of
3311 (calli_tbr_rel): Add missing operand 2.
3312 (call_valuei_tbr_rel): Add missing operand 3.
3314 2017-11-23 Jakub Jelinek <jakub@redhat.com>
3317 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
3318 bitpos/bitsize covering the whole destination, use store_expr only if
3319 the complex mode is the same. Otherwise, use expand_normal and if
3320 it returns CONCAT, subreg each part separately instead of trying to
3321 subreg the whole result.
3323 2017-11-23 Richard Biener <rguenther@suse.de>
3325 PR tree-optimization/23094
3326 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
3327 come here from walking over backedges in the first iteration.
3328 (vn_reference_lookup_3): Skip clobbers that store the same value.
3330 2017-11-23 Richard Biener <rguenther@suse.de>
3332 PR tree-optimization/81403
3333 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
3334 a block we need a leader relative to.
3335 (phi_translate_1): For nary processing require a leader from
3336 get_representative_for given we run expression simplification
3337 using match-and-simplify. Remove previous fix.
3339 2017-11-22 Jeff Law <law@redhat.com>
3341 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
3342 Use new method allocate_value_range rather than accessing the
3343 vrp_value_range_pool data member directly.
3344 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
3345 to use extract_range_from_stmt method to avoid need for
3346 extract_range_from_assignment method.
3347 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
3348 method rather than setting values_propgated data member directly.
3349 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
3350 and values propagated data members and extract_range_from_assignment
3351 method. Reorder private data members to conform to standards.
3352 Add new methods set_lattice_propagation_complete and
3353 allocate_value_range.
3355 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
3357 PR rtl-optimization/83030
3358 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
3359 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
3360 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
3361 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
3362 CROSSING_JUMP_P flag is already set before setting it.
3363 * cfgrtl.c (fixup_partition_crossing): Likewise.
3364 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
3367 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3369 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
3370 Handle the case where both arguments are using gen_const_vec_series.
3372 2017-11-22 David Malcolm <dmalcolm@redhat.com>
3375 * pretty-print.c (pp_format): Move quoting implementation to
3376 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
3377 to pass address of "quote" local.
3378 (pp_begin_quote): New function.
3379 (pp_end_quote): New function.
3380 * pretty-print.h (printer_fn): Convert penultimate param from bool
3382 (pp_begin_quote): New decl.
3383 (pp_end_quote): New decl.
3384 * tree-diagnostic.c (default_tree_printer): Convert penultimate
3385 param from bool to bool *.
3386 * tree-diagnostic.h (default_tree_printer): Likewise.
3388 2017-11-22 Jeff Law <law@redhat.com>
3390 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
3391 Initialize vr_values.
3392 (evrp_range_analyzer::try_find_new_range): Call methods attached to
3393 vr_values via vr_values class instance rather than delegators.
3394 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
3395 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
3396 (evrp_range_analyzer::push_value_range): Likewise.
3397 (evrp_range_analyzer::pop_value_range): Likewise.
3398 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
3399 most delegators. Those remaining are exposed as public interfaces.
3400 Make vr_values a pointer and private.
3401 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
3403 (evrp_range_analyzer::get_vr_value): New method.
3404 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
3405 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
3406 (class evrp_dom_walker): Attach evrp_folder class, initialize
3407 it in the ctor. Remove temporary delegators.
3408 (evrp_dom_walker::before_dom_children): Call methods in attached
3409 evrp_range_analyzer class via class instance pointer. Use
3410 free value_range_constant_singleton to remove need for
3411 op_with_constant_singleton_value delegator method. Do not
3412 create a vrp_prop class instance for every call! Narrow
3413 scope of a couple variables.
3414 (evrp_dom_walker::cleanup): Call methods in attached
3415 evrp_range_analyzer class via class instance pointer.
3416 * vr-values.h (class vr_values): Privatize many methods and
3419 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3421 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
3422 * cfgexpand.c (expand_debug_expr): Remove handling for them.
3423 * expr.c (expand_expr_real_2): Likewise.
3424 * fold-const.c (const_unop): Likewise.
3425 * optabs-tree.c (optab_for_tree_code): Likewise.
3426 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3427 * tree-inline.c (estimate_operator_cost): Likewise.
3428 * tree-pretty-print.c (dump_generic_node): Likewise.
3429 (op_code_prio): Likewise.
3430 (op_symbol_code): Likewise.
3431 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
3432 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
3433 * internal-fn.c (direct_internal_fn_optab): New function.
3434 (direct_internal_fn_array, direct_internal_fn_supported_p
3435 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
3436 * fold-const-call.c (fold_const_reduction): New function.
3437 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
3439 * tree-vect-loop.c: Include internal-fn.h.
3440 (reduction_code_for_scalar_code): Rename to...
3441 (reduction_fn_for_scalar_code): ...this and return an internal
3443 (vect_model_reduction_cost): Take an internal_fn rather than
3445 (vect_create_epilog_for_reduction): Likewise. Build calls rather
3447 (vectorizable_reduction): Use internal functions rather than tree
3448 codes for the reduction operation. Update calls to the functions
3450 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
3451 Use calls to internal functions rather than REDUC tree codes.
3452 * config/aarch64/aarch64-simd.md: Update comment accordingly.
3454 2017-11-22 Olivier Hainque <hainque@adacore.com>
3456 * config/vxworks.c (vxworks_override_options): Pick default
3457 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
3458 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
3459 DWARF_GNAT_ENCODINGS_DEFAULT.
3460 * config/vxworksae.h: Likewise.
3462 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
3464 PR tree-optimization/83104
3465 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
3468 2017-11-22 Marek Polacek <polacek@redhat.com>
3469 H.J. Lu <hongjiu.lu@intel.com>
3470 Jason Merrill <jason@redhat.com>
3475 * calls.c (initialize_argument_information): Call
3476 warn_parameter_passing_abi target hook.
3477 (store_one_arg): Use 0 for empty record size. Don't push 0 size
3478 argument onto stack.
3479 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
3480 * common.opt: Update -fabi-version description.
3481 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
3482 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
3484 (ix86_is_empty_record): New function.
3485 (ix86_warn_parameter_passing_abi): New function.
3486 (TARGET_EMPTY_RECORD_P): Redefine.
3487 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
3488 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
3489 * doc/tm.texi: Regenerated.
3490 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
3491 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
3492 * dwarf2out.c (get_ultimate_context): Move to tree.c.
3493 * explow.c (hard_function_value): Call arg_int_size_in_bytes
3494 instead of int_size_in_bytes.
3495 * expr.c (copy_blkmode_to_reg): Likewise.
3496 * function.c (aggregate_value_p): Return 0 for empty types.
3497 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
3499 (locate_and_pad_parm): Call arg size_in_bytes instead
3501 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
3502 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
3503 * target.def (empty_record_p, warn_parameter_passing_abi): New target
3505 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
3506 (std_gimplify_va_arg_expr): Skip empty records. Call
3507 arg_size_in_bytes instead size_in_bytes.
3508 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
3509 * tree-core.h (tree_type_common): Add empty_flag.
3510 (tree_decl_common): Update comments.
3511 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
3513 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3514 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
3516 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
3517 * tree.c (default_is_empty_type): New function.
3518 (default_is_empty_record): New function.
3519 (arg_int_size_in_bytes): New function.
3520 (arg_size_in_bytes): New function.
3521 (get_ultimate_context): New function.
3522 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
3523 TRANSLATION_UNIT_WARN_EMPTY_P.
3524 (default_is_empty_record, arg_int_size_in_bytes,
3525 arg_size_in_bytes, get_ultimate_context): Declare.
3527 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3529 * config/arm/arm.c (cmse_clear_registers): New function.
3530 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
3531 code by call to cmse_clear_registers.
3532 (cmse_nonsecure_entry_clear_before_return): Likewise.
3534 2017-11-22 Tamar Christina <tamar.christina@arm.com>
3536 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
3537 (vdot_s32, vdotq_s32): New.
3538 (vdot_lane_u32, vdotq_lane_u32): New.
3539 (vdot_lane_s32, vdotq_lane_s32): New.
3542 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3545 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
3546 for unsigned values with fewer HWIs than the precision.
3547 (test_overflow): New function.
3548 (wide_int_cc_tests): Call it.
3550 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
3552 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
3553 has the same mode class as Pmode.
3555 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3557 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
3558 padding_bits_to_clear_ptr.
3559 (cmse_nonsecure_entry_clear_before_return): Likewise.
3561 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3563 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
3564 auto_sbitap instead of integer bitfield to control register needing
3567 2017-11-22 Jakub Jelinek <jakub@redhat.com>
3569 PR tree-optimization/83044
3570 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
3571 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
3572 that need the upper bound. Subtract offset from
3573 get_addr_base_and_unit_offset only if positive and subtract it
3574 before division by eltsize rather than after it.
3577 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
3578 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
3579 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
3583 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
3584 expand_binop with *mul_widen_optab, make sure at least one of the
3585 operands doesn't have VOIDmode.
3588 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
3590 PR rtl-optimization/82044
3591 PR tree-optimization/82042
3592 * dse.c (record_store): Check for overflow.
3593 (check_mem_read_rtx): Properly check for overflow if width == -1, call
3594 add_wild_read instead of clear_rhs_from_active_local_stores on
3595 overflow and log it into dump_file.
3597 2017-11-22 Richard Biener <rguenther@suse.de>
3599 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
3600 fake edges to exit when looking for a place to insert.
3601 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
3603 (insert_into_preds_of_block): Commit edge insertion immediately,
3604 assert that doesn't require new BBs.
3605 (fini_pre): Release expressions.
3606 (pass_pre::execute): Shuffle things around a bit, if the fn
3607 is too large do not compute AVAIL either as this is really the
3610 2017-11-22 Richard Biener <rguenther@suse.de>
3612 PR tree-optimization/83089
3613 * tree-if-conv.c (pass_if_conversion::execute): If anything
3614 changed reset SCEV and free the number of iteration estimates.
3616 2017-11-21 Martin Sebor <msebor@redhat.com>
3618 PR tree-optimization/82945
3619 * calls.h (warn_nonstring_bound): Remove unused function.
3621 2017-11-21 Martin Sebor <msebor@redhat.com>
3623 PR tree-optimization/82945
3624 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
3625 * calls.h (maybe_warn_nonstring_arg): Declare new function.
3626 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
3628 (initialize_argument_information): Call maybe_warn_nonstring_arg.
3629 * calls.h (get_attr_nonstring_decl): Declare new function.
3630 * doc/extend.texi (attribute nonstring): Update.
3631 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
3632 get_attr_nonstring_decl and handle it.
3633 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
3634 detection of nul-termination.
3635 (strlen_to_stridx): Change to a pointer.
3636 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
3637 (pass_strlen::execute): Same.
3639 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
3641 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
3642 for the new option -mprefer-vector-width=[none|128|256|512].
3643 * config/i386/i386.c (ix86_target_string): remove old style options
3644 -mprefer-avx256 and make -mprefer-avx128 as alias.
3645 (ix86_option_override_internal): Apply defaults for the
3646 -mprefer-vector-width=[128|256] option.
3647 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
3648 Implement macros to work with -mprefer-vector-width=.
3649 * config/i386/i386.opt: Implemented option
3650 -mprefer-vector-width=[none|128|256|512].
3651 * doc/invoke.texi: Documentation for
3652 -mprefer-vector-width=[none|128|256|512].
3654 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
3656 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
3658 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
3660 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
3662 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
3664 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
3666 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
3667 * expr.c (expand_expr_real_2): Likewise.
3668 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
3669 fold_binary_loc): Likewise.
3670 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
3671 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
3672 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
3673 MINUS_EXPR transformations.
3674 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
3675 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
3676 * tree-inline.c (estimate_operator_cost): Likewise.
3677 * tree-pretty-print.c (dump_generic_node, op_code_prio,
3678 op_symbol_code): Likewise.
3679 * tree-vect-stmts.c (vectorizable_operation): Likewise.
3680 * vr-values.c (extract_range_from_binary_expr): Likewise.
3681 * varasm.c (initializer_constant_valid_p_1): Likewise.
3682 * tree.def: New tree code POINTER_DIFF_EXPR.
3684 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
3686 * config/i386/i386.md (*bswap<mode>2_movbe): Add
3687 integer suffix to movbe mnemonic.
3688 (*bswaphi2_movbe): Ditto.
3689 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
3691 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3694 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
3695 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
3696 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
3698 2017-11-21 Martin Liska <mliska@suse.cz>
3700 * tree-inline.c (expand_call_inline): Remove not needed
3703 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
3704 Jakub Jelinek <jakub@redhat.com>
3707 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
3708 Remove static keyword from f variable.
3710 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3712 PR tree-optimization/83086
3713 * gimple-ssa-store-merging.c
3714 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
3715 rather than n.base_addr.
3717 2017-11-21 Martin Liska <mliska@suse.cz>
3719 PR rtl-optimization/82044
3720 PR tree-optimization/82042
3721 * dse.c (check_mem_read_rtx): Check for overflow.
3723 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3725 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
3728 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3730 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
3731 and its corresponding call arg location note.
3733 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
3734 Andrew Burgess <andrew.burgess@embecosm.com>
3736 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
3738 (arc_return_slot_offset): Likewise.
3739 (arc_eh_return_address_location): New declaration.
3740 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
3741 (MUST_SAVE_REGISTER): Add exception handler case.
3742 (MUST_SAVE_RETURN_ADDR): Likewise.
3743 (arc_frame_pointer_required): Likewise.
3744 (arc_frame_pointer_needed): New function.
3745 (arc_compute_frame_size): Changed.
3746 (arc_expand_prologue): Likewise.
3747 (arc_expand_epilogue): Likewise.
3748 (arc_initial_elimination_offset): Likewise.
3749 (arc_return_slot_offset): Delete.
3750 (arc_eh_return_address_location): New function.
3751 (arc_builtin_setjmp_frame_value): Likewise.
3752 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
3753 (EH_RETURN_STACKADJ_RTX): Define.
3754 (EH_RETURN_HANDLER_RTX): Likewise.
3755 * config/arc/arc.md (eh_return): Delete.
3757 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
3759 * print-tree.h (debug_vec_tree): Remove prototype.
3760 * gdbinit.in (pvt): Remove macro.
3762 2017-11-21 Jakub Jelinek <jakub@redhat.com>
3764 PR tree-optimization/83047
3765 * gimple-ssa-store-merging.c
3766 (imm_store_chain_info::output_merged_store): If the loads with the
3767 same vuse are in different basic blocks, for load_gsi pick a load
3768 location that is dominated by the other loads.
3771 * config/i386/i386.c (ix86_memmodel_check): Start
3772 -Winvalid-memory-model diagnostics with lowercase letter.
3775 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
3776 set in_cold_section_p to the partition containing loc_list->first.
3777 When seeing loc_list->last_before_switch node, update secname and
3778 perform range_across_switch second partition handling only after that.
3781 * run-rtl-passes.c: Include debug.h.
3782 (run_rtl_passes): Call debug_hooks->assembly_start.
3783 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
3787 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
3788 OPTAB_DIRECT in calls to expand_simple_binop.
3790 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3793 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
3795 2017-11-20 David Malcolm <dmalcolm@redhat.com>
3798 * spellcheck.h (best_match::blithely_get_best_candidate): New
3801 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3803 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
3804 with lower case letter.
3806 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
3808 * config/i386/i386.md (bswaphi2): New expander.
3809 (*bswaphi2_movbe): New insn pattern.
3810 (bswaphi -> rorhi pepehole2): New peephole pattern.
3812 2017-11-20 Jeff Law <law@redhat.com>
3814 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
3815 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
3816 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
3817 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
3818 gimple-ssa-evrp-analyze.h.
3820 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
3821 set BB_VISITED here.
3822 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
3824 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3826 * tree-predcom.c: Add general comment on Store-Store chains.
3827 (split_data_refs_to_components): Postpone clearing eliminate_store_p
3829 (get_chain_last_ref_at): Rename into...
3830 (get_chain_last_write_at): ...this.
3831 (get_chain_last_write_before_load): New function.
3832 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
3833 CT_STORE_STORE when write reference is added.
3834 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
3835 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
3836 (initialize_root_vars_store_elim_1): Ditto.
3837 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
3838 definition is created.
3839 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
3840 chain by replacing it with dominant stored value.
3842 2017-11-20 Bin Cheng <bin.cheng@arm.com>
3844 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
3846 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
3848 * vr-values.c (extract_range_from_binary_expr): Use a full range
3851 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
3853 * config/arm/arm.md (R4_REGNUM): Define constant.
3854 (nonsecure_call_internal): Remove r4 clobber.
3855 (nonsecure_call_value_internal): Likewise.
3856 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
3857 clobber and resequence match_operands.
3858 (nonsecure_call_value_reg_thumb1_v5): Likewise.
3859 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
3860 (nonsecure_call_value_reg_thumb2): Likewise.
3862 2017-11-20 Jakub Jelinek <jakub@redhat.com>
3864 PR tree-optimization/78821
3865 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
3866 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
3867 to the address of the base rather than the base itself.
3868 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
3869 (find_bswap_or_nop_finalize): New function.
3870 (find_bswap_or_nop): Use it.
3871 (bswap_replace): Return a tree rather than bool, change first
3872 argument from gimple * to gimple_stmt_iterator, allow inserting
3873 into an empty sequence, allow ins_stmt to be NULL - then emit
3874 all stmts into gsi. Fix up MEM_REF address gimplification.
3875 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
3876 (struct store_immediate_info): Add N and INS_STMT non-static
3878 (store_immediate_info::store_immediate_info): Initialize them
3879 from newly added ctor args.
3880 (merged_store_group::apply_stores): Formatting fixes. Sort by
3882 (stmts_may_clobber_ref_p): For stores call also
3883 refs_anti_dependent_p.
3884 (gather_bswap_load_refs): New function.
3885 (imm_store_chain_info::try_coalesce_bswap): New method.
3886 (imm_store_chain_info::coalesce_immediate_stores): Use it.
3887 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
3888 (imm_store_chain_info::output_merged_store): Fail if number of
3889 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
3890 and NOP_EXPR rhs_code.
3891 (pass_store_merging::process_store): Compute n and ins_stmt, if
3892 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
3893 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
3895 (pass_store_merging::execute): Calculate dominators.
3897 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
3898 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
3899 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
3900 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3901 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
3902 class pass_optimize_bswap, bswap_replace,
3903 pass_optimize_bswap::execute): Moved to ...
3904 * gimple-ssa-store-merging.c: ... this file.
3905 Include optabs-tree.h.
3906 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
3907 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
3908 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
3909 anonymous namespace, remove static keywords.
3910 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
3911 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
3913 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
3916 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
3918 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
3920 * vec.h (debug_helper): New function.
3921 (DEFINE_DEBUG_VEC): New macro.
3922 * hash-set.h (debug_helper): New function.
3923 (DEFINE_DEBUG_HASH_SET): New macro.
3924 * cfg.c (debug_slim (edge)): New function.
3925 Call DEFINE_DEBUG_VEC for edges.
3926 Call DEFINE_DEBUG_HASH_SET for edges.
3927 * cfghooks.c (debug_slim (basic_block)): New function.
3928 Call DEFINE_DEBUG_VEC for basic blocks.
3929 Call DEFINE_DEBUG_HASH_SET for basic blocks.
3930 * print-tree.c (debug_slim): New function to handle trees.
3931 Call DEFINE_DEBUG_VEC for trees.
3932 Call DEFINE_DEBUG_HASH_SET for trees.
3933 (debug (vec<tree, va_gc>) &): Remove.
3934 (debug (<vec<tree, va_gc>) *): Remove.
3935 * print-rtl.c (debug_slim): New function to handle const_rtx.
3936 Call DEFINE_DEBUG_VEC for rtx_def.
3937 Call DEFINE_DEBUG_VEC for rtx_insn.
3938 Call DEFINE_DEBUG_HASH_SET for rtx_def.
3939 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
3940 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
3941 (debug (vec<rtx_insn *> *ptr): Remove.
3942 (debug_insn_vector): Remove.
3943 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
3945 2017-11-20 Tom de Vries <tom@codesourcery.com>
3947 PR rtl-optimization/82020
3948 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
3949 IF_THEN_ELSE condition.
3951 2017-11-19 Jeff Law <law@redhat.com>
3953 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
3954 of degenerates resulting from ignoring an edge.
3956 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3959 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
3961 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3964 * profile-count.c (profile_count::to_sreal_scale): Fix return value
3965 for uninitialied counts.
3967 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3970 * tree-inline.c (estimate_num_insns): Set to 1 at least.
3972 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
3975 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
3977 2017-11-19 Tom de Vries <tom@codesourcery.com>
3979 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
3982 2017-11-19 Tom de Vries <tom@codesourcery.com>
3984 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
3987 2017-11-19 Tom de Vries <tom@codesourcery.com>
3989 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
3992 2017-11-19 Tom de Vries <tom@codesourcery.com>
3994 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
3996 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
3998 2017-11-19 Tom de Vries <tom@codesourcery.com>
4000 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
4002 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
4003 semicolon after MCORE_EXPORT_NAME call.
4005 2017-11-19 Tom de Vries <tom@codesourcery.com>
4008 * vmsdbgout.c (vmsdbgout_early_finish): New function.
4009 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
4011 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
4013 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
4014 (cgraph_edge::clone): Cleanup updating of profile.
4015 * ipa-cp.c (update_profiling_info): Likewise.
4016 * ipa-inline-transform.c (inline_transform): Likewise.
4017 * ipa-inline.c (inline_small_functions): Add missing space to dump.
4018 * ipa-split.c (execute_split_functions): Do not split when function
4020 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
4021 * profile-count.c (profile_count::dump): Add global0.
4022 (profile_count::to_cgraph_frequency): Do not ICE when entry is
4024 (profile_count::to_sreal_scale): Likewise.
4025 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
4026 (profile_count::combine_with_ipa_count): New function.
4027 * profile-count.h (profile_guessed_global0adjusted): New.
4028 (profile_count::adjusted_zero): New.
4029 (profile_count::global0adjusted): New.
4030 (profile_count::combine_with_ipa_count): New.
4031 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
4032 correct profile of return block of split functions.
4033 (copy_cfg_body): Remove unused profile_count.
4034 (copy_body): Likewise.
4035 (expand_call_inline): Update.
4036 (tree_function_versioning): Update.
4038 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
4040 * hash-set.h (hash_set::empty): New.
4041 * tree-ssa-threadbackward.h: Delete.
4042 * tree-ssa-threadbackward.c (class thread_jumps): New.
4043 Move max_threaded_paths into class.
4044 (fsm_find_thread_path): Remove arguments that are now in class.
4045 (profitable_jump_thread_path): Rename to...
4046 (thread_jumps::profitable_jump_thread_path): ...this.
4047 (convert_and_register_jump_thread_path): Rename to...
4048 (thread_jumps::convert_and_register_current_path): ...this.
4049 (check_subpath_and_update_thread_path): Rename to...
4050 (thread_jumps::check_subpath_and_update_thread_path): ...this.
4051 (register_jump_thread_path_if_profitable): Rename to...
4052 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
4053 (handle_phi): Rename to...
4054 (thread_jumps::handle_phi): ...this.
4055 (handle_assignment): Rename to...
4056 (thread_jumps::handle_assignment): ...this.
4057 (fsm_find_control_statement_thread_paths): Rename to...
4058 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
4059 (find_jump_threads_backwards): Rename to...
4060 (thread_jumps::find_jump_threads_backwards): ...this.
4061 Initialize path local data.
4062 (pass_thread_jumps::execute): Call find_jump_threads_backwards
4063 from within thread_jumps class.
4064 (pass_early_thread_jumps::execute): Same.
4066 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4068 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
4070 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4072 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
4073 to canonical location.
4075 2017-11-17 Steve Ellcey <sellcey@cavium.com>
4078 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
4080 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
4082 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
4084 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
4085 rcrt1.o%s/grcrt1.o%s for -static-pie.
4087 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4089 * i386.c (ix86_multiplication_cost, ix86_division_cost,
4090 ix86_shift_rotate_cost): Break out from ...
4091 (ix86_rtx_costs): ... here.
4092 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
4095 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4097 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
4098 when entry block was promoted unlikely.
4099 (estimate_bb_frequencies): Increase frequency scale.
4100 * profile-count.h (profile_count): Export precision info.
4102 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4104 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
4105 disturbin profile of entry block.
4107 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4109 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
4112 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
4114 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
4115 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
4117 2017-11-17 Jeff Law <law@redhat.com>
4119 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
4120 from evrp_dom_walker class. Various methods moved into new class.
4121 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
4122 (evrp_range_analyzer::enter): New method.
4123 (evrp_range_analyzer::leave): New method.
4124 (evrp_dom_walker): Remove delegators no longer needed by this class.
4125 Replace vr_values data member with evrp_range_analyzer
4127 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
4128 method extracted from evrp_dom_walker::before_dom_children.
4129 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
4130 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
4132 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
4133 Add private copy constructor and move assignment operators.
4134 Privatize methods and class data where trivially possible.
4135 (evrp_dom_walker::cleanup): New function, extracted from
4136 execute_early_vrp. Simplify access to class data.
4138 * vr-values.h (get_output_for_vrp): Prototype.
4139 * vr-values.c (get_output_for_vrp): New function extracted from
4140 vrp_visit_assignment_or_call and extract_range_from_stmt.
4141 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
4143 2017-11-17 Luis Machado <luis.machado@linaro.org>
4145 * config/aarch64/aarch64.c
4146 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
4147 (qdf24xx_tunings) <autoprefetcher_model>: Set to
4148 tune_params::AUTOPREFETCHER_WEAK.
4150 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4153 * config/arm/arm.c (arm_valid_target_attribute_rec):
4154 Parse "arch=" and "+<ext>".
4155 (arm_valid_target_attribute_tree): Re-init global options.
4156 (arm_option_override): Make non-static.
4157 (arm_options_perform_arch_sanity_checks): Make errors fatal.
4158 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
4159 (__ARM_FEATURE_CRC32): Support undef.
4160 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
4161 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
4163 2017-11-17 David Malcolm <dmalcolm@redhat.com>
4165 * gdbinit.in (break-on-diagnostic): New command.
4167 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4169 * config/i386/i386.c (ix86_expand_epilogue): Change simple
4170 return to indirect jump for EH return if control-flow
4171 protection is enabled. Change explicit 'false' argument in
4172 pro_epilogue_adjust_stack with a value of
4174 * config/i386/i386.md (simple_return_indirect_internal):
4175 Remove SImode restriction to support 64-bit.
4177 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
4179 * combine.c (added_notes_insn): New.
4180 (try_combine): Handle added_notes_insn like added_links_insn.
4181 Rewrite return value code.
4182 (distribute_notes): Set added_notes_insn to the earliest insn we added
4185 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
4187 PR rtl-optimization/82621
4188 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
4189 dest of one of those SETs is unused.
4191 2017-11-17 Richard Biener <rguenther@suse.de>
4194 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
4195 * tree-pretty-print.c (dump_generic_node): Handle
4196 annot_expr_parallel_kind.
4197 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
4198 * gimplify.c (gimple_boolify): Likewise.
4200 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4202 * config.gcc (extra_headers): Add cet.h for x86 targets.
4203 * config/i386/cet.h: New file.
4204 * doc/install.texi: Add --enable-cet/--disable-cet.
4206 2017-11-17 Richard Biener <rguenther@suse.de>
4208 PR tree-optimization/83017
4209 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
4210 (gen_parallel_loop): Properly count iterations.
4211 (parallelize_loops): Handle loop->can_be_parallel independent
4212 of flag_loop_parallelize_all. Make static profitability test match
4214 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
4215 * invoke.texi (parloops-min-per-thread): Document.
4217 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
4219 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
4220 upstreaming review comments.
4222 2017-11-17 Tamar Christina <tamar.christina@arm.com>
4224 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
4225 with fast unaligned access.
4226 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
4228 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
4230 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
4231 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
4233 2017-11-17 Richard Biener <rguenther@suse.de>
4235 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
4236 folding of references.
4238 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
4241 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
4242 of replacing call to strncmp with corresponding call to strcmp when
4245 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
4247 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
4248 option prefer-avx256 for skylake-avx512 configuration.
4249 * config/i386/i386.c (ix86_option_override_internal): Ditto.
4250 (get_builtin_code_for_version): Ditto.
4252 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4253 Monk Chiang <sh.chiang04@gmail.com>
4255 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
4256 (FIXED_REGISTERS): Reserve more register numbers.
4257 (CALL_USED_REGISTERS): Likewise.
4258 (REG_ALLOC_ORDER): Likewise.
4259 (REG_CLASS_CONTENTS): Likewise.
4260 (REGISTER_NAMES): Likewise.
4262 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
4263 Kito Cheng <kito.cheng@gmail.com>
4265 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
4267 * config/nds32/iterators.md: Add vector mode iterators and attributes.
4269 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
4271 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
4272 parameter list for vec_splats.
4274 2017-11-16 Joseph Myers <joseph@codesourcery.com>
4276 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
4278 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
4280 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
4283 * doc/install.texi: Document incompatibility of Perl >=5.6.26
4284 with the required version of automake 1.11.6.
4286 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
4288 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
4290 (DU_C2_3_power9): Correct reservation combinations.
4291 (FP_DIV_power9, VEC_DIV_power9): New.
4292 (power9-alu): Split out rotate/shift...
4293 (power9-rot): ...to here, correct dispatch resource.
4294 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
4296 (power9-fp): Correct latency.
4297 (power9-sdiv): Add div/sqrt resource.
4298 (power9-ddiv): Correct latency, add div/sqrt resource.
4299 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
4300 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
4302 (power9-qpdiv, power9-qpmul): Adjust resource usage.
4304 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
4306 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
4307 switch statement mapping KF built-ins to TF built-ins if we don't
4308 have the proper ISA 3.0 assembler support.
4310 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4312 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
4313 (lower_emutls_function_body): Do not compute it.
4315 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4317 * ipa-split.c (split_bb_info): Turn time to sreal.
4318 (split_point): Likewise.
4319 (dump_split_point): Likewise.
4320 (fine_split_points): Likewise.
4321 (execute_split_functions): Only zero split_bbs; turn time to sreals.
4323 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4325 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
4328 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4330 * predict.c (combine_predictions_for_bb): Preserve zero predicted
4332 (expensive_function_p): Remove useless assert.
4333 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
4335 2017-11-16 Martin Sebor <msebor@redhat.com>
4337 PR tree-optimization/82588
4338 PR tree-optimization/82583
4339 * tree-vrp.c (check_array_ref): Handle flexible array members,
4340 string literals, and inner indices.
4341 (search_for_addr_array): Add detail to diagnostics.
4343 2017-11-16 Nathan Sidwell <nathan@acm.org>
4347 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
4348 (SET_DECL_ASSEMBLER_NAME): Forward to
4349 overwrite_decl_assembler_name.
4350 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
4351 (overwrite_decl_assembler_name): Declare.
4352 * tree.c (overwrite_decl_assembler_name): New.
4353 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
4354 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
4355 (LANG_HOOKS_INITIALIZER): Add it.
4356 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
4357 * langhooks.c (lhd_set_decl_assembler_name): Use
4358 SET_DECL_ASSEMBLER_NAME.
4359 (lhd_overwrite_decl_assembler_name): Default implementation.
4361 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4362 Jackson Woodruff <jackson.woodruff@arm.com>
4364 PR tree-optimization/71026
4365 * match.pd: Canonicalize constant multiplies in division.
4367 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
4369 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
4370 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
4371 Remove OPT_fomit_frame_pointer entry.
4372 * common/config/arc/arc-common.c: Likewise.
4373 * common/config/arm/arm-common.c: Likewise.
4374 * common/config/avr/avr-common.c: Likewise.
4375 * common/config/c6x/c6x-common.c: Likewise.
4376 * common/config/cr16/cr16-common.c: Likewise.
4377 * common/config/cris/cris-common.c: Likewise.
4378 * common/config/epiphany/epiphany-common.c: Likewise.
4379 * common/config/fr30/fr30-common.c: Likewise.
4380 * common/config/frv/frv-common.c: Likewise.
4381 * common/config/ia64/ia64-common.c: Likewise.
4382 * common/config/iq2000/iq2000-common.c: Likewise.
4383 * common/config/lm32/lm32-common.c: Likewise.
4384 * common/config/m32r/m32r-common.c: Likewise.
4385 * common/config/mcore/mcore-common.c: Likewise.
4386 * common/config/microblaze/microblaze-common.c: Likewise.
4387 * common/config/mips/mips-common.c: Likewise.
4388 * common/config/mmix/mmix-common.c: Likewise.
4389 * common/config/mn10300/mn10300-common.c: Likewise.
4390 * common/config/nios2/nios2-common.c: Likewise.
4391 * common/config/pa/pa-common.c: Likewise.
4392 * common/config/pdp11/pdp11-common.c: Likewise.
4393 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
4394 * common/config/riscv/riscv-common.c: Likewise.
4395 * common/config/rs6000/rs6000-common.c: Likewise.
4396 * common/config/rx/rx-common.c: Likewise.
4397 * common/config/s390/s390-common.c: Likewise.
4398 * common/config/sh/sh-common.c: Likewise.
4399 * common/config/sparc/sparc-common.c: Likewise.
4400 * common/config/tilegx/tilegx-common.c: Likewise.
4401 * common/config/tilepro/tilepro-common.c: Likewise.
4402 * common/config/v850/v850-common.c: Likewise.
4403 * common/config/visium/visium-common.c: Likewise.
4404 * common/config/xstormy16/xstormy16-common.c: Likewise.
4405 * common/config/xtensa/xtensa-common.c: Likewise.
4406 * invoke.texi (-fomit-frame-pointer): Update documentation.
4408 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
4410 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
4412 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4414 * cfg.c (scale_bbs_frequencies_int,
4415 cale_bbs_frequencies_gcov_type): Remove.
4416 * cfg.h (scale_bbs_frequencies_int,
4417 cale_bbs_frequencies_gcov_type): Remove.
4419 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4421 * tree-ssa-loop-manip.c
4422 (scale_dominated_blocks_in_loop): Update to profile counts.
4423 (tree_transform_and_unroll_loop): Likewise.
4425 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4427 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
4428 scale_bbs_frequencies_int.
4430 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4432 * final.c (compute_alignments): Use counts rather than frequencies.
4434 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4436 * cfgloopanal.c: Include sreal.h
4437 (average_num_loop_insns): Use counts and sreal for accounting.
4439 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4441 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
4444 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4445 Kito Cheng <kito.cheng@gmail.com>
4447 * config/nds32/constraints.md: Provide more constraints.
4448 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
4449 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
4450 support constraints usage.
4452 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4454 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
4456 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
4457 Kito Cheng <kito.cheng@gmail.com>
4459 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
4460 * config/nds32/nds32.opt: Refine the layout.
4461 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
4462 TARGET_EXT_STRING): Support new options.
4463 * config/nds32/nds32.h: Likewise.
4464 * config/nds32/nds32.md: Likewise.
4465 * config/nds32/nds32-predicates.c: Likewise.
4466 * config/nds32/constraints.md: Likewise.
4467 * common/config/nds32/nds32-common.c: Likewise.
4469 2017-11-16 Julia Koval <julia.koval@intel.com>
4472 * config/i386/gfniintrin.h: Add sse check.
4473 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
4475 2017-11-16 Julia Koval <julia.koval@intel.com>
4477 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
4478 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
4479 (ix86_handle_option): Handle -mavx512vbmi2.
4480 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
4481 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4482 * config/i386/i386-c.c (__AVX512VBMI2__): New.
4483 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
4484 (ix86_valid_target_attribute_inner_p): Ditto.
4485 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
4486 * config/i386/i386.opt (mavx512vbmi2): New option.
4487 * doc/invoke.texi: Add new option.
4489 2017-11-16 Julia Koval <julia.koval@intel.com>
4491 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
4492 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
4493 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
4494 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
4495 _mm512_gf2p8mul_epi8): New intrinsics.
4496 * config/i386/i386-builtin-types.def
4497 (V64QI_FTYPE_V64QI_V64QI): New type.
4498 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
4499 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
4500 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
4501 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
4502 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
4503 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4505 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
4507 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
4508 explicitly as a stream of bytes.
4510 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4512 * config/rs6000/altivec.h (vec_xst_be): New #define.
4513 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
4514 and externalize from *altivec_vperm_<mode>_internal.
4515 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
4517 (XL_BE_V8HI): Likewise.
4518 (XL_BE_V4SI): Likewise.
4519 (XL_BE_V4SI): Likewise.
4520 (XL_BE_V2DI): Likewise.
4521 (XL_BE_V4SF): Likewise.
4522 (XL_BE_V2DF): Likewise.
4523 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
4524 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
4525 all array entries with these keys: VSX_BUILTIN_VEC_XL,
4526 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
4527 VSX_BUILTIN_VEC_XST_BE.
4528 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
4529 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
4531 (altivec_init_builtins): Replace conditional calls to def_builtin
4532 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
4533 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
4534 with unconditional calls. Remove calls to def_builtin for
4535 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
4536 __builtin_vec_xst_be.
4537 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
4538 to define_expand, and add alternate RTL generation for P8.
4539 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
4540 vsx_ld_elemrev_v8hi.
4541 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
4542 add alternate RTL generation for P8.
4543 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
4544 vsx_ld_elemrev_v16qi.
4545 (vsx_st_elemrev_v8hi): Convert define_insn
4546 to define_expand, and add alternate RTL generation for P8.
4547 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
4548 vsx_st_elemrev_v8hi.
4549 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
4550 add alternate RTL generation for P8.
4551 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
4552 vsx_st_elemrev_v16qi.
4554 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
4557 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
4558 TARGET_AVX512ER check.
4559 (ix86_option_override_internal): Set MASK_VZEROUPPER if
4560 neither -mzeroupper nor -mno-zeroupper is used and
4561 TARGET_EMIT_VZEROUPPER is set.
4562 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
4563 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
4565 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
4567 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4568 folding of vector compares.
4569 (fold_build_vec_cmp): New helper function.
4570 (fold_compare_helper): New helper function.
4571 (builtin_function_type): Add compare builtins to the list of functions
4572 having unsigned arguments. Cosmetic updates to comment indentation.
4573 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
4574 the not+eq combination.
4576 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4578 PR tree-optimization/82726
4579 PR tree-optimization/70754
4580 * tree-predcom.c (order_drefs_by_pos): New function.
4581 (combine_chains): Move code setting has_max_use_after to...
4582 (try_combine_chains): ...here. New parameter. Sort combined chains
4583 according to position information.
4584 (tree_predictive_commoning_loop): Update call to above function.
4585 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
4587 2017-11-15 Bin Cheng <bin.cheng@arm.com>
4589 PR tree-optimization/82726
4591 2017-01-23 Bin Cheng <bin.cheng@arm.com>
4593 PR tree-optimization/70754
4594 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
4595 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
4596 combined stmt before it if not NULL.
4597 (combine_chains): Process refs reversely and compute dominance point
4601 2017-02-23 Bin Cheng <bin.cheng@arm.com>
4603 PR tree-optimization/79663
4604 * tree-predcom.c (combine_chains): Process refs in reverse order
4605 only for ZERO length chains, and add explaining comment.
4607 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4609 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
4610 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
4611 * doc/invoke.texi (ARM Options): Add armv8.3-a.
4613 2017-11-15 Tamar Christina <tamar.christina@arm.com>
4615 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
4617 2017-11-15 Martin Liska <mliska@suse.cz>
4619 * tree-cfg.c (pass_warn_function_return::execute):
4620 Compare warn_return_type for greater than zero.
4622 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
4626 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
4627 to return true on Xeon and not on Xeon Phi.
4628 (ix86_check_avx256_register): Changed to ...
4629 (ix86_check_avx_upper_register): ... this. Add extra check for
4630 VALID_AVX512F_REG_OR_XI_MODE.
4631 (ix86_avx_u128_mode_needed): Changed
4632 ix86_check_avx256_register to ix86_check_avx_upper_register.
4633 (ix86_check_avx256_stores): Changed to ...
4634 (ix86_check_avx_upper_stores): ... this. Changed
4635 ix86_check_avx256_register to ix86_check_avx_upper_register.
4636 (ix86_avx_u128_mode_after): Changed
4637 avx_reg256_found to avx_upper_reg_found. Changed
4638 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
4639 (ix86_avx_u128_mode_entry): Changed
4640 ix86_check_avx256_register to ix86_check_avx_upper_register.
4641 (ix86_avx_u128_mode_exit): Ditto.
4642 * config/i386/i386.h: (host_detect_local_cpu): New define.
4644 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
4646 * config/arm/xgene1.md (xgene1): Split into automatons
4647 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
4648 (xgene1_f_load): Adjust reservations and/or types.
4649 (xgene1_f_store): Likewise.
4650 (xgene1_load_pair): Likewise.
4651 (xgene1_store_pair): Likewise.
4652 (xgene1_fp_load1): Likewise.
4653 (xgene1_load1): Likewise.
4654 (xgene1_store1): Likewise.
4655 (xgene1_move): Likewise.
4656 (xgene1_alu): Likewise.
4657 (xgene1_simd): Likewise.
4658 (xgene1_bfm): Likewise.
4659 (xgene1_neon_load1): Likewise.
4660 (xgene1_neon_store1): Likewise.
4661 (xgene1_neon_logic): Likewise.
4662 (xgene1_neon_st1): Likewise.
4663 (xgene1_neon_ld1r): Likewise.
4664 (xgene1_alu_cond): Added.
4665 (xgene1_shift_reg): Likwise.
4666 (xgene1_bfx): Likewise.
4667 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
4669 2017-11-15 Jakub Jelinek <jakub@redhat.com>
4672 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
4674 (can_widen_mult_without_libcall): New function.
4675 (expand_mul_overflow): If only checking unsigned mul overflow,
4676 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
4677 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
4678 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
4679 (expand_DIVMOD): Formatting fix.
4680 * expmed.h (expand_mult): Add NO_LIBCALL argument.
4681 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
4682 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
4684 PR tree-optimization/82977
4685 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
4686 constructed temporary to strlen_to_stridx.put.
4688 2017-11-15 Martin Liska <mliska@suse.cz>
4690 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
4691 * configure: Regenerate.
4693 2017-11-15 Martin Liska <mliska@suse.cz>
4696 * config/sh/sh-mem.cc: Use proper probability for
4699 2017-11-14 Jeff Law <law@redhat.com>
4701 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
4702 the red zone for stack_clash_protection_final_dynamic_probe targets
4703 when the total dynamic stack size is zero bytes.
4705 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
4706 blocks is post order.
4708 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
4710 * dumpfile.h (TDF_COMPARE_DEBUG): New.
4711 * final.c (rest_of_clean_state): Set it for the
4712 -fcompare-debug dump.
4713 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
4714 class when TDF_COMPARE_DEBUG is set.
4716 * dwarf2out.c (gen_producer_string): Discard
4719 2017-11-15 Joseph Myers <joseph@codesourcery.com>
4722 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
4723 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
4724 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
4725 (__TGMATH_REAL_2_3): Remove macros.
4726 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
4727 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
4729 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
4731 (remquo): Define using __TGMATH_REAL_3.
4733 2017-11-14 Jeff Law <law@redhat.com>
4735 * vr-values.c: New file with contents extracted from tree-vrp.c.
4736 * Makefile.in (OBJS): Add vr-values.o
4737 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
4738 (set_value_range, set_and_canonicalize_value_range): Likewise.
4739 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
4740 (value_range_constant_singleton, symbolic_range_p): Likewise.
4741 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
4742 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
4743 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
4744 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
4745 (find_case_label_range, find_case_label_index): Likewise.
4746 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
4747 (range_int_cst_singleton_p, value_inside_range): Likewise.
4748 (get_single_symbol): Likewise.
4749 (switch_update): Move structure definition here.
4750 (to_remove_edges, to_update_switch_stmts): Provide externs.
4751 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
4752 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
4753 (vrp_val_is_min, set_value_range): Likewise.
4754 (set_and_canonicalize_value_range, copy_value_range): Likewise.
4755 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
4756 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
4757 (range_is_nonnull, range_int_cst_p): Likewwise.
4758 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
4759 (get_single_symbol, operand_less_p): Likewise
4760 (compare_values_warnv, compare_values): Likewise.
4761 (value_inside_range, value_range_constant_singleton): Likewise.
4762 (zero_nonzero_bitgs_from_vr): Likewise.
4763 (extract_range_from_binary_expr_1): Likewise.
4764 (overflow_comparison_p): Likewise.
4765 (to_remove_edges, to_update_switch_stmts): Likewise.
4766 (find_case_label-index, find_case_label_range): Likewise.
4767 (switch_update, set_value_range_to_nonnegative): Remove.
4768 (set_value_range_to_truthvalue): Likewise.
4769 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
4770 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
4771 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
4772 (find_case_label_ranges, test_for_singularity): Likewise.
4773 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
4774 (x_vr_values): Move to its remaining use site.
4776 2017-11-10 Jeff Law <law@redhat.com>
4778 * vr-values.h (VR_INITIALIZER): Move #define here.
4779 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
4780 * Makefile.in (OBJS): Add tree-evrp.o
4781 * tree-vrp.h (assert_info): Move structure definition here.
4782 (set_value_range_to_varying): Prototype.
4783 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4784 (infer_value_range, register_edge_assert_for): Likewise.
4785 (stmt_interesting_for_vrp): Likewise.
4786 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
4787 (set_value_range_to_varying): No longer static.
4788 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
4789 (infer_value_range, register_edge_assert_for): Likewise.
4791 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4793 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
4794 generate the XXBRD instruction.
4796 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
4797 (rs6000_builtin_type_compatible): Treat _Float128 and long double
4798 as being compatible if -mabi=ieeelongdouble.
4799 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
4800 to setup float128 built-ins with hardware support.
4801 (BU_FLOAT128_HW_2): Likewise.
4802 (BU_FLOAT128_HW_3): Likewise.
4803 (BU_FLOAT128_HW_VSX_1): Likewise.
4804 (BU_FLOAT128_HW_VSX_2): Likewise.
4805 (scalar_extract_expq): Change float128 built-in functions to
4806 accommodate having both KFmode and TFmode functions. Use the
4807 KFmode variant as the default.
4808 (scalar_extract_sigq): Likewise.
4809 (scalar_test_neg_qp): Likewise.
4810 (scalar_insert_exp_q): Likewise.
4811 (scalar_insert_exp_qp): Likewise.
4812 (scalar_test_data_class_qp): Likewise.
4813 (sqrtf128_round_to_odd): Delete processing the round to odd
4814 built-in functions as special built-in functions, and define them
4815 as float128 built-ins. Use the KFmode variant as the default.
4816 (truncf128_round_to_odd): Likewise.
4817 (addf128_round_to_odd): Likewise.
4818 (subf128_round_to_odd): Likewise.
4819 (mulf128_round_to_odd): Likewise.
4820 (divf128_round_to_odd): Likewise.
4821 (fmaf128_round_to_odd): Likewise.
4822 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
4823 support for KFmode and TFmode xststdcqp calls.
4824 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
4825 point, switch the built-in handlers for the get/set float128
4826 exponent, get float128 mantissa, float128 test built-ins, and the
4827 float128 round to odd built-in functions. Eliminate creating the
4828 float128 round to odd built-in functions as special built-ins.
4829 (rs6000_init_builtins): Eliminate special creation of the float128
4830 round to odd built-in functions.
4831 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
4832 function insns to support both TFmode and KFmode variants.
4833 (xsxsigqp_<mode>): Likewise.
4834 (xsiexpqpf_<mode>): Likewise.
4835 (xsiexpqp_<mode>): Likewise.
4836 (xststdcqp_<mode>): Likewise.
4837 (xststdcnegqp_<mode>): Likewise.
4838 (xststdcqp_<mode>): Likewise.
4840 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4842 * tree-ssa-threadupdate.c (compute_path_counts): Remove
4843 unused path_in_freq_ptr parameter.
4844 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
4846 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
4848 * ipa-inline.c (edge_badness): Dump sreal frequency.
4849 (compute_inlined_call_time): Match natural implementaiton ...
4850 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
4851 forgotten division by CGRAPH_FREQ_BASE.
4853 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4855 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
4856 Solaris 11. Update comment.
4857 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
4859 * config/sol2.h (STARTFILE_SPEC): Likewise.
4860 * configure: Regenerate.
4862 2017-11-14 Carl Love <cel@us.ibm.com>
4864 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
4865 le_ and be_ prefixes to swap* variables. Remove
4866 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
4869 2017-11-14 Jason Merrill <jason@redhat.com>
4871 Support GTY((cache)) on hash_map.
4872 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
4873 (ggc_cache_remove): Override it instead of ggc_mx.
4874 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
4875 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
4876 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
4877 (simple_cache_map_traits): Override maybe_mx.
4878 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
4879 (hash_map): Friend gt_cleare_cache.
4880 (gt_cleare_cache): New.
4881 * tree.h (tree_cache_traits): New hash_map traits class.
4882 (tree_cache_map): New typedef.
4884 2017-11-14 Richard Biener <rguenther@suse.de>
4886 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
4887 paramter and handling.
4888 (cleanup_control_flow_bb): Likewise.
4889 (cleanup_control_flow_pre): New helper performing a DFS walk
4890 to call cleanup_control_flow_bb in PRE order.
4891 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
4892 via cleanup_control_flow_pre.
4894 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
4896 * config/aarch64/aarch64-simd.md
4897 (aarch64_simd_bsl<mode>_internal): Remove DImode.
4898 (*aarch64_simd_bsl<mode>_alt): Likewise.
4899 (aarch64_simd_bsldi_internal): New.
4900 (aarch64_simd_bsldi_alt): Likewise.
4902 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4904 * tracer.c (better_p): Do not compare frequencies.
4905 * reg-stack.c (better_edge): Likewise.
4906 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
4909 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4911 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
4912 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
4913 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
4914 * lto-streamer-in.c (input_function): Use update_max_bb_count.
4915 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
4916 * predict.c (maybe_hot_frequency_p): Inline to ...
4917 (maybe_hot_count_p): ... here; rewrite to counts.
4918 (counts_to_freqs): Rename to ...
4919 (update_max_bb_count): ... this one.
4920 (expensive_function_p): Use counts.
4921 (estimate_bb_frequencies): Update.
4922 (rebuild_frequencies): Update.
4923 * predict.h (counts_to_freqs): Rename to ...
4924 (update_max_bb_count): ... this one.
4925 * profile.c (compute_branch_probabilities): Add debug info
4926 * tree-inline.c (expand_call_inline): Update debug info.
4927 (optimize_inline_calls): Use update_max_bb_count..
4928 (tree_function_versioning): Use update_max_bb_count..
4929 * value-prof.c (gimple_value_profile_transformations):
4930 Do not use update_max_bb_count.
4932 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4934 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
4935 always use frequencies.
4937 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4939 * bb-reorder.c: Remove frequencies from comments.
4940 (better_edge_p): Use profile counts.
4941 (find_traces): Dump profile counts.
4942 (rotate_loop): Use profile counts.
4943 (find_traces_1_round): Likewise.
4944 (connect_better_edge_p): Use counts instead of probabilities for
4946 (copy_bb_p): Drop early check for non-0 frequency.
4947 (sanitize_hot_paths): Update comments.
4949 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
4951 * ipa-split.c (struct split_point): Add count.
4952 (consider_split): Do not compute incoming frequency; compute incoming
4953 count and store it to split_point.
4954 (split_function): Set count of the call to split part correctly.
4956 2017-11-13 Carl Love <cel@us.ibm.com>
4958 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
4960 2017-11-13 Tom Tromey <tom@tromey.com>
4962 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
4964 2017-11-13 Carl Love <cel@us.ibm.com>
4966 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
4967 Add support for builtins:
4968 unsigned int vec_first_{,miss}_match_{,or_eos}index,
4969 vector {un,}signed {char,int,short},
4970 vector {un,}signed {char,int,short}) arguments.
4971 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
4972 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
4973 Add BU_P9V_AV_2 expansions for the builtins.
4974 * config/rs6000/altivec.h (vec_first_match_index,
4975 vec_first_mismatch_index, vec_first_match_or_eos_index,
4976 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
4977 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
4978 new extern declaration.
4979 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
4980 * config/rs6000/vsx.md (first_match_index_<mode>,
4981 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
4982 first_mismatch_or_eos_index_<mode>): Add define expand.
4983 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
4984 * doc/extend.texi: Update the built-in documenation file for the new
4987 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
4989 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
4990 into the min/max operations for _Float<N> and _Float<N>X types.
4992 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
4995 * dwarf2out.c (do_eh_frame): New static variable.
4996 (dwarf2out_begin_prologue): Set it.
4997 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
4999 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5001 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
5003 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
5005 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
5008 * debug.h (dwarf2out_do_eh_frame): Declare.
5009 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
5010 (dwarf2out_do_frame): Use it.
5011 (dwarf2out_do_cfi_asm): Likewise.
5012 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
5013 (dwarf2out_assembly_start): Likewise.
5014 (dwarf2out_begin_prologue): Fix comment.
5015 * toplev.c (compile_file): Always call dwarf2out_frame_finish
5016 if the target needs either debug or unwind DWARF2 info.
5017 * lto-opts.c (lto_write_options): Do not save -fexceptions,
5018 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
5019 -ftrapping-math, -ftrapv and -fwrapv.
5021 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
5023 * cgraph.c (cgraph_edge::sreal_frequency): New function.
5024 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
5025 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
5026 (estimate_edge_size_and_time): Likewise.
5027 (ipa_merge_fn_summary_after_inlining): Likewise.
5028 * ipa-inline.c (cgraph_freq_base_rec): Remove.
5029 (compute_uninlined_call_time): Use sreal_frequency.
5030 (compute_inlined_call_time): Likewise.
5031 (ipa_inline): Do not initialize cgraph_freq_base_rec.
5032 * profile-count.c: Include sreal.h.
5033 (profile_count::to_sreal_scale): New.
5034 * profile-count.h: Forward declare sreal.
5035 (profile_count::to_sreal_scale): Declare.
5037 2017-11-13 Nathan Sidwell <nathan@acm.org>
5039 * diagnostic.c (maybe_line_and_column): New.
5040 (diagnostic_get_location_text): Use it.
5041 (diagnostic_report_current_module): Likewise.
5042 (test_diagnostic_get_location_text): Add tests.
5044 2017-11-13 Luis Machado <luis.machado@linaro.org>
5046 * doc/md.texi (Specifying processor pipeline description): Fix
5047 incorrect latency for the div instruction example.
5049 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5051 PR tree-optimization/78821
5052 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
5053 that bit_not_p is the same.
5054 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
5055 (split_group): Count precisely bit_not_p bits in each statement.
5056 (invert_op): New function.
5057 (imm_store_chain_info::output_merged_store): Use invert_op to
5058 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
5059 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
5061 2017-11-13 Martin Liska <mliska@suse.cz>
5063 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
5064 (struct source_info): Likewise.
5065 (add_branch_counts): Likewise.
5066 (add_line_counts): Likewise.
5067 (function_summary): Likewise.
5068 (output_intermediate_line): Likewise.
5069 (generate_results): Likewise.
5071 2017-11-13 Martin Liska <mliska@suse.cz>
5073 * gcov.c (struct block_info): Remove typedef for block_t.
5074 (struct line_info): Likewise.
5075 (line_info::has_block): Likewise.
5076 (EXIT_BLOCK): Likewise.
5077 (unblock): Likewise.
5078 (circuit): Likewise.
5079 (get_cycles_count): Likewise.
5080 (process_file): Likewise.
5081 (read_graph_file): Likewise.
5082 (solve_flow_graph): Likewise.
5083 (find_exception_blocks): Likewise.
5084 (add_line_counts): Likewise.
5085 (accumulate_line_info): Likewise.
5086 (output_line_details): Likewise.
5088 2017-11-13 Martin Liska <mliska@suse.cz>
5090 * gcov.c (struct arc_info): Remove typedef for arc_t.
5091 (struct line_info): Likewise.
5092 (add_branch_counts): Likewise.
5093 (output_branch_count): Likewise.
5094 (function_info::~function_info): Likewise.
5095 (circuit): Likewise.
5096 (output_intermediate_line): Likewise.
5097 (read_graph_file): Likewise.
5098 (solve_flow_graph): Likewise.
5099 (find_exception_blocks): Likewise.
5100 (add_line_counts): Likewise.
5101 (accumulate_line_info): Likewise.
5102 (output_line_details): Likewise.
5103 (output_function_details): Likewise.
5105 2017-11-13 Martin Liska <mliska@suse.cz>
5107 * gcov.c (struct function_info): Remove typedef for function_t.
5108 (struct source_info): Likewise.
5109 (source_info::get_functions_at_location): Likewise.
5110 (solve_flow_graph): Likewise.
5111 (find_exception_blocks): Likewise.
5112 (add_line_counts): Likewise.
5113 (output_intermediate_file): Likewise.
5114 (process_file): Likewise.
5115 (generate_results): Likewise.
5116 (release_structures): Likewise.
5117 (read_graph_file): Likewise.
5118 (read_count_file): Likewise.
5119 (accumulate_line_counts): Likewise.
5120 (output_lines): Likewise.
5122 2017-11-13 Martin Liska <mliska@suse.cz>
5124 * gcov.c (function_info::function_info): Remove num_counts
5125 and add vector<gcov_type>.
5126 (function_info::~function_info): Use the vector.
5127 (process_file): Likewise.
5128 (read_graph_file): Likewise.
5129 (read_count_file): Likewise.
5130 (solve_flow_graph): Likewise.
5132 2017-11-13 Martin Liska <mliska@suse.cz>
5134 * gcov.c (function_info::is_artificial): New function.
5135 (process_file): Erase all artificial early.
5136 (generate_results): Skip as all artificial are already
5139 2017-11-13 Martin Liska <mliska@suse.cz>
5141 * gcov.c (read_graph_file): Store to global vector of functions.
5142 (read_count_file): Iterate the vector.
5143 (process_file): Likewise.
5144 (generate_results): Likewise.
5145 (release_structures): Likewise.
5147 2017-11-13 Jakub Jelinek <jakub@redhat.com>
5149 PR tree-optimization/82954
5150 * gimple-ssa-store-merging.c
5151 (imm_store_chain_info::coalesce_immediate_stores): If
5152 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
5154 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
5156 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
5157 Upddate call to ENDIAN_LANE_N.
5158 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
5159 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
5160 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
5161 and use aarch64_endian_lane_rtx.
5162 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
5164 2017-11-12 Tom de Vries <tom@codesourcery.com>
5166 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
5168 2017-11-12 Tom de Vries <tom@codesourcery.com>
5170 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
5172 2017-11-12 Tom de Vries <tom@codesourcery.com>
5174 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
5176 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
5177 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
5178 * defaults.h (ASM_OUTPUT_LABELREF): Same.
5180 2017-11-11 Martin Sebor <msebor@redhat.com>
5183 * doc/extend.texi (attribute nonstring): Remove spurious argument.
5186 * prefic.c (translate_name): Replace strncpy with memcpy to
5187 avoid -Wstringop-truncation.
5189 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5191 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
5193 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5195 * predict.c (maybe_hot_frequency_p): Do not use cfun.
5197 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5199 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
5202 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
5204 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
5205 enable generating XXBRH if the value is in a vector register.
5206 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
5207 value is in a vector register.
5208 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
5209 register to register bswap64's instead of doing the GPR sequence
5210 used on previous machines.
5211 (bswapdi2_xxbrd): New insn.
5212 (bswapdi2_reg): Disallow on ISA 3.0.
5213 (register to register bswap64 splitter): Do not split the insn on
5214 ISA 3.0 systems that use XXBRD.
5216 2017-11-10 Martin Sebor <msebor@redhat.com>
5219 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
5221 (find_subframework_file): Same.
5223 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
5225 * auto-profile.c (afdo_indirect_call): Drop frequency.
5226 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
5227 (cgraph_node::create_edge): Drop frequency argument.
5228 (cgraph_node::create_indirect_edge): Drop frequency argument.
5229 (cgraph_edge::make_speculative): Drop frequency arguments.
5230 (cgraph_edge::resolve_speculation): Do not update frequencies
5231 (cgraph_edge::dump_edge_flags): Do not dump frequency.
5232 (cgraph_node::dump): Check consistency in IPA mode.
5233 (cgraph_edge::maybe_hot_p): Use IPA counter.
5234 (cgraph_edge::verify_count_and_frequency): Rename to ...
5235 (cgraph_edge::verify_count): ... this one; drop frequency checking.
5236 (cgraph_node::verify_node): Update.
5237 * cgraph.h (struct cgraph_edge): Drop frequency.
5238 (cgraph_edge::frequency): New function.
5239 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
5241 (cgraph_edge::rebuild_edges): Likewise.
5242 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
5243 (duplicate_thunk_for_node): Do not pass frequency.
5244 (cgraph_node::create_clone): Scale only counts.
5245 (cgraph_node::create_virtual_clone): Do not pass frequency.
5246 (cgraph_node::create_edge_including_clones): Do not pass frequency.
5247 (cgraph_node::create_version_clone): Do not pass frequency.
5248 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
5249 (cgraph_node::expand_thunk): Do not pass frequency.
5250 (cgraph_node::create_wrapper): Do not pass frequency.
5251 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
5253 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
5254 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
5255 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
5256 (gather_caller_stats): Use frequency function.
5257 (ipcp_cloning_candidate_p): Use frequency function.
5258 (ipcp_propagate_stage): Use frequency function.
5259 (get_info_about_necessary_edges): Use frequency function.
5260 (update_profiling_info): Update only IPA profile.
5261 (update_specialized_profile): Use frequency functoin.
5262 (perhaps_add_new_callers): Update only IPA profile.
5263 * ipa-devirt.c (ipa_devirt): Use IPA profile.
5264 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
5265 (dump_ipa_call_summary): Use frequency function.
5266 (estimate_edge_size_and_time): Use frequency function.
5267 (ipa_merge_fn_summary_after_inlining): Use frequency function.
5268 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
5269 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
5270 (update_noncloned_counts): ... ths one; scale counts only.
5271 (clone_inlined_nodes): Do not scale frequency.
5272 (inline_call): Do not pass frequency.
5273 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
5274 (compute_inlined_call_time): Use IPA profile.
5275 (want_inline_small_function_p): Use IPA profile.
5276 (want_inline_self_recursive_call_p): Use IPA profile.
5277 (edge_badness): Use IPA profile.
5278 (lookup_recursive_calls): Use IPA profile.
5279 (recursive_inlining): Do not pass frequency.
5280 (resolve_noninline_speculation): Do not update frequency.
5281 (inline_small_functions): Collect max of IPA profile.
5282 (dump_overall_stats): Dump IPA porfile.
5283 (dump_inline_stats): Dump IPA porfile.
5284 (ipa_inline): Collect IPA stats.
5285 * ipa-inline.h (clone_inlined_nodes): Update prototype.
5286 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
5287 (ipa_propagate_frequency): Use frequency function.
5288 (ipa_profile): Cleanup.
5289 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
5290 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
5291 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
5292 (input_node): Do not stream frequency.
5293 (input_edge): Do not stream frequency.
5294 (merge_profile_summaries): Scale only IPA profiles.
5295 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
5296 * predict.c (drop_profile): Do not recompute frequency.
5297 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
5298 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
5299 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
5300 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
5301 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
5302 * tree-inline.c (copy_bb): Do not scale frequency.
5303 (expand_call_inline): Do not scale frequency.
5304 (tree_function_versioning): Do not scale frequency.
5305 * ubsan.c (ubsan_create_edge): Do not pass frequency.
5307 2017-11-10 Julia Koval <julia.koval@intel.com>
5309 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
5310 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
5311 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
5312 (_mm256_maskz_gf2p8affine_epi64_epi8)
5313 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
5314 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
5315 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
5316 (__builtin_ia32_vgf2p8affineqb_v32qi)
5317 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
5318 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
5320 2017-11-10 Tamar Christina <tamar.christina@arm.com>
5324 (arm_option_override): Refactor.
5325 (arm_option_reconfigure_globals): New.
5326 (arm_options_perform_arch_sanity_checks): New.
5327 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
5329 (arm_options_perform_arch_sanity_checks): Likewise
5331 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
5333 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
5334 (power9-qpmul): New.
5335 * rs6000/rs6000.md ("type" attr): Add qmul.
5336 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
5337 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
5338 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
5340 2017-11-10 Martin Sebor <msebor@redhat.com>
5343 * builtins.c (compute_objsize): Handle arrays that
5344 compute_builtin_object_size likes to fail for. Make extern.
5345 * builtins.h (compute_objsize): Declare.
5346 (check_strncpy_sizes): New function.
5347 (expand_builtin_strncpy): Call check_strncpy_sizes.
5348 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
5349 -Wstringop-truncation.
5350 (gimple_fold_builtin_strncat): Same.
5351 * gimple.c (gimple_build_call_from_tree): Set call location.
5352 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
5353 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
5354 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
5355 (handle_builtin_strlen): Use strlen_to_stridx.
5356 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
5358 Use strlen_to_stridx.
5359 (pass_strlen::execute): Release strlen_to_stridx.
5360 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
5361 (-Wstringop-truncation): Document new option.
5363 2017-11-10 Martin Liska <mliska@suse.cz>
5365 PR gcov-profile/82702
5366 * gcov.c (main): Handle intermediate files in a different
5368 (get_gcov_intermediate_filename): New function.
5369 (output_gcov_file): Remove support of intermediate files.
5370 (generate_results): Allocate intermediate file.
5371 (release_structures): Clean-up properly fn_end.
5372 (output_intermediate_file): Start iterating with line 1.
5374 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5376 PR tree-optimization/82929
5377 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5378 ops_swapped_p non-static data member.
5379 (store_immediate_info::store_immediate_info): Clear it.
5380 (imm_store_chain_info::coalesce_immediate_stores): If swapping
5381 ops set ops_swapped_p.
5382 (count_multiple_uses): Handle ops_swapped_p.
5384 2017-11-10 Martin Liska <mliska@suse.cz>
5386 * coverage.c (coverage_init): Stream information about
5387 support of has_unexecuted_blocks.
5388 * doc/gcov.texi: Document that.
5389 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
5390 * gcov.c (read_graph_file): Likewise.
5391 (output_line_beginning): Fix a small issue with
5394 2017-11-10 Bin Cheng <bin.cheng@arm.com>
5396 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
5397 reference of trivial component.
5399 2017-11-10 Jakub Jelinek <jakub@redhat.com>
5402 * gimple-ssa-store-merging.c
5403 (pass_store_merging::terminate_all_aliasing_chains): For
5404 gimple_store_p stmts also call refs_output_dependent_p.
5406 PR rtl-optimization/82913
5407 * compare-elim.c (try_merge_compare): Punt if def_insn is not
5410 2017-11-09 Jeff Law <law@redhat.com>
5412 * vr-values.h: New file with vr_values class.
5413 * tree-vrp.c: Include vr-values.h
5414 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
5415 data objects into the vr_values class.
5416 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
5417 (get_value_range): Make it a member function within vr_values class.
5418 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
5419 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
5420 (op_with_constant_singleton_value_range): Likewise.
5421 (extract_range_for_var_from_comparison_expr): Likewise.
5422 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
5423 (extract_range_from_binary_expr): Likewise.
5424 (extract_range_from_unary_expr): Likewise.
5425 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
5426 (check_for_binary_op_overflow, extract_range_basic): Likewise.
5427 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
5428 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
5429 (compare_name_with_value, compare_names): Likewise.
5430 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
5431 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
5432 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
5433 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
5434 (extract_range_from_phi_node): Likewise.
5435 (simplify_truth_ops_using_ranges): Likewise.
5436 (simplify_div_or_mod_using_ranges): Likewise.
5437 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
5438 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
5439 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
5440 (simplify_float_conversion_using_ranges): Likewise.
5441 (simplify_internal_call_using_ranges): Likewise.
5442 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
5443 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
5444 poitner via x_vr_values for calls into gimple folder.
5445 (vrp_initialize_lattice): Make this the vr_values ctor.
5446 (vrp_free_lattice): Make this the vr_values dtor.
5447 (set_vr_value): New function.
5448 (class vrp_prop): Add vr_values data member. Add various member
5449 functions as well as member functions that delegate to vr_values.
5450 (check_array_ref): Make a member function within vrp_prop class.
5451 (search_for_addr_array, vrp_initialize): Likewise.
5452 (vrp_finalize): Likewise. Revamp to avoid direct access to
5453 vr_value, values_propagated, etc.
5454 (check_array_bounds): Extract vrp_prop class instance pointer from
5455 walk info structure. Use it to call member functions.
5456 (check_all_array_refs): Make a member function within vrp_prop class.
5457 Smuggle class instance pointer via walk info structure.
5458 (x_vr_values): New local static.
5459 (vrp_valueize): Use x_vr_values to get class instance.
5460 (vr_valueize_1): Likewise.
5461 (class vrp_folder): Add vr_values data member. Add various member
5462 functions as well as member functions that delegate to vr_values.
5463 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
5464 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
5465 class instance from vr_values. Use it to call member functions.
5466 (vrp_dom_walker): Add vr_values data member.
5467 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
5468 instance via x_vr_values.
5469 (identify_jump_threads): Accept vr_values as argument. Store
5470 it into the walker structure.
5471 (evrp_dom_walker): Add vr_values class data member. Add various
5473 (evrp_dom_walker::try_find_new_range): Use vr_values data
5474 member to access the memory allocator.
5475 (evrp_dom_walker::before_dom_children): Store vr_values class
5476 instance into the vrp_folder class.
5477 (evrp_dom_walker::push_value_range): Rework to avoid direct
5478 access to num_vr_values and vr_value.
5479 (evrp_dom_walker::pop_value_range): Likewise.
5480 (execute_early_vrp): Remove call to vrp_initialize_lattice.
5481 Use vr_values to get to dump_all_value_ranges member function.
5482 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
5483 and simplify_cond_using_ranges_2 via vrp_prop class instance.
5484 Pass vr_values class instance down to identify_jump_threads.
5485 Remove call to vrp_free_lattice.
5486 (debug_all_value_ranges): Remove.
5488 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
5489 (vrp_folder): Likewise.
5491 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
5492 Get it from the existing bitmap instead.
5493 (vrp_intersect_ranges_1): Likewise.
5495 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5497 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5499 (store_immediate_info::store_immediate_info): Add bitnotp argument,
5500 set bit_not_p to it.
5501 (imm_store_chain_info::coalesce_immediate_stores): Break group
5502 if bit_not_p is different.
5503 (count_multiple_uses, split_group,
5504 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
5505 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
5506 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
5507 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
5508 to store_immediate_info ctor.
5510 2017-11-09 Jim Wilson <jimw@sifive.com>
5512 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
5513 (scan_prog_file): Likewise.
5515 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5517 * bb-reorder.c (max_entry_frequency): Remove.
5518 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
5519 connect_traces, push_to_next_round_p): Remove prototypes.
5520 (find_traces_1_round): Use counts only.
5521 (push_to_next_round_p): Likewise.
5522 (find_traces): Likewise.
5523 (rotate_loop): Likewise.
5524 (find_traces_1_round): Likewise.
5525 (connect_traces): Likewise.
5526 (edge_order): Likewise.
5528 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
5530 * config/arm/arm.c (output_return_instruction): Add comments to
5531 indicate requirement for cmse_nonsecure_entry return to account
5532 for the size of clearing instruction output here.
5533 (thumb_exit): Likewise.
5534 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
5535 return in hardfloat mode.
5537 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
5539 * config/rs6000/rs6000.c (machine_function): Add a bool,
5540 "toc_is_wrapped_separately".
5541 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
5542 if it wasn't explicitly set or unset, we are optimizing for speed, and
5543 doing separate shrink-wrapping.
5544 (rs6000_get_separate_components): Enable the TOC component if
5545 saving the TOC register in the prologue.
5546 (rs6000_components_for_bb): Handle the TOC component.
5547 (rs6000_emit_prologue_components): Store the TOC register where needed.
5548 (rs6000_set_handled_components): Mark TOC as handled, if handled.
5549 (rs6000_emit_prologue): Don't save the TOC if that is already done.
5551 2017-11-09 Martin Jambor <mjambor@suse.cz>
5553 * ipa-param-manipulation.c: New file.
5554 * ipa-param-manipulation.h: Likewise.
5555 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
5556 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
5557 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
5558 (ipa_parm_adjustment): Likewise.
5559 (ipa_parm_adjustment_vec): Likewise.
5560 (ipa_get_vector_of_formal_parms): Moved declaration to
5561 ipa-param-manipulation.h.
5562 (ipa_get_vector_of_formal_parm_types): Likewise.
5563 (ipa_modify_formal_parameters): Likewise.
5564 (ipa_modify_call_arguments): Likewise.
5565 (ipa_combine_adjustments): Likewise.
5566 (ipa_dump_param_adjustments): Likewise.
5567 (ipa_modify_expr): Likewise.
5568 (ipa_get_adjustment_candidate): Likewise.
5569 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
5570 ipa-param-manipulation.c.
5571 (ipa_get_vector_of_formal_parm_types): Likewise.
5572 (ipa_modify_formal_parameters): Likewise.
5573 (ipa_modify_call_arguments): Likewise.
5574 (ipa_modify_expr): Likewise.
5575 (get_ssa_base_param): Likewise.
5576 (ipa_get_adjustment_candidate): Likewise.
5577 (index_in_adjustments_multiple_times_p): Likewise.
5578 (ipa_combine_adjustments): Likewise.
5579 (ipa_dump_param_adjustments): Likewise.
5580 * tree-sra.c: Also include ipa-param-manipulation.h
5581 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
5584 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5585 Alan Hayward <alan.hayward@arm.com>
5586 David Sherwood <david.sherwood@arm.com>
5588 * doc/sourcebuild.texi (vect_masked_store): Document.
5590 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5591 Alan Hayward <alan.hayward@arm.com>
5592 David Sherwood <david.sherwood@arm.com>
5594 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
5596 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5597 Alan Hayward <alan.hayward@arm.com>
5598 David Sherwood <david.sherwood@arm.com>
5600 * doc/sourcebuild.texi (vect_variable_length): Document.
5602 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5603 Alan Hayward <alan.hayward@arm.com>
5604 David Sherwood <david.sherwood@arm.com>
5606 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
5608 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5609 Alan Hayward <alan.hayward@arm.com>
5610 David Sherwood <david.sherwood@arm.com>
5612 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
5614 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5615 Alan Hayward <alan.hayward@arm.com>
5616 David Sherwood <david.sherwood@arm.com>
5618 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
5619 previously undocumented selectors.
5620 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
5622 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5624 * doc/rtl.texi (const_vector): Say that elements can be
5625 const_wide_ints too.
5626 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
5627 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
5628 (gen_vec_duplicate): Use it instead of CONSTANT_P.
5629 * optabs.c (expand_vector_broadcast): Likewise.
5631 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5632 Alan Hayward <alan.hayward@arm.com>
5633 David Sherwood <david.sherwood@arm.com>
5635 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
5636 scaled index even if the unscaled address was invalid.
5637 Don't increase the complexity of using a scale in that case.
5639 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
5640 Alan Hayward <alan.hayward@arm.com>
5641 David Sherwood <david.sherwood@arm.com>
5643 * doc/rtl.texi: Rewrite the subreg rules so that they partition
5644 the inner register into REGMODE_NATURAL_SIZE bytes rather than
5645 UNITS_PER_WORD bytes.
5646 * emit-rtl.c (validate_subreg): Divide subregs into blocks
5647 based on REGMODE_NATURAL_SIZE of the inner mode.
5648 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
5649 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
5650 * expmed.c (lowpart_bit_field_p): Divide the value up into
5651 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
5652 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
5653 whether something is likely to occupy more than one register.
5655 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
5658 * ipa-inline-transform.c (update_noncloned_frequencies): Use
5659 profile_count::adjust_for_ipa_scaling.
5660 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
5661 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
5663 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
5665 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5667 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
5668 (split_group): Add total_orig and total_new arguments, estimate the
5669 number of statements related to the store group without store merging
5670 and with store merging.
5671 (imm_store_chain_info::output_merged_store): Adjust split_group
5672 callers, punt if estimated number of statements with store merging
5673 is not smaller than estimated number of statements without it.
5675 (handled_load): Remove has_single_use checks.
5676 (pass_store_merging::process_store): Likewise.
5678 2017-11-09 Richard Biener <rguenther@suse.de>
5680 PR tree-optimization/82902
5681 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
5683 2017-11-09 Martin Liska <mliska@suse.cz>
5686 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
5689 2017-11-09 Martin Liska <mliska@suse.cz>
5691 PR tree-optimization/82669
5692 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
5694 2017-11-09 Martin Liska <mliska@suse.cz>
5696 PR gcov-profile/48463
5697 * coverage.c (coverage_begin_function): Output also end locus
5698 of a function and information whether the function is
5700 * gcov-dump.c (tag_function): Parse and print the information.
5701 * gcov.c (INCLUDE_MAP): Add include.
5702 (INCLUDE_SET): Likewise.
5703 (struct line_info): Move earlier in the source file because
5704 of vector<line_info> in function_info structure.
5705 (line_info::line_info): Likewise.
5706 (line_info::has_block): Likewise.
5707 (struct source_info): Add new member index.
5708 (source_info::get_functions_at_location): New function.
5709 (function_info::group_line_p): New function.
5710 (output_intermediate_line): New function.
5711 (output_intermediate_file): Use the mentioned function.
5712 (struct function_start): New.
5713 (struct function_start_pair_hash): Likewise.
5714 (process_file): Add code that identifies group functions.
5715 Assign lines either to global or function scope.
5716 (generate_results): Skip artificial functions.
5717 (find_source): Assign index for each source file.
5718 (read_graph_file): Read new flag artificial and end_line.
5719 (add_line_counts): Assign it either to global of function scope.
5720 (accumulate_line_counts): Isolate core of the function to
5721 accumulate_line_info and call it for both function and global
5723 (accumulate_line_info): New function.
5724 (output_line_beginning): Fix GNU coding style.
5725 (print_source_line): New function.
5726 (output_line_details): Likewise.
5727 (output_function_details): Likewise.
5728 (output_lines): Iterate both source (global) scope and function
5730 (struct function_line_start_cmp): New class.
5731 * doc/gcov.texi: Reflect changes in documentation.
5733 2017-11-09 Jakub Jelinek <jakub@redhat.com>
5736 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
5737 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
5738 and similarly for not instead of neg.
5740 2017-11-08 Andi Kleen <ak@linux.intel.com>
5742 * config/i386/i386.opt: Add -mforce-indirect-call.
5743 * config/i386/predicates.md: Check for flag_force_indirect_call.
5744 * doc/invoke.texi: Document -mforce-indirect-call
5746 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
5748 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
5750 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
5751 config/riscv/riscv.c (predict.h): New include.
5752 (riscv_slow_unaligned_access_p): No longer static.
5753 (riscv_block_move_straight): Add require.
5754 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
5756 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5759 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
5760 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
5761 nonimmediate_operand predicate for operand 1 instead of
5764 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5766 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
5768 * config/aarch64/constraints.md (Uml): New constraint.
5769 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
5772 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5774 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
5775 of two vec_duplicates into a vec_concat.
5777 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5779 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5780 Simplify vec_merge of vec_duplicate and vec_concat.
5781 * config/aarch64/constraints.md (Utq): New constraint.
5782 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
5785 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5787 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
5788 Simplify vec_merge of vec_duplicate and const_vector.
5789 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
5791 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
5792 mode iterator. Update predicate on operand 1 to
5793 handle non-const_vec constants. Delete constraints.
5794 (*aarch64_combinez_be<mode>): Likewise for operand 2.
5796 2017-11-08 Jakub Jelinek <jakub@redhat.com>
5798 PR tree-optimization/78821
5799 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
5801 (store_operand_info::store_operand_info): Initialize it to false.
5802 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
5803 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
5804 store in the group, and if chain_info is non-NULL, to ignore altogether
5806 (compatible_load_p): Fail if bit_not_p does not match.
5807 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
5808 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
5809 (pass_store_merging::process_store): Adjust
5810 terminate_all_aliasing_chains calls to pass NULL in all current spots,
5811 call terminate_all_aliasing_chains newly when adding a store into
5812 a chain with non-NULL chain_info.
5814 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5816 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
5818 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
5820 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
5822 (aarch64_layout_frame): Initialise emit_frame_chain.
5823 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
5824 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
5826 2017-11-08 Martin Liska <mliska@suse.cz>
5828 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
5829 of gimple_call_internal_p.
5831 2017-11-07 Tom de Vries <tom@codesourcery.com>
5833 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
5835 2017-11-07 Tom de Vries <tom@codesourcery.com>
5837 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
5840 2017-11-08 Martin Liska <mliska@suse.cz>
5843 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
5845 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5847 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
5849 2017-11-07 Jakub Jelinek <jakub@redhat.com>
5852 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
5853 (*cmp<mode>_ccz_1): New insn with $k alternative.
5856 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
5857 RTX_COMM_COMPARE as commutative as well.
5858 (ix86_binary_operator_ok): Formatting fix.
5859 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
5860 *<code><mode>3<mask_name><round_saeonly_name>,
5861 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
5862 *<s>mul<mode>3_highpart<mask_name>,
5863 *vec_widen_umult_even_v16si<mask_name>,
5864 *vec_widen_umult_even_v8si<mask_name>,
5865 *vec_widen_umult_even_v4si<mask_name>,
5866 *vec_widen_smult_even_v16si<mask_name>,
5867 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
5868 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
5869 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
5870 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
5871 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
5872 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
5873 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
5874 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
5875 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
5876 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
5877 ix86_binary_operator_ok. Formatting fixes.
5878 (*<plusminus_insn><mode>3<mask_name><round_name>,
5879 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
5882 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
5884 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
5885 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
5886 eq and ne if TARGET_ISEL.
5887 (cmp): New code_iterator.
5888 (UNS, UNSU_, UNSIK): New code_attrs.
5889 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
5890 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
5892 ("eq<mode>3"): ... this.
5893 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
5895 ("ne<mode>3"): ... this.
5897 2017-11-07 Julia Koval <julia.koval@intel.com>
5900 * common/config/i386/i386-common.c
5901 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
5902 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
5903 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
5904 * config/i386/i386.opt: Ditto.
5905 * config/i386/i386.c (ix86_target_string): Ditto.
5906 (ix86_option_override_internal): Ditto.
5907 (ix86_init_mpx_builtins): Move MPX to args2.
5908 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
5909 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
5910 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
5911 __builtin_ia32_vgf2p8affineinvqb_v32qi,
5912 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
5913 __builtin_ia32_vgf2p8affineinvqb_v16qi,
5914 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
5916 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
5919 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
5920 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
5921 (zero-extendsidi peephole2): Remove peephole.
5923 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
5926 * stor-layout.c: Include attribs.h.
5927 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
5928 explicit lookup of "aligned" attribute.
5930 2017-11-07 Andrew Waterman <andrew@sifive.com>
5932 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
5933 (riscv_expand_block_move): Likewise.
5934 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
5936 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
5937 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
5938 * config/riscv/riscv.c (riscv_block_move_straight): New function.
5939 (riscv_adjust_block_mem): Likewise.
5940 (riscv_block_move_loop): Likewise.
5941 (riscv_expand_block_move): Likewise.
5942 * config/riscv/riscv.md (movmemsi): New pattern.
5944 2017-11-07 Michael Clark <michaeljclark@mac.com>
5946 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
5947 (MUSL_DYNAMIC_LINKER): Likewise.
5949 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
5951 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
5952 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
5955 2017-11-07 Richard Biener <rguenther@suse.de>
5957 * match.pd: Fix build.
5959 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
5960 Jackson Woodruff <jackson.woodruff@arm.com>
5962 PR tree-optimization/71026
5963 * match.pd: Canonicalize negate in division.
5965 2017-11-07 Sudakshina Das <sudi.das@arm.com>
5968 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
5970 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5972 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
5973 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
5975 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
5977 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
5978 non-scalar integral types.
5979 * match.pd (negate_expr_p): Handle MINUS_EXPR.
5980 (-(A-B), -(~A)): New transformations.
5982 2017-11-07 Tom de Vries <tom@codesourcery.com>
5984 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
5985 semicolon after "do {} while (0)".
5986 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5987 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5988 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5989 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5990 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5992 2017-11-07 Tom de Vries <tom@codesourcery.com>
5994 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
5995 after "do {} while (0)".
5996 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5997 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5998 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
5999 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6000 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
6002 2017-11-07 Tom de Vries <tom@codesourcery.com>
6005 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
6007 (arm_rtx_costs_internal): Add missing semicolon after
6008 HANDLE_NARROW_SHIFT_ARITH call.
6010 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6012 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
6013 disable isel if it was not set explicitly.
6015 2017-11-06 James Bowman <james.bowman@ftdichip.com>
6017 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
6018 (add_type_attribute) likewise.
6020 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
6022 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
6023 of struct ix86_frame.
6024 (ix86_initial_elimination_offset): Likewise.
6025 (ix86_expand_split_stack_prologue): Likewise.
6027 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
6029 * tree-vrp.h (enum value_range_type): Update stale comment.
6031 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6032 Alan Hayward <alan.hayward@arm.com>
6033 David Sherwood <david.sherwood@arm.com>
6035 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
6036 (aarch64_expand_vec_perm_const): Take the number of units too.
6037 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
6038 (aarch64_expand_vec_perm_const): Likewise.
6039 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
6040 (vec_perm<mode>): Update accordingly.
6042 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6043 Alan Hayward <alan.hayward@arm.com>
6044 David Sherwood <david.sherwood@arm.com>
6046 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
6047 Take the number of units too.
6048 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
6049 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
6050 but check for a vector mode before rather than after the call.
6051 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
6052 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
6053 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
6054 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
6055 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
6056 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
6057 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
6058 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
6059 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
6060 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
6061 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
6062 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
6063 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
6064 (aarch64_sqdmull2_n<mode>): Update accordingly.
6066 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6067 Alan Hayward <alan.hayward@arm.com>
6068 David Sherwood <david.sherwood@arm.com>
6070 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
6071 the number of units too.
6072 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
6073 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
6074 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
6075 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
6076 (vec_store_lanesxi<mode>): Update accordingly.
6078 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6079 Alan Hayward <alan.hayward@arm.com>
6080 David Sherwood <david.sherwood@arm.com>
6082 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
6083 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
6084 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
6085 of units rather than the mode.
6086 * config/aarch64/iterators.md (nunits): New mode attribute.
6087 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
6088 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
6089 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
6090 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
6091 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
6092 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
6093 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
6094 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
6095 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
6096 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
6097 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
6098 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
6099 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
6100 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
6101 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
6102 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
6103 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
6104 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
6105 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
6106 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
6107 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
6108 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
6109 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
6110 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
6111 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
6112 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
6113 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
6114 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
6115 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
6116 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
6117 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
6118 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
6119 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
6120 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
6121 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
6122 (aarch64_simd_vec_setv2di): Likewise.
6124 2017-11-06 Carl Love <cel@us.ibm.com>
6126 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
6128 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
6129 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
6131 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
6133 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
6134 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
6135 Add power 8 macro expansions.
6136 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
6137 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
6138 power 8 instructions. (VSX_XXBR): Add iterator.
6140 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
6142 * config/arm/arm.md (predicable_short_it): Change default to "no",
6143 improve documentation, remove uses that are identical to the default.
6144 (enabled_for_depr_it): Rename to enabled_for_short_it.
6145 * gcc/config/arm/arm-fixed.md (predicable_short_it):
6146 Remove default uses.
6147 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
6148 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
6149 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
6150 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
6152 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
6155 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
6156 float128 helper macros, which are no longer used after deleting
6157 the old 'q' built-in functions, and moving the round to odd
6158 built-in functions to being special built-in functions.
6159 (BU_FLOAT128_2): Likewise.
6160 (BU_FLOAT128_1_HW): Likewise.
6161 (BU_FLOAT128_2_HW): Likewise.
6162 (BU_FLOAT128_3_HW): Likewise.
6163 (FABSQ): Delete old 'q' built-in functions.
6164 (COPYSIGNQ): Likewise.
6165 (SQRTF128_ODD): Move round to odd built-in functions to be
6166 special built-in functions, so that we can handle
6167 -mabi=ieeelongdouble.
6168 (TRUNCF128_ODD): Likewise.
6169 (ADDF128_ODD): Likewise.
6170 (SUBF128_ODD): Likewise.
6171 (MULF128_ODD): Likewise.
6172 (DIVF128_ODD): Likewise.
6173 (FMAF128_ODD): Likewise.
6174 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
6175 built-in names to 'f128'.
6176 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
6177 old 'q' built-in functions, as the machine independent code for
6178 'f128' built-in functions handles this.
6179 (rs6000_expand_builtin): Add expansion for float128 round to odd
6180 functions, keying off on -mabi=ieeelongdouble of whether to use
6181 the KFmode or TFmode variant.
6182 (rs6000_init_builtins): Initialize the _Float128 round to odd
6184 * doc/extend.texi (PowerPC Built-in Functions): Document the old
6185 _Float128 'q' built-in functions are now mapped into the new
6186 'f128' built-in functions.
6188 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
6190 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
6191 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
6193 2017-11-06 David Malcolm <dmalcolm@redhat.com>
6196 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
6197 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
6198 * toplev.c: Include "ipa-fnsummary.h".
6199 (toplev::finalize): Call ipa_fnsummary_c_finalize.
6201 2017-11-06 Jakub Jelinek <jakub@redhat.com>
6203 PR tree-optimization/82838
6204 * gimple-ssa-store-merging.c
6205 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
6206 on a separate gimple_seq which is then appended to seq.
6208 2017-11-06 Jeff Law <law@redhat.com>
6211 * config/i386/i386.c (PROBE_INTERVAL): Remove.
6212 (get_probe_interval): New functions.
6213 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
6214 (ix86_adjust_stack_and_probe): Likewise.
6215 (output_adjust_stack_and_probe): Likewise.
6216 (ix86_emit_probe_stack_range): Likewise.
6217 (ix86_expand_prologue): Likewise.
6219 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6221 PR tree-optimization/82816
6222 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
6223 if the modes of the two types are the same.
6224 (convert_plusminus_to_widen): Likewise.
6226 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6228 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
6230 (usadv16qi): New define_expand.
6231 (usadv8hi): New define_expand.
6233 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
6236 * ipa-inline-transform.c (update_noncloned_frequencies): Always
6238 (inline_transform): Likewise.
6239 * predict.c (counts_to_freqs): Remove useless conditional.
6240 * profile-count.h (profile_count::apply_scale): Move sanity check.
6241 * tree-inline.c (copy_bb): Always scale.
6242 (copy_cfg_body): Likewise.
6244 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
6247 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
6249 (*cmp_ite0): Add enabled_for_depr_it attribute.
6250 (*cmp_ite1): Likewise.
6252 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
6254 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
6257 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
6259 * tree-vrp.c (vrp_int_const_binop): Return true on success and
6260 return the value by pointer.
6261 (extract_range_from_multiplicative_op_1): Update accordingly.
6262 Return as soon as an operation fails.
6264 2017-11-05 Tom de Vries <tom@codesourcery.com>
6267 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
6268 (DEF_SANITIZER_BUILTIN): ... here.
6269 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
6270 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
6272 2017-11-05 Tom de Vries <tom@codesourcery.com>
6275 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
6277 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
6278 ASM_OUTPUT_BEFORE_CASE_LABEL call.
6279 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
6281 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6282 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6283 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
6285 2017-11-05 Tom de Vries <tom@codesourcery.com>
6288 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
6291 2017-11-04 Michael Clark <michaeljclark@mac.com>
6293 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
6294 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
6296 (*addsi3_extended): Likewise.
6297 (*addsi3_extended2): Likewise.
6298 (<optab>si3): Likewise.
6299 (<optab>di3): Likewise.
6300 (<optab><mode>3): Likewise.
6301 (<*optabe>si3_internal): Likewise.
6302 (zero_extendqi<SUPERQI:mode>2): Likewise.
6303 (*add<mode>hi3): Likewise.
6304 (*xor<mode>hi3): Likewise.
6305 (<optab>di3): Likewise.
6306 (*<optab>si3_extend): Likewise.
6307 (*sge<u>_<X:mode><GPR:mode>): Likewise.
6308 (*slt<u>_<X:mode><GPR:mode>): Likewise.
6309 (*sle<u>_<X:mode><GPR:mode>): Likewise.
6311 2017-11-04 Andrew Waterman <andrew@sifive.com>
6313 * config/riscv/riscv.c (riscv_option_override): Conditionally set
6314 TARGET_STRICT_ALIGN based upon -mtune argument.
6316 2017-11-04 Andrew Waterman <andrew@sifive.com>
6318 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
6320 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
6322 * config/i386/i386.c (choose_basereg): Use optional scratch
6323 register and add assertion.
6324 (x86_emit_outlined_ms2sysv_save): Use scratch register when
6325 needed, and don't allocate stack.
6326 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
6327 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
6328 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
6330 2017-11-03 Jeff Law <law@redhat.com>
6332 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
6333 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
6334 to probe at the start of a noreturn function.
6336 2017-11-03 Jakub Jelinek <jakub@redhat.com>
6338 PR tree-optimization/78821
6339 * gimple-ssa-store-merging.c: Update the file comment.
6340 (MAX_STORE_ALIAS_CHECKS): Define.
6341 (struct store_operand_info): New type.
6342 (store_operand_info::store_operand_info): New constructor.
6343 (struct store_immediate_info): Add rhs_code and ops data members.
6344 (store_immediate_info::store_immediate_info): Add rhscode, op0r
6345 and op1r arguments to the ctor, initialize corresponding data members.
6346 (struct merged_store_group): Add load_align_base and load_align
6348 (merged_store_group::merged_store_group): Initialize them.
6349 (merged_store_group::do_merge): Update them.
6350 (merged_store_group::apply_stores): Pick the constant for
6351 encode_tree_to_bitpos from one of the two operands, or skip
6352 encode_tree_to_bitpos if neither operand is a constant.
6353 (class pass_store_merging): Add process_store method decl. Remove
6354 bool argument from terminate_all_aliasing_chains method decl.
6355 (pass_store_merging::terminate_all_aliasing_chains): Remove
6356 var_offset_p argument and corresponding handling.
6357 (stmts_may_clobber_ref_p): New function.
6358 (compatible_load_p): New function.
6359 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
6360 if there is overlap and rhs_code is not INTEGER_CST. For
6361 non-overlapping stores terminate group if rhs is not mergeable.
6362 (get_alias_type_for_stmts): Change first argument from
6363 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
6364 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
6365 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
6367 (get_location_for_stmts): Change first argument from
6368 auto_vec<gimple *> & to vec<gimple *> &.
6369 (struct split_store): Remove orig_stmts data member, add orig_stores.
6370 (split_store::split_store): Create orig_stores rather than orig_stmts.
6371 (find_constituent_stmts): Renamed to ...
6372 (find_constituent_stores): ... this. Change second argument from
6373 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
6374 to info structures rather than the statements.
6375 (split_group): Rename ALLOW_UNALIGNED argument to
6376 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
6377 it. Adjust find_constituent_stores caller.
6378 (imm_store_chain_info::output_merged_store): Handle rhs_code other
6379 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
6380 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
6381 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
6382 (mem_valid_for_store_merging): New function.
6383 (handled_load): New function.
6384 (pass_store_merging::process_store): New method.
6385 (pass_store_merging::execute): Use process_store method. Adjust
6386 terminate_all_aliasing_chains caller.
6388 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6390 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
6391 Return true for more constants, symbols and label references.
6392 (aarch64_valid_floating_const): Remove unused function.
6394 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6397 * config/aarch64/aarch64.c (aarch64_layout_frame):
6398 Undo forcing of LR at bottom of frame.
6400 2017-11-03 Jeff Law <law@redhat.com>
6403 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
6404 for int_registers_saved.
6406 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
6407 extracted from tree-ssa-dom.c.
6408 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
6409 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
6410 (record_equivalences_from_incoming_edge): Add additional argument
6411 to single_pred_edge_ignoring_loop_edges call.
6412 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
6413 (uncprop_dom_walker::before_dom_children): Add additional argument
6414 to single_pred_edge_ignoring_loop_edges call.
6415 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
6416 single_pred_edge_ignoring_loop_edges rather than open coding.
6417 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
6419 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
6421 * match.pd (-(-A)): Rewrite.
6423 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6425 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
6426 (rs6000_emit_int_cmove): New declaration.
6427 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
6428 (rs6000_emit_sISEL): Delete.
6429 (rs6000_emit_int_cmove): Make non-static.
6430 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
6431 instead of rs6000_emit_sISEL.
6433 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
6435 * asan.c (create_cond_insert_point): Maintain profile.
6436 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
6438 * basic-block.h (struct basic_block_def): Remove frequency.
6439 (EDGE_FREQUENCY): Use to_frequency
6440 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
6442 (find_traces): Update to use to_frequency.
6443 (find_traces_1_round): Likewise; use only IPA counts.
6444 (bb_to_key): Likewise.
6445 (connect_traces): Use IPA counts only.
6446 (copy_bb_p): Update to use to_frequency.
6447 (fix_up_crossing_landing_pad): Likewise.
6448 (sanitize_hot_paths): Likewise.
6449 * bt-load.c (basic_block_freq): Likewise.
6450 * cfg.c (init_flow): Set count_max to uninitialized.
6451 (check_bb_profile): Remove frequencies; check counts.
6452 (dump_bb_info): Do not dump frequencies.
6453 (update_bb_profile_for_threading): Update counts only.
6454 (scale_bbs_frequencies_int): Likewise.
6455 (MAX_SAFE_MULTIPLIER): Remove.
6456 (scale_bbs_frequencies_gcov_type): Update counts only.
6457 (scale_bbs_frequencies_profile_count): Update counts only.
6458 (scale_bbs_frequencies): Update counts only.
6459 * cfg.h (struct control_flow_graph): Add count-max.
6460 (update_bb_profile_for_threading): Update prototype.
6461 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
6462 (find_many_sub_basic_blocks): Likewise.
6463 * cfgcleanup.c (try_forward_edges): Likewise.
6464 (try_crossjump_to_edge): Likewise.
6465 * cfgexpand.c (expand_gimple_cond): Likewise.
6466 (expand_gimple_tailcall): Likewise.
6467 (construct_init_block): Likewise.
6468 (construct_exit_block): Likewise.
6469 * cfghooks.c (verify_flow_info): Check consistency of counts.
6470 (dump_bb_for_graph): Do not dump frequencies.
6471 (split_block_1): Do not update frequencies.
6472 (split_edge): Do not update frequencies.
6473 (make_forwarder_block): Do not update frequencies.
6474 (duplicate_block): Do not update frequencies.
6475 (account_profile_record): Do not update frequencies.
6476 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
6477 for global heuristics.
6478 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
6479 (expected_loop_iterations_unbounded): Use counts only.
6480 * cfgloopmanip.c (scale_loop_profile): Simplify.
6481 (create_empty_loop_on_edge): Simplify
6483 (duplicate_loop_to_header_edge): Simplify
6484 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
6485 (update_br_prob_note): Take care of removing note when profile
6487 (relink_block_chain): Do not dump frequency.
6488 (rtl_account_profile_record): Use to_frequency.
6489 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
6490 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
6491 (cgraph_update_edges_for_call_stmt_node): Likewise.
6492 (cgraph_edge::verify_count_and_frequency): Update.
6493 (cgraph_node::verify_node): Temporarily disable frequency verification.
6494 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
6495 to_cgraph_frequency.
6496 (cgraph_edge::rebuild_edges): Convert to ipa counts.
6497 * cgraphunit.c (init_lowered_empty_function): Do not initialize
6499 (cgraph_node::expand_thunk): Update profile.
6500 * except.c (dw2_build_landing_pads): Do not update frequency.
6501 * final.c (compute_alignments): Use to_frequency.
6502 (dump_basic_block_info): Do not dump frequency.
6503 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
6504 (dump_gimple_bb_header): Do not dump frequency.
6505 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
6507 * gimple-streamer-in.c (input_bb): Do not stream frequency.
6508 * gimple-streamer-out.c (output_bb): Do not stream frequency.
6509 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
6510 (init_before_recovery): Do not update frequency.
6511 (sched_create_recovery_edges): Do not update frequency.
6512 * hsa-gen.c (convert_switch_statements): Do not update frequency.
6513 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
6514 (ipa_cp_c_finalize): Set max_count to uninitialized.
6515 * ipa-fnsummary.c (get_minimal_bb): Use counts.
6516 (param_change_prob): Use counts.
6517 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
6519 * ipa-split.c (consider_split): Use to_frequency.
6520 (split_function): Use to_frequency.
6521 * ira-build.c (loop_compare_func): Likewise.
6522 (mark_loops_for_removal): Likewise.
6523 (mark_all_loops_for_removal): Likewise.
6524 * loop-doloop.c (doloop_modify): Do not update frequency.
6525 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
6527 * lto-streamer-in.c (input_function): Update count_max.
6528 * omp-expand.c (expand_omp_taskreg): Update count_max.
6529 * omp-simd-clone.c (simd_clone_adjust): Update profile.
6530 * predict.c (maybe_hot_frequency_p): Use to_frequency.
6531 (maybe_hot_count_p): Use ipa counts only.
6532 (maybe_hot_bb_p): Simplify.
6533 (maybe_hot_edge_p): Simplify.
6534 (probably_never_executed): Do not take frequency argument.
6535 (probably_never_executed_bb_p): Do not pass frequency.
6536 (probably_never_executed_edge_p): Likewise.
6537 (combine_predictions_for_bb): Check that profile is nonzero.
6538 (propagate_freq): Do not set frequency.
6539 (drop_profile): Simplify.
6540 (counts_to_freqs): Simplify.
6541 (expensive_function_p): Use to_frequency.
6542 (propagate_unlikely_bbs_forward): Simplify.
6543 (determine_unlikely_bbs): Simplify.
6544 (estimate_bb_frequencies): Add hack to silence graphite issues.
6545 (compute_function_frequency): Use ipa counts.
6546 (pass_profile::execute): Update.
6547 (rebuild_frequencies): Use counts only.
6548 (force_edge_cold): Use counts only.
6549 * profile-count.c (profile_count::dump): Dump new count types.
6550 (profile_count::differs_from_p): Check compatiblity.
6551 (profile_count::to_frequency): New function.
6552 (profile_count::to_cgraph_frequency): New function.
6553 * profile-count.h (struct function): Declare.
6554 (enum profile_quality): Add profile_guessed_local and
6555 profile_guessed_global0.
6556 (class profile_proability): Decrease number of bits to 29;
6557 update from_reg_br_prob_note and to_reg_br_prob_note.
6558 (class profile_count: Update comment; decrease number of bits
6559 to 61. Check compatibility.
6560 (profile_count::compatible_p): New private member function.
6561 (profile_count::ipa_p): New member function.
6562 (profile_count::operator<): Handle global zero correctly.
6563 (profile_count::operator>): Handle global zero correctly.
6564 (profile_count::operator<=): Handle global zero correctly.
6565 (profile_count::operator>=): Handle global zero correctly.
6566 (profile_count::nonzero_p): New member function.
6567 (profile_count::force_nonzero): New member function.
6568 (profile_count::max): New member function.
6569 (profile_count::apply_scale): Handle IPA scalling.
6570 (profile_count::guessed_local): New member function.
6571 (profile_count::global0): New member function.
6572 (profile_count::ipa): New member function.
6573 (profile_count::to_frequency): Declare.
6574 (profile_count::to_cgraph_frequency): Declare.
6575 * profile.c (OVERLAP_BASE): Delete.
6576 (compute_frequency_overlap): Delete.
6577 (compute_branch_probabilities): Do not use compute_frequency_overlap.
6578 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
6579 * sched-ebb.c (rank): Use counts only.
6580 * shrink-wrap.c (handle_simple_exit): Use counts only.
6581 (try_shrink_wrapping): Use counts only.
6582 (place_prologue_for_one_component): Use counts only.
6583 * tracer.c (find_best_predecessor): Use to_frequency.
6584 (find_trace): Use to_frequency.
6585 (tail_duplicate): Use to_frequency.
6586 * trans-mem.c (expand_transaction): Do not update frequency.
6587 * tree-call-cdce.c: Do not update frequency.
6588 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
6589 (gimple_merge_blocks): Likewise.
6590 (gimple_split_edge): Likewise.
6591 (gimple_duplicate_sese_region): Likewise.
6592 (gimple_duplicate_sese_tail): Likewise.
6593 (move_sese_region_to_fn): Likewise.
6594 (gimple_account_profile_record): Likewise.
6595 (insert_cond_bb): Likewise.
6596 * tree-complex.c (expand_complex_div_wide): Likewise.
6597 * tree-eh.c (lower_resx): Update profile.
6598 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
6600 (initialize_cfun): Do not initialize frequencies
6601 (freqs_to_counts): Delete.
6602 (copy_cfg_body): Ignore count parameter.
6603 (copy_body): Update.
6604 (expand_call_inline): Update count_max.
6605 (optimize_inline_calls): Update count_max.
6606 (tree_function_versioning): Update count_max.
6607 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
6608 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
6610 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
6611 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
6612 (try_peel_loop): Likewise.
6613 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
6615 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
6616 (tree_transform_and_unroll_loop): Do not use frequencies
6617 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
6618 Use reliable prediction only.
6619 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
6620 * tree-ssa-sink.c (select_best_block): Use to_frequency.
6621 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
6622 probability scaling.
6623 * tree-ssa-threadupdate.c (create_block_for_threading): Do
6624 not update frequency
6625 (any_remaining_duplicated_blocks): Likewise.
6626 (update_profile): Likewise.
6627 (estimated_freqs_path): Delete.
6628 (freqs_to_counts_path): Delete.
6629 (clear_counts_path): Delete.
6630 (ssa_fix_duplicate_block_edges): Likewise.
6631 (duplicate_thread_path): Likewise.
6632 * tree-switch-conversion.c (gen_inbound_check): Use counts.
6633 * tree-tailcall.c (decrease_profile): Do not update frequency.
6634 (eliminate_tail_call): Likewise.
6635 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
6636 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
6637 (optimize_mask_stores): Likewise.
6638 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
6639 * ubsan.c (ubsan_expand_null_ifn): Update profile.
6640 (ubsan_expand_ptr_ifn): Update profile.
6641 * value-prof.c (gimple_ic): Simplify.
6642 * value-prof.h (gimple_ic): Update prototype.
6643 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
6644 * ipa-inline.c (compute_uninlined_call_time): Be sure that
6646 (want_inline_self_recursive_call_p): Likewise.
6647 (resolve_noninline_speculation): Only cummulate defined counts.
6648 (inline_small_functions): Use nonzero_p.
6649 (ipa_inline): Do not access freed node.
6651 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
6653 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6654 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
6656 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
6658 * config/riscv/riscv.c (riscv_legitimize_move): Handle
6659 non-legitimate address.
6661 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6663 * config/rs6000/rs6000.md (*lt0_disi): Delete.
6664 (*lt0_<mode>di, *lt0_<mode>si): New.
6666 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
6668 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
6669 TARGET_PAIRED_FLOAT.
6671 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
6672 Jim Wilson <jim.wilson@linaro.org>
6674 * config/aarch64/aarch64-cores.def (saphira): New CPU.
6675 * config/aarch64/aarch64-tune.md: Regenerated.
6676 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
6677 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
6679 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
6681 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
6682 (arc_expand_prologue): Restore blink for millicode.
6683 * config/arc/linux.h (LINK_EH_SPEC): Defined.
6685 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
6688 * config/i386/i386.c (ix86_vector_duplicate_value): Use
6689 gen_vec_duplicate after forcing the scalar into a register.
6691 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
6693 * combine (try_combine): Print the insns input to try_combine to the
6696 2017-11-02 Steve Ellcey <sellcey@cavium.com>
6699 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
6700 Remove second argument from aarch64_process_target_attr call.
6701 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
6703 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
6705 (aarch64_handle_attr_arch): Remove second argument.
6706 (aarch64_handle_attr_cpu): Ditto.
6707 (aarch64_handle_attr_tune): Ditto.
6708 (aarch64_handle_attr_isa_flags): Ditto.
6709 (aarch64_process_one_target_attr): Ditto.
6710 (aarch64_process_target_attr): Ditto.
6711 (aarch64_option_valid_attribute_p): Remove second argument.
6712 on aarch64_process_target_attr call.
6714 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6716 * diagnostic.c: Include "selftest-diagnostic.h".
6717 (selftest::assert_location_text): New function.
6718 (selftest::test_diagnostic_get_location_text): New function.
6719 (selftest::diagnostic_c_tests): Call it.
6721 2017-11-02 David Malcolm <dmalcolm@redhat.com>
6723 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
6724 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
6725 (class selftest::test_diagnostic_context): Move to...
6726 * selftest-diagnostic.c: New file.
6727 * selftest-diagnostic.h: New file.
6729 2017-11-02 James Bowman <james.bowman@ftdichip.com>
6731 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
6732 offset range for FT32B.
6733 * config/ft32/ft32.h: option "mcompress" enables relaxation.
6734 * config/ft32/ft32.md: Add TARGET_NOPM.
6735 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
6736 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
6738 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
6740 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
6742 2017-11-02 Jeff Law <law@redhat.com>
6744 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
6745 virtual keyword on FINAL OVERRIDE members.
6747 * tree-ssa-propagate.h (ssa_propagation_engine): Group
6748 virtuals together. Add virtual destructor.
6749 (substitute_and_fold_engine): Similarly.
6751 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
6753 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
6755 2017-11-02 Richard Biener <rguenther@suse.de>
6757 PR tree-optimization/82795
6758 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
6760 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6762 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
6764 (gcc_GAS_CHECK_FEATURE): Remove.
6765 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
6766 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
6767 * configure: Regenerate.
6769 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
6771 * config/arc/arc.c (hwloop_optimize): Account for empty
6774 2017-11-02 Richard Biener <rguenther@suse.de>
6777 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
6778 Truncate ARRAY_REF index and element size.
6780 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6782 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
6784 2017-11-01 Jeff Law <law@redhat.com>
6786 * tree-ssa-ccp.c (ccp_folder): New class derived from
6787 substitute_and_fold_engine.
6788 (ccp_folder::get_value): New member function.
6789 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
6790 (ccp_fold_stmt): Remove prototype.
6791 (ccp_finalize): Call substitute_and_fold from the ccp_class.
6792 * tree-ssa-copy.c (copy_folder): New class derived from
6793 substitute_and_fold_engine.
6794 (copy_folder::get_value): Renamed from get_value.
6795 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
6796 * tree-vrp.c (vrp_folder): New class derived from
6797 substitute_and_fold_engine.
6798 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
6799 (vrp_folder::get_value): New member function.
6800 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
6801 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
6802 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
6803 provide a class interface to folder/substitute routines.
6804 (ssa_prop_fold_stmt_fn): Remove typedef.
6805 (ssa_prop_get_value_fn): Likewise.
6806 (subsitute_and_fold): Remove prototype.
6807 (replace_uses_in): Likewise.
6808 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
6809 Renamed from replace_uses_in. Call the virtual member function
6810 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
6811 (substitute_and_fold_dom_walker): Remove initialization of
6812 data member entries for calbacks. Add substitute_and_fold_engine
6813 member and initialize it.
6814 (substitute_and_fold_dom_walker::before_dom_children0: Use the
6815 member functions for get_value, replace_phi_args_in c
6816 replace_uses_in, and fold_stmt calls.
6817 (substitute_and_fold_engine::substitute_and_fold): Renamed from
6818 substitute_and_fold. Remove assert. Update ctor call.
6820 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
6821 (ssa_prop_visit_phi_fn): Likewise.
6822 (class ssa_propagation_engine): New class to provide an interface
6824 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
6826 (ssa_prop_visit_phi): Likewise.
6827 (ssa_propagation_engine::simulate_stmt): Moved into class.
6828 Call visit_phi/visit_stmt from the class rather than via
6829 file scoped static variables.
6830 (ssa_propagation_engine::simulate_block): Moved into class.
6831 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
6832 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
6833 set file scoped statics for the visit_stmt/visit_phi callbacks.
6834 * tree-complex.c (complex_propagate): New class derived from
6835 ssa_propagation_engine.
6836 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
6837 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
6838 (tree_lower_complex): Call ssa_propagate via the complex_propagate
6840 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
6841 ssa_propagation_engine.
6842 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
6843 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
6844 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
6845 * tree-ssa-copy.c (copy_prop): New class derived from
6846 ssa_propagation_engine.
6847 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
6848 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
6849 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
6850 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
6851 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
6852 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
6853 (execute_vrp): Call ssa_propagate from the vrp_prop class.
6855 2017-11-01 Jakub Jelinek <jakub@redhat.com>
6857 PR rtl-optimization/82778
6858 PR rtl-optimization/82597
6859 * compare-elim.c (struct comparison): Add in_a_setter field.
6860 (find_comparison_dom_walker::before_dom_children): Remove killed
6861 bitmap and df_simulate_find_defs call, instead walk the defs.
6862 Compute last_setter and initialize in_a_setter. Merge definitions
6863 with first initialization for a few variables.
6864 (try_validate_parallel): Use insn_invalid_p instead of
6865 recog_memoized. Return insn rather than just the pattern.
6866 (try_merge_compare): Fix up comment. Don't uselessly test if
6867 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
6869 (execute_compare_elim_after_reload): Remove df_chain_add_problem
6872 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6873 Alan Hayward <alan.hayward@arm.com>
6874 David Sherwood <david.sherwood@arm.com>
6876 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
6877 aarch64_hard_regno_nregs to get the number of registers
6880 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6881 Alan Hayward <alan.hayward@arm.com>
6882 David Sherwood <david.sherwood@arm.com>
6884 * config/aarch64/constraints.md (Upl): Rename to...
6886 * config/aarch64/aarch64.md
6887 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
6890 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6891 Alan Hayward <alan.hayward@arm.com>
6892 David Sherwood <david.sherwood@arm.com>
6894 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
6895 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
6898 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6899 Alan Hayward <alan.hayward@arm.com>
6900 David Sherwood <david.sherwood@arm.com>
6902 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
6903 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
6904 (aarch64_evpc_dup): Generate rtl direcly, rather than using
6906 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
6907 of a single element.
6908 * config/aarch64/iterators.md: Add a comment above the permute
6909 unspecs to say that they are generated directly by
6910 aarch64_expand_vec_perm_const.
6911 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
6913 2017-11-01 Nathan Sidwell <nathan@acm.org>
6915 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
6917 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
6919 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
6920 and medany code models, and describe what they do.
6922 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6924 Revert accidental duplicate:
6926 * combine.c (can_change_dest_mode): Reject changes in
6927 REGMODE_NATURAL_SIZE.
6929 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
6931 PR rtl-optimization/64682
6932 PR rtl-optimization/69567
6933 PR rtl-optimization/69737
6934 PR rtl-optimization/82683
6935 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
6936 register mentioned in the note, drop the note, unless it came from I3,
6937 in which case it should go to I3 again.
6939 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6941 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
6942 and return false if not.
6943 (clear_bytes_written_by, live_bytes_read): Update accordingly.
6945 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6947 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
6948 range is known to be empty.
6950 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6951 Alan Hayward <alan.hayward@arm.com>
6952 David Sherwood <david.sherwood@arm.com>
6954 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
6955 and CONST_VECTOR_NUNITS instead of computing the number of units from
6956 the byte sizes of the vector and element.
6957 (simplify_binary_operation_1): Likewise.
6958 (simplify_const_binary_operation): Likewise.
6959 (simplify_ternary_operation): Likewise.
6961 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6962 Alan Hayward <alan.hayward@arm.com>
6963 David Sherwood <david.sherwood@arm.com>
6965 * var-tracking.c (INT_MEM_OFFSET): Replace with...
6966 (int_mem_offset): ...this new function.
6967 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
6968 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
6969 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
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 * lower-subreg.c (interesting_mode_p): New function.
6977 (compute_costs, find_decomposable_subregs, decompose_register)
6978 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
6979 (resolve_clobber, dump_choices): Use it.
6981 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6982 Alan Hayward <alan.hayward@arm.com>
6983 David Sherwood <david.sherwood@arm.com>
6985 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
6987 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6988 Alan Hayward <alan.hayward@arm.com>
6989 David Sherwood <david.sherwood@arm.com>
6991 * alias.c (find_base_value, find_base_term): Only process integer
6992 truncations. Check the precision rather than the size.
6994 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
6995 Alan Hayward <alan.hayward@arm.com>
6996 David Sherwood <david.sherwood@arm.com>
6998 * machmode.h (is_narrower_int_mode): New function
6999 * optabs.c (expand_float, expand_fix): Use it.
7000 * dwarf2out.c (rotate_loc_descriptor): Likewise.
7002 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7003 Alan Hayward <alan.hayward@arm.com>
7004 David Sherwood <david.sherwood@arm.com>
7006 * rtl.h (narrower_subreg_mode): New function.
7007 * ira-color.c (update_costs_from_allocno): Use it.
7009 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7010 Alan Hayward <alan.hayward@arm.com>
7011 David Sherwood <david.sherwood@arm.com>
7013 * optabs-query.h (convert_optab_p): New function, split out from...
7014 (convert_optab_handler): ...here.
7015 (widening_optab_handler): Delete.
7016 (find_widening_optab_handler): Remove permit_non_widening parameter.
7017 (find_widening_optab_handler_and_mode): Likewise. Provide an
7018 override that operates on mode class wrappers.
7019 * optabs-query.c (widening_optab_handler): Delete.
7020 (find_widening_optab_handler_and_mode): Remove permit_non_widening
7021 parameter. Assert that the two modes are the same class and that
7022 the "from" mode is narrower than the "to" mode. Use
7023 convert_optab_handler instead of widening_optab_handler.
7024 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
7025 instead of widening_optab_handler.
7026 * expr.c (expand_expr_real_2): Update calls to
7027 find_widening_optab_handler.
7028 * optabs.c (expand_widen_pattern_expr): Likewise.
7029 (expand_binop_directly): Take the insn_code as a parameter.
7030 (expand_binop): Only call find_widening_optab_handler for
7031 conversion optabs; use optab_handler otherwise. Update calls
7032 to find_widening_optab_handler and expand_binop_directly.
7033 Use convert_optab_handler instead of widening_optab_handler.
7034 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
7035 find_widening_optab_handler and use scalar_mode rather than
7037 (convert_plusminus_to_widen): Likewise.
7039 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7040 Alan Hayward <alan.hayward@arm.com>
7041 David Sherwood <david.sherwood@arm.com>
7043 * machmode.h (fixed_size_mode): New class.
7044 * rtl.h (get_pool_mode): Return fixed_size_mode.
7045 * gengtype.c (main): Add fixed_size_mode.
7046 * target.def (get_raw_result_mode): Return a fixed_size_mode.
7047 (get_raw_arg_mode): Likewise.
7048 * doc/tm.texi: Regenerate.
7049 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
7050 * targhooks.c (default_get_reg_raw_mode): Likewise.
7051 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
7052 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
7053 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
7054 (msp430_get_raw_result_mode): Likewise.
7055 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
7056 * dbxout.c (dbxout_parms): Require fixed-size modes.
7057 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
7058 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7059 * omp-low.c (lower_oacc_reductions): Likewise.
7060 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
7061 (simplify_subreg): Update accordingly.
7062 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
7063 (force_const_mem): Update accordingly. Return NULL_RTX for modes
7064 that aren't fixed-size.
7065 (get_pool_mode): Return a fixed_size_mode.
7066 (output_constant_pool_2): Take a fixed_size_mode.
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 * doc/rtl.texi (vec_series): Document.
7073 (const): Say that the operand can be a vec_series.
7074 * rtl.def (VEC_SERIES): New rtx code.
7075 * rtl.h (const_vec_series_p_1): Declare.
7076 (const_vec_series_p): New function.
7077 * emit-rtl.h (gen_const_vec_series): Declare.
7078 (gen_vec_series): Likewise.
7079 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
7080 (gen_vec_series): Likewise.
7081 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
7082 * simplify-rtx.c (simplify_unary_operation): Handle negations
7084 (simplify_binary_operation_series): New function.
7085 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
7086 (test_vector_ops_series): New function.
7087 (test_vector_ops): Call it.
7088 * config/powerpcspe/altivec.md (altivec_lvsl): Use
7089 gen_const_vec_series.
7090 (altivec_lvsr): Likewise.
7091 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
7093 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7094 Alan Hayward <alan.hayward@arm.com>
7095 David Sherwood <david.sherwood@arm.com>
7097 * doc/rtl.texi (const): Update description of address constants.
7098 Say that vector constants are allowed too.
7099 * common.md (E, F): Use CONSTANT_P instead of checking for
7101 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
7102 checking for CONST_VECTOR.
7103 * expmed.c (make_tree): Use build_vector_from_val for a CONST
7105 * expr.c (expand_expr_real_2): Check for vector modes instead
7106 of checking for CONST_VECTOR.
7107 * rtl.h (const_vec_p): New function.
7108 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
7109 (unwrap_const_vec_duplicate): Handle them here too.
7111 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7112 David Malcolm <dmalcolm@redhat.com>
7113 Alan Hayward <alan.hayward@arm.com>
7114 David Sherwood <david.sherwood@arm.com>
7116 * rtl.h (vec_duplicate_p): New function.
7117 * selftest-rtl.c (assert_rtx_eq_at): New function.
7118 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
7119 (assert_rtx_eq_at): Declare.
7120 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
7121 * selftest-run-tests.c (selftest::run_tests): Call it.
7122 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
7123 (simplify_unary_operation_1): Recursively handle vector duplicates.
7124 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
7126 (simplify_subreg): Handle subregs of vector duplicates.
7127 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
7128 (selftest::simplify_rtx_c_tests): New functions.
7130 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7131 Alan Hayward <alan.hayward@arm.com>
7132 David Sherwood <david.sherwood@arm.com>
7134 * emit-rtl.h (gen_const_vec_duplicate): Declare.
7135 (gen_vec_duplicate): Likewise.
7136 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
7138 (gen_const_vector): ...here.
7139 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
7140 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
7141 whose elements are all equal.
7142 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
7143 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
7144 (simplify_relational_operation): Likewise.
7145 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
7147 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
7148 (aarch64_expand_vector_init): Likewise.
7149 * config/arm/arm.c (neon_vdup_constant): Likewise.
7150 (neon_expand_vector_init): Likewise.
7151 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
7152 (arm_block_set_unaligned_vect): Likewise.
7153 (arm_block_set_aligned_vect): Likewise.
7154 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
7155 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
7156 (expand_vec_perm_even_odd_pack): Likewise.
7157 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
7158 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
7159 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
7160 gen_const_vec_duplicate.
7161 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
7162 * config/mips/mips.c (mips_gen_const_int_vector): Use
7163 gen_const_vec_duplicate.
7164 (mips_expand_vector_init): Use CONST0_RTX.
7165 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
7166 (define_split): Use gen_const_vec_duplicate.
7167 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
7168 (define_split): Use gen_const_vec_duplicate.
7169 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
7170 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
7171 * config/spu/spu.c (spu_const): Likewise.
7173 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
7174 Alan Hayward <alan.hayward@arm.com>
7175 David Sherwood <david.sherwood@arm.com>
7177 * combine.c (can_change_dest_mode): Reject changes in
7178 REGMODE_NATURAL_SIZE.
7180 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
7182 * configure.ac (--enable-libssp): New.
7183 (gcc_cv_libc_provides_ssp): Check for explicit setting before
7184 trying to determine target-specific default. Adjust indentation.
7185 * configure: Regenerated.
7186 * doc/install.texi (Configuration): Expand --disable-libssp
7189 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
7191 config/i386/i386.c (ix86_expand_epilogue): Correct stack
7194 2017-10-31 Martin Jambor <mjambor@suse.cz>
7197 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
7199 2017-10-31 David Malcolm <dmalcolm@redhat.com>
7201 * auto-profile.c (autofdo_source_profile::read): Use
7202 UNKNOWN_LOCATION rather than 0.
7203 * diagnostic-core.h (warning_at_rich_loc): Rename to...
7204 (warning_at): ...this overload.
7205 (warning_at_rich_loc_n): Rename to...
7206 (warning_n): ...this overload.
7207 (error_at_rich_loc): Rename to...
7208 (error_at): ...this overload.
7209 (pedwarn_at_rich_loc): Rename to...
7210 (pedwarn): ...this overload.
7211 (permerror_at_rich_loc): Rename to...
7212 (permerror): ...this overload.
7213 (inform_at_rich_loc): Rename to...
7214 (inform): ...this overload.
7215 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
7216 (diagnostic_n_impl_richloc): Rename to...
7217 (diagnostic_n_impl): ...this rich_location *-based decl.
7218 (inform_at_rich_loc): Rename to...
7219 (inform): ...this, and add an assertion.
7220 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
7221 (warning_at_rich_loc): Rename to...
7222 (warning_at): ...this, and add an assertion.
7223 (warning_at_rich_loc_n): Rename to...
7224 (warning_n): ...this, and add an assertion.
7225 (warning_n): Update location_t-based implementation for removal of
7226 location_t-based diagnostic_n_impl.
7227 (pedwarn_at_rich_loc): Rename to...
7228 (pedwarn): ...this, and add an assertion.
7229 (permerror_at_rich_loc): Rename to...
7230 (permerror): ...this, and add an assertion.
7231 (error_n): Update for removal of location_t-based diagnostic_n_impl.
7232 (error_at_rich_loc): Rename to...
7233 (error_at): ...this, and add an assertion.
7234 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
7235 (driver::do_spec_on_infiles): Likewise.
7236 * substring-locations.c (format_warning_va): Update for renaming
7237 of inform_at_rich_loc.
7239 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
7241 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
7242 _Float<N>X built-in functions so that the variant without the
7243 "__builtin_" prefix is only enabled for the GNU C and Objective C
7244 languages when they are in non-strict ANSI/ISO mode.
7245 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
7246 * target.def (floatn_builtin_p): Add a target hook to control
7247 whether _Float<N> and _Float<N>X built-in functions without the
7248 "__builtin_" prefix are enabled, and return true for C and
7249 Objective C in the default hook. Include langhooks.h in
7251 * targhooks.h (default_floatn_builtin_p): Likewise.
7252 * targhooks.c (default_floatn_builtin_p): Likewise.
7253 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
7254 floatn_builtin_p target hook.
7255 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
7257 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
7258 Eric Botcazou <ebotcazou@adacore.com>
7260 PR rtl-optimization/81803
7261 * lra-constraints.c (curr_insn_transform): Also reload the whole
7262 register for a strict subreg no wider than a word if this is for
7263 a WORD_REGISTER_OPERATIONS target.
7265 2017-10-31 Jason Merrill <jason@redhat.com>
7267 * gdbinit.in: Skip over inlines from timevar.h.
7269 2017-10-31 Martin Liska <mliska@suse.cz>
7271 * doc/gcov.texi: Document new option.
7272 * gcov.c (print_usage): Likewise print it.
7273 (process_args): Support the argument.
7274 (format_count): New function.
7275 (format_gcov): Use the function.
7277 2017-10-31 Martin Liska <mliska@suse.cz>
7279 * gcov.c (struct name_map): do not use typedef.
7280 Define operator== and operator<.
7281 (name_search): Remove.
7282 (name_sort): Remove.
7283 (main): Do not allocate names.
7284 (process_file): Add vertical space.
7285 (generate_results): Use std::find.
7286 (release_structures): Do not release memory.
7287 (find_source): Use std::find.
7289 2017-10-31 Martin Liska <mliska@suse.cz>
7291 * gcov.c (struct line_info): Remove it's typedef.
7292 (line_info::line_info): Add proper ctor.
7293 (line_info::has_block): Do not use a typedef.
7294 (struct source_info): Do not use typedef.
7295 (circuit): Likewise.
7296 (get_cycles_count): Likewise.
7297 (output_intermediate_file): Iterate via vector iterator.
7298 (add_line_counts): Use std::vector methods.
7299 (accumulate_line_counts): Likewise.
7300 (output_lines): Likewise.
7302 2017-10-31 Martin Liska <mliska@suse.cz>
7304 * gcov.c (struct source_info): Remove typedef.
7305 (source_info::source_info): Add proper ctor.
7306 (accumulate_line_counts): Use struct, not it's typedef.
7307 (output_gcov_file): Likewise.
7308 (output_lines): Likewise.
7309 (main): Do not allocate an array.
7310 (output_intermediate_file): Use size of vector container.
7311 (process_file): Resize the vector.
7312 (generate_results): Do not preallocate, use newly added vector
7314 (release_structures): Do not release sources.
7315 (find_source): Use vector methods.
7316 (add_line_counts): Do not use typedef.
7318 2017-10-31 Martin Liska <mliska@suse.cz>
7320 * doc/gcov.texi: Document that.
7321 * gcov.c (add_line_counts): Mark lines with a non-executed
7323 (output_line_beginning): Handle such lines.
7324 (output_lines): Pass new argument.
7325 (output_intermediate_file): Print it in intermediate format.
7327 2017-10-31 Martin Liska <mliska@suse.cz>
7329 * color-macros.h: New file.
7330 * diagnostic-color.c: Factor out color related to macros to
7332 * doc/gcov.texi: Document -k option.
7333 * gcov.c (INCLUDE_STRING): Include string.h.
7334 (print_usage): Add -k option.
7335 (process_args): Parse it.
7336 (pad_count_string): New function.
7337 (output_line_beginning): Likewise.
7338 (DEFAULT_LINE_START): New macro.
7339 (output_lines): Support color output.
7341 2017-10-31 Martin Liska <mliska@suse.cz>
7343 PR gcov-profile/82633
7344 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
7345 their interaction with GCOV infrastructure.
7346 * configure.ac: Add -fkeep-{inline,static}-functions to
7348 * configure: Regenerate.
7350 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
7353 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
7355 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
7358 * config/rs6000/rs6000.md (allocate_stack): Force update interval
7359 into a register if it does not fit into an immediate offset field.
7361 2017-10-31 Olivier Hainque <hainque@adacore.com>
7363 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
7365 2017-10-31 Julia Koval <julia.koval@intel.com>
7367 * config.gcc: Add gfniintrin.h.
7368 * config/i386/gfniintrin.h: New.
7369 * config/i386/i386-builtin-types.def
7370 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
7371 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
7372 __builtin_ia32_vgf2p8affineinvqb_v32qi,
7373 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
7374 __builtin_ia32_vgf2p8affineinvqb_v16qi,
7375 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
7376 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
7377 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
7378 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
7379 V64QI_FTYPE_V64QI_V64QI_INT): New types.
7380 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
7381 * config/i386/immintrin.h: Include gfniintrin.h.
7382 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
7384 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
7386 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
7388 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7390 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
7391 (arm_ashldi3_1bit): Remove pattern.
7392 (ashrdi3): Remove shift by 1 expansion.
7393 (arm_ashrdi3_1bit): Remove pattern.
7394 (lshrdi3): Remove shift by 1 expansion.
7395 (arm_lshrdi3_1bit): Remove pattern.
7396 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
7397 cost of ashldi3 by 1.
7398 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
7399 (<shift>di3_neon): Likewise.
7401 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7403 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
7404 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
7405 and (*aarch64_simd_mov<VQ:mode>).
7406 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
7407 pattern alternative.
7408 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
7409 attributes to match pattern alternative.
7411 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
7413 * config.gcc (powerpc*-*-*): Add emmintrin.h.
7414 * config/rs6000/emmintrin.h: New file.
7415 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
7417 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
7419 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
7420 * (movdi_vfp_cortexa8): Remove pattern.
7422 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7424 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
7426 (Specific, alpha*-dec-osf5.1): Remove.
7427 (Specific, mips-sgi-irix5): Remove.
7428 (Specific, mips-sgi-irix6): Remove.
7430 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7433 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
7434 arguments to clear_bit_region_be.
7436 2017-10-30 Jim Wilson <wilson@tuliptree.org>
7438 * gimplify.c: Include memmodel.h.
7440 2017-10-30 Martin Jambor <mjambor@suse.cz>
7442 * omp-grid.c (grid_attempt_target_gridification): Also insert a
7443 condition whether loop should be executed at all.
7445 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
7447 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7448 gimple folding of vec_madd() intrinsics.
7449 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
7450 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
7451 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
7453 2017-10-30 Richard Biener <rguenther@suse.de>
7455 PR tree-optimization/82762
7457 2017-10-23 Richard Biener <rguenther@suse.de>
7459 PR tree-optimization/82129
7461 2017-08-01 Richard Biener <rguenther@suse.de>
7463 PR tree-optimization/81181
7464 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
7465 (compute_antic): ... end of iteration here.
7467 2017-10-30 Joseph Myers <joseph@codesourcery.com>
7469 * doc/invoke.texi (C Dialect Options): Document -std=c17,
7470 -std=iso9899:2017 and -std=gnu17.
7471 * doc/standards.texi (C Language): Document C17 support.
7472 * doc/cpp.texi (Overview): Mention -std=c17.
7473 (Standard Predefined Macros): Document C11 and C17 values of
7474 __STDC_VERSION__. Do not refer to C99 support as incomplete.
7475 * doc/extend.texi (Inline): Do not list individual options for
7476 standards newer than C99.
7477 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
7479 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
7482 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
7484 * asan.c (asan_finish_file): Align asan globals array by shadow
7487 2017-10-30 Jakub Jelinek <jakub@redhat.com>
7490 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
7491 (struct store_immediate_info): Add bitregion_start and bitregion_end
7493 (store_immediate_info::store_immediate_info): Add brs and bre
7494 arguments and initialize bitregion_{start,end} from those.
7495 (struct merged_store_group): Add bitregion_start, bitregion_end,
7496 align_base and mask fields. Drop unnecessary struct keyword from
7497 struct store_immediate_info. Add do_merge method.
7498 (clear_bit_region_be): Use memset instead of loop storing zeros.
7499 (merged_store_group::do_merge): New method.
7500 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
7501 stores as long as the surrounding bitregions have no gaps.
7502 (merged_store_group::merge_overlapping): Use do_merge.
7503 (merged_store_group::apply_stores): Test that bitregion_{start,end}
7504 is byte aligned, rather than requiring that start and width are
7505 byte aligned. Drop unnecessary struct keyword from
7506 struct store_immediate_info. Allocate and populate also mask array.
7507 Make start of the arrays relative to bitregion_start rather than
7508 start and size them according to bitregion_{end,start} difference.
7509 (struct imm_store_chain_info): Drop unnecessary struct keyword from
7510 struct store_immediate_info.
7511 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
7512 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
7513 struct keyword from struct store_immediate_info.
7514 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
7515 between stores as long as the surrounding bitregions have no gaps.
7517 (struct split_store): Add orig non-static data member.
7518 (split_store::split_store): Initialize orig to false.
7519 (find_constituent_stmts): Return store_immediate_info *, non-NULL
7520 if there is exactly a single original stmt. Change stmts argument
7521 to pointer from reference, if NULL, don't push anything to it. Add
7522 first argument, use it to optimize skipping over orig stmts that
7523 are known to be before bitpos already. Simplify.
7524 (split_group): Return unsigned int count how many stores are or
7525 would be needed rather than a bool. Add allow_unaligned argument.
7526 Change split_stores argument from reference to pointer, if NULL,
7527 only do a dry run computing how many stores would be produced.
7528 Rewritten algorithm to use both alignment and misalign if
7529 !allow_unaligned and handle bitfield stores with gaps.
7530 (imm_store_chain_info::output_merged_store): Set start_byte_pos
7531 from bitregion_start instead of start. Compute allow_unaligned
7532 here, if true, do 2 split_group dry runs to compute which one
7533 produces fewer stores and prefer aligned if equal. Punt if
7534 new count is bigger or equal than original before emitting any
7535 statements, rather than during that. Remove no longer needed
7536 new_ssa_names tracking. Replace num_stmts with
7537 split_stores.length (). Use 32-bit stack allocated entries
7538 in split_stores auto_vec. Try to reuse original store lhs/rhs1
7539 if possible. Handle bitfields with gaps.
7540 (pass_store_merging::execute): Ignore bitsize == 0 stores.
7541 Compute bitregion_{start,end} for the stores and construct
7542 store_immediate_info with that. Formatting fixes.
7544 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
7547 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
7548 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
7550 2017-10-29 Jim Wilson <wilson@tuliptree.org>
7552 * gimplify.c: Include tm_p.h.
7554 * common.opt (gcoff): Re-add as ignored option.
7555 (gcoff1, gcoff2, gcoff3): Likewise.
7557 * Makefile.in (OBJS): Delete sdbout.o.
7558 (GTFILES): Delete $(srcdir)/sdbout.c.
7559 * debug.h: Delete sdb_debug_hooks.
7560 * final.c: Delete sdbout.h include.
7561 (final_scan_insn): Delete SDB_DEBUG check.
7562 (rest_of_clean_state): Likewise.
7563 * output.h: Delete sdb_begin_function_line.
7566 * toplev.c: Delete sdbout.h include.
7567 (process_options): Delete SDB_DEBUG check.
7568 * tree-core.h (tree_type_common): Delete pointer field of
7570 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
7571 TYPE_SYMTAB_POINTER.
7572 * tree.h (TYPE_SYMTAB_POINTER): Delete.
7573 (TYPE_SYMTAB_IS_POINTER): Delete.
7574 (TYPE_SYMTAB_IS_DIE): Renumber.
7575 * xcoffout.c: Refer to former sdbout.c file.
7576 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
7578 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
7579 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
7580 (Debugging Options): Delete -gcoff.
7581 (-gstabs): Delete SDB reference.
7583 (-gcoff@var{level}): Delete.
7584 * doc/passes.texi (Debugging information output): Delete SDB and
7585 sdbout.c references.
7586 * doc/tm.texi: Regenerate.
7587 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
7588 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
7590 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
7591 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
7592 (SDB_DEBUGGING_INFO): Delete.
7593 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
7594 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
7595 * target.def (output_source_filename): Delete COFF reference.
7597 * common.opt (gcoff): Delete.
7598 (gxcoff+): Update Negative chain.
7599 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
7601 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
7602 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
7603 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
7605 (expand_function_start): Change sdb reference to past tense.
7606 (expand_function_end): Change sdb reference to past tense.
7607 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
7608 * opts.c (debug_type_names): Delete coff entry.
7609 (common_handle_option): Delete OPT_gcoff case.
7610 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
7612 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
7613 * config/cris/cris.h: Delete SDB reference in comment.
7614 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
7615 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
7616 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
7617 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
7619 (ix86_expand_prologue): Likewise.
7620 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
7621 * config/ia64/ia64.h: Likewise.
7622 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
7623 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
7625 * config/mmix/mmix.h: Likewise.
7626 * config/nds32/nds32.c: Likewise.
7627 * config/stormy/storym16.h: Likewise.
7628 * config/visium/visium.h: Likewise.
7629 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
7631 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
7633 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
7634 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
7635 FRAME_POINTER_REGNUM point at high end of local var area.
7637 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
7639 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
7640 Move comment around. Do not reset best_edge for a copiable
7641 destination if the copy would cause a partition change.
7642 (better_edge_p): Remove redundant check.
7644 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7646 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
7648 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
7650 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
7651 for math functions that have _Float<N> and _Float<N>X variants.
7652 (mathfn_built_in_2): Add support for math functions that have
7653 _Float<N> and _Float<N>X variants.
7654 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
7655 (expand_builtin_mathfn_ternary): Add support for fma with
7656 _Float<N> and _Float<N>X variants.
7657 (expand_builtin): Likewise.
7658 (fold_builtin_3): Likewise.
7659 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
7660 create math function _Float<N> and _Float<N>X variants as external
7662 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
7663 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
7664 the __builtin_ prefix and if not strict ansi, without the prefix.
7665 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
7666 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
7667 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
7668 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
7669 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
7670 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
7671 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
7672 function signatures for fma _Float<N> and _Float<N>X variants.
7673 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
7674 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
7675 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
7676 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
7677 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
7678 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
7679 * gencfn-macros.c (print_case_cfn): Add support for math functions
7680 that have _Float<N> and _Float<N>X variants.
7681 (print_define_operator_list): Likewise.
7682 (fltfn_suffixes): Likewise.
7684 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
7685 for math functions that have _Float<N> and _Float<N>X variants.
7686 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
7687 and _Float<N>X variants.
7688 (COPYSIGN): Likewise.
7691 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
7692 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
7694 (integer_valued_read_call_p): Likewise.
7695 * fold-const-call.c (fold_const_call_ss): Likewise.
7696 (fold_const_call_sss): Add support for copysign, fmin, and fmax
7697 _Float<N> and _Float<N>X variants.
7698 (fold_const_call_ssss): Add support for fma _Float<N> and
7699 _Float<N>X variants.
7700 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
7701 support for copysign and fma _Float<N> and _Float<N>X variants.
7702 (backprop::process_builtin_call_use): Likewise.
7703 * tree-call-cdce.c (can_test_argument_range); Add support for
7704 sqrt _Float<N> and _Float<N>X variants.
7705 (edom_only_function): Likewise.
7706 (get_no_error_domain): Likewise.
7707 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
7708 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
7709 copysign _Float<N> and _Float<N>X variants.
7710 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
7711 handled by machine independent code.
7712 (FMAF128): Likewise.
7713 * doc/cpp.texi (Common Predefined Macros): Document defining
7714 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
7715 fma _Float<N> and _Float<N>X variants.
7717 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
7720 * config/i386/i386-modes.def (CCFPU): Remove definition.
7721 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
7722 (ix86_cc_modes_compatible): Ditto.
7723 (ix86_expand_carry_flag_compare): Ditto.
7724 (ix86_expand_int_movcc): Ditto.
7725 (ix86_expand_int_addcc): Ditto.
7726 (ix86_reverse_condition): Ditto.
7727 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
7728 Return true/false for unordered/ordered fp comparisons.
7729 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
7730 (ix86_prepare_fp_compare_args): Update for rename.
7731 (ix86_expand_fp_compare): Update for rename. Generate unordered
7732 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
7733 (ix86_expand_sse_compare_and_jump): Ditto.
7734 * config/i386/predicates.md (fcmov_comparison_operator):
7735 Remove CCFPU mode handling.
7736 (ix86_comparison_operator): Ditto.
7737 (ix86_carry_flag_operator): Ditto.
7738 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
7739 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
7740 (*cmpu<mode>_cc_i387): Ditto.
7741 (FPCMP): Remove mode iterator.
7742 (unord): Remove mode attribute.
7743 (unord_subst): New define_subst transformation
7744 (unord): New define_subst attribute.
7746 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
7747 (*cmpi<unord>xf_i387): Ditto.
7748 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
7749 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
7750 using unord_subst transformation.
7751 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
7752 (round_saeonly): Also handle CCFP mode.
7753 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
7754 Remove UNSPEC_SAHF unspec handling.
7756 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7758 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
7760 2017-10-27 Jeff Law <law@redhat.com>
7762 * gimple-ssa-sprintf.c: Include domwalk.h.
7763 (class sprintf_dom_walker): New class, derived from dom_walker.
7764 (sprintf_dom_walker::before_dom_children): New function.
7765 (struct call_info): Moved into sprintf_dom_walker class
7766 (compute_formath_length, handle_gimple_call): Likewise.
7767 (sprintf_length::execute): Call the dominator walker rather
7768 than walking the statements.
7770 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
7771 gimple statement locations.
7772 (check_array_bounds): Corresponding changes. Get the statement's
7773 location directly from wi->stmt.
7775 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
7778 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
7780 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
7782 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
7783 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
7785 2017-10-27 Jakub Jelinek <jakub@redhat.com>
7788 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
7789 * config/i386/i386.c (maybe_get_pool_constant): Removed.
7790 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
7791 maybe_get_pool_constant.
7792 * config/i386/predicates.md (zero_extended_scalar_load_operand):
7795 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7797 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
7798 2.26 caveat. Update gas and gld versions.
7799 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
7802 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7804 * cgraph.h (set_malloc_flag): Declare.
7805 * cgraph.c (set_malloc_flag_1): New function.
7806 (set_malloc_flag): Likewise.
7807 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
7808 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
7810 (read_ipa_call_summary): Add support for reading is_return_callee.
7811 (write_ipa_call_summary): Stream is_return_callee.
7812 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
7813 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
7814 ipa-prop.h, ipa-fnsummary.h.
7815 (pure_const_names): Change to static.
7816 (malloc_state_e): Define.
7817 (malloc_state_names): Define.
7818 (funct_state_d): Add field malloc_state.
7819 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
7820 (check_retval_uses): New function.
7821 (malloc_candidate_p): Likewise.
7822 (analyze_function): Add support for malloc attribute.
7823 (pure_const_write_summary): Stream malloc_state.
7824 (pure_const_read_summary): Add support for reading malloc_state.
7825 (dump_malloc_lattice): New function.
7826 (propagate_malloc): New function.
7827 (warn_function_malloc): New function.
7828 (ipa_pure_const::execute): Call propagate_malloc and
7829 ipa_free_fn_summary.
7830 (pass_local_pure_const::execute): Add support for malloc attribute.
7831 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
7832 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
7834 2017-10-27 Martin Liska <mliska@suse.cz>
7836 PR gcov-profile/82457
7837 * doc/invoke.texi: Document that one needs a non-strict ISO mode
7838 for fork-like functions to be properly instrumented.
7840 2017-10-27 Richard Biener <rguenther@suse.de>
7843 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
7844 info when we redirected EH.
7846 2017-10-26 Michael Collison <michael.collison@arm.com>
7848 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
7850 (<optab>_trunchf<GPI:mode>2: New pattern.
7851 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
7852 * config/aarch64/iterators.md (wv): New mode attribute.
7853 (vf, VF): New mode attributes.
7854 (vgp, VGP): New mode attributes.
7855 (s): Update attribute with SImode and DImode prefixes.
7857 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7859 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
7860 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
7861 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
7862 (nios2_option_overide): Initialize it. Don't allow R0-relative
7863 addressing with PIC.
7864 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
7865 (nios2_symbolic_constant_p): Likewise.
7866 (nios2_legitimate_address_p): Likewise.
7867 (nios2_r0rel_section_name_p): New.
7868 (nios2_symbol_ref_in_r0rel_data_p): New.
7869 (nios2_emit_move_sequence): Handle r0rel_constant_p.
7870 (r0rel_constant_p): New.
7871 (nios2_print_operand_address): Handle r0rel_constant_p.
7872 (nios2_cdx_narrow_form_p): Likewise.
7873 * config/nios2/nios2.opt (mr0rel-sec=): New option.
7874 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
7875 (Nios II Options): Document -mr0rel-sec.
7877 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
7879 * config/nios2/nios2.c: Include xregex.h.
7880 (nios2_gprel_sec_regex): New.
7881 (nios2_option_overide): Initialize it. Don't allow GP-relative
7882 addressing with PIC.
7883 (nios2_small_section_name_p): Check for regex match.
7884 * config/nios2/nios2.opt (mgprel-sec=): New option.
7885 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
7886 (Nios II Options): Document -mgprel-sec.
7888 2017-10-26 Jim Wilson <wilson@tuliptree.org>
7890 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
7892 2017-10-26 Tom de Vries <tom@codesourcery.com>
7894 PR tree-optimization/82707
7895 * gimple.c (gimple_copy): Fix unsharing of
7896 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
7898 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
7900 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
7901 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
7902 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
7903 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
7904 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
7905 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
7906 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
7907 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
7908 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
7909 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
7910 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
7911 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
7912 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
7913 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
7914 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
7915 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
7916 _mm512_mask_cmpunord_ps_mask): New intrinsics.
7918 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7920 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
7922 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
7923 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
7924 warning to rs6000.c. Remove the Undocumented flag, since it has
7926 (-mabi=ibmlongdouble): Likewise.
7927 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
7928 already set, set the default format for long double.
7929 (rs6000_debug_reg_global): Print whether long double is IBM or
7931 (rs6000_option_override_internal): Rework setting long double
7932 format. Only warn if the user is changing the long double default
7933 and they did not use -Wno-psabi.
7934 * doc/invoke.texi (PowerPC options): Update the documentation for
7935 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
7937 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7938 Alan Hayward <alan.hayward@arm.com>
7939 David Sherwood <david.sherwood@arm.com>
7941 * rtl.h (wider_subreg_mode): New function.
7942 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
7943 rather than an unsigned int *.
7944 * ira-color.c (regno_max_ref_width): Replace with...
7945 (regno_max_ref_mode): ...this new variable.
7946 (coalesced_pseudo_reg_slot_compare): Update accordingly.
7947 Use wider_subreg_mode.
7948 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
7949 rather than an unsigned int *.
7950 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
7951 (process_alt_operands): Likewise.
7952 (invariant_p): Likewise.
7953 * lra-spills.c (assign_mem_slot): Likewise.
7954 (add_pseudo_to_slot): Likewise.
7955 * lra.c (collect_non_operand_hard_regs): Likewise.
7956 (add_regs_to_insn_regno_info): Likewise.
7957 * reload1.c (regno_max_ref_width): Replace with...
7958 (regno_max_ref_mode): ...this new variable.
7959 (reload): Update accordingly. Update call to
7960 ira_sort_regnos_for_alter_reg.
7961 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
7962 (init_eliminable_invariants): Update to use regno_max_ref_mode.
7963 (scan_paradoxical_subregs): Likewise.
7965 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7967 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
7968 (aarch64_frame): Add emit_frame_chain boolean.
7969 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
7970 Move eh_return case to aarch64_layout_frame.
7971 (aarch64_layout_frame): Initialize emit_frame_chain.
7972 (aarch64_expand_prologue): Use emit_frame_chain.
7974 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7976 * config/aarch64/aarch64.c (aarch64_layout_frame):
7977 Ensure LR is always stored at the bottom of the callee-saves.
7978 Remove rarely used frame layout which saves callee-saves at top of
7979 frame, so the store of LR can be used as a valid probe in all cases.
7981 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
7983 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
7984 Improve unaligned TImode/TFmode base/offset split.
7986 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
7987 Alan Hayward <alan.hayward@arm.com>
7988 David Sherwood <david.sherwood@arm.com>
7990 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
7991 * combine.c (find_single_use_1): Likewise.
7992 (expand_field_assignment): Likewise.
7993 (move_deaths): Likewise.
7994 * lra-constraints.c (simplify_operand_subreg): Likewise.
7995 (curr_insn_transform): Likewise.
7996 * lra.c (collect_non_operand_hard_regs): Likewise.
7997 (add_regs_to_insn_regno_info): Likewise.
7998 * rtlanal.c (reg_referenced_p): Likewise.
7999 (covers_regno_no_parallel_p): Likewise.
8001 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8003 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
8004 Don't print any bits outside the precision of the value.
8005 * wide-int.cc (test_printing): Add some new tests.
8007 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8009 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
8010 supports -xbrace_comment option.
8011 * configure: Regenerate.
8012 * config.in: Regenerate.
8013 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
8014 (ASM_CPU_SPEC): Use it.
8016 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
8018 * target.def (static_rtx_alignment): New hook.
8019 * targhooks.h (default_static_rtx_alignment): Declare.
8020 * targhooks.c (default_static_rtx_alignment): New function.
8021 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
8022 * doc/tm.texi: Regenerate.
8023 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
8024 instead of targetm.constant_alignment. Remove call to
8026 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8027 (cris_preferred_mininum_alignment): New function, split out from...
8028 (cris_constant_alignment): ...here.
8029 (cris_static_rtx_alignment): New function.
8030 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
8032 (ix86_constant_alignment): ...here.
8033 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8034 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8035 (mmix_static_rtx_alignment): New function.
8036 * config/spu/spu.c (spu_static_rtx_alignment): New function.
8037 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
8039 2017-10-26 Tamar Christina <tamar.christina@arm.com>
8042 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
8043 Add flag_trapping_math and flag_fp_int_builtin_inexact.
8045 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
8047 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
8048 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
8049 mark as a sign-extending load.
8050 (local_pic_load_u): Define.
8052 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
8055 * fold-const.c (operand_equal_for_comparison_p): Also return true
8056 if ARG0 is a simple variant of ARG1 with narrower precision.
8057 (fold_ternary_loc): Always pass unstripped operands to the predicate.
8059 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
8061 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
8063 * i386.h (processor_costs): Add gather_static, gather_per_elt,
8064 scatter_static, scatter_per_elt.
8065 * x86-tune-costs.h: Add new cost entries.
8067 2017-10-25 Richard Biener <rguenther@suse.de>
8069 * tree-ssa-sccvn.h (vn_eliminate): Declare.
8070 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
8071 class pass_fre): Move to ...
8072 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
8073 class pass_fre): ... here and adjust for statistics.
8075 2017-10-25 Jakub Jelinek <jakub@redhat.com>
8078 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
8079 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
8080 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
8081 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
8084 2017-10-25 Richard Biener <rguenther@suse.de>
8086 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
8087 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
8088 eliminate_push_avail, eliminate_insert): Move inside...
8089 (class eliminate_dom_walker): ... this class in preparation
8091 (fini_eliminate): Remove by merging with ...
8092 (eliminate): ... this function. Adjust for class changes.
8093 (pass_pre::execute): Remove fini_eliminate call.
8094 (pass_fre::execute): Likewise.
8096 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8099 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
8100 (VPERMI2, VPERMI2I): New mode iterators.
8101 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
8102 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
8104 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
8105 mode iterator. Remove 3 old define_insn patterns.
8106 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
8107 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
8108 VPERMI2 mode iterator, remove the other two expanders.
8109 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
8110 to use VPERMI2 mode iterator, add another alternative for vpermi2*
8111 instructions, remove the other two patterns.
8112 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
8113 mode iterator, remove the other two patterns.
8114 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
8115 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
8116 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
8117 and adjust argument order accordingly.
8118 (ix86_expand_vec_perm): Adjust caller.
8119 (expand_vec_perm_1): Likewise.
8120 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
8121 (expand_vec_perm_vpermt2_vpshub2): ... this.
8122 (ix86_expand_vec_perm_const_1): Adjust caller.
8123 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
8126 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
8127 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
8128 (vec_shl_<mode>): Remove unused expander.
8129 (avx512bw_<shift_insn><mode>3): New define_insn.
8130 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
8131 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
8133 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
8136 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
8139 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8141 PR rtl-optimization/82396
8142 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
8143 (autopref_multipass_init): Simplify initialization.
8144 (autopref_rank_data): Simplify sort order.
8145 * gcc/sched-int.h (autopref_multipass_data_): Remove
8146 multi_mem_insn_p, min_offset and max_offset.
8148 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
8151 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8152 Check special value of flag_omit_frame_pointer.
8153 (aarch64_can_eliminate): Likewise.
8154 (aarch64_override_options_after_change_1): Simplify handling of
8155 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
8157 2017-10-24 Richard Biener <rguenther@suse.de>
8159 PR tree-optimization/82697
8160 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
8161 zero for conditional load and unconditional store.
8163 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8165 * doc/install.texi: Document bootstrap-cet.
8167 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
8170 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
8171 ENDBR instruction at function entrance if function is only
8174 2017-10-24 Jakub Jelinek <jakub@redhat.com>
8177 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
8178 patterns to better describe from which operation the CF is computed.
8179 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
8180 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
8181 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
8182 is 0, use _0 suffixed expanders instead of emitting a comparison
8185 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
8187 * config/i386/i386.md(*movsf_internal, *movdf_internal):
8188 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
8190 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
8193 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
8194 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
8195 * loop-iv.c (iv_get_reaching_def): Likewise.
8196 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
8197 variable is promoted and the partition contains undefined values.
8199 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8201 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
8203 (nios2_address_cost): Define.
8204 (nios2_legitimize_address): Recognize (exp + constant) directly.
8205 (TARGET_ADDRESS_COST): Define.
8207 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8209 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
8210 (nios2_symbolic_memory_operand_p): Declare.
8211 (nios2_split_large_constant): Declare.
8212 (nios2_split_symbolic_memory_operand): Declare.
8213 * config/nios2/nios2.c: Adjust includes.
8214 (nios2_symbolic_constant_allowed): New.
8215 (nios2_symbolic_constant_p): New.
8216 (nios2_plus_symbolic_constant_p): New.
8217 (nios2_valid_addr_expr_p): Recognize addresses involving
8219 (nios2_legitimate_address_p): Likewise, also LO_SUM.
8220 (nios2_symbolic_memory_operand_p): New.
8221 (nios2_large_constant_p): New.
8222 (nios2_split_large_constant): New.
8223 (nios2_split_plus_large_constant): New.
8224 (nios2_split_symbolic_memory_operand): New.
8225 (nios2_legitimize_address): Code refactoring. Handle addresses
8226 involving symbolic constants.
8227 (nios2_emit_move_sequence): Likewise.
8228 (nios2_print_operand): Improve error output.
8229 (nios2_print_operand_address): Handle LO_SUM.
8230 (nios2_cdx_narrow_form_p): Likewise.
8231 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
8232 operands involving symbolic constants.
8233 (movhi_internal, movsi_internal): Likewise.
8234 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
8235 (extendhisi2, extendqi<mode>2): Likewise.
8237 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8239 * tree-pass.h (PROP_rtl_split_insns): Define.
8240 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
8242 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
8244 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
8246 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8249 * target.def (const_not_ok_for_debug_p): Default to
8250 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
8251 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
8252 * targhooks.c (default_const_not_ok_for_debug_p): New function.
8253 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
8254 which targetm.const_not_ok_for_debug_p returned true.
8255 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
8257 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
8259 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
8260 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
8261 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
8263 (ix86_const_not_ok_for_debug_p): New function.
8264 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
8265 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
8267 2017-10-23 David Malcolm <dmalcolm@redhat.com>
8270 * system.h: Conditionally include "unique-ptr.h" if
8271 INCLUDE_UNIQUE_PTR is defined.
8272 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
8273 of defining INCLUDE_UNIQUE_PTR before including "system.h".
8275 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
8277 * config/rl78/rl78.md: New define_expand "subdi3".
8279 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
8282 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
8283 DF_REF_INSN if DF_REF_INSN_INFO is false.
8285 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
8287 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
8288 xmm_move instead of sse_move.
8289 (sse_store_index): New function.
8290 (ix86_register_move_cost): Be more sensible about mismatch stall;
8291 model AVX moves correctly; make difference between sse->integer and
8293 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
8294 moves; make difference between SSE and AVX.
8295 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
8296 and zmm_move. Increase size of sse load and store tables;
8297 add unaligned load and store tables; add ssemmx_to_integer.
8298 * x86-tune-costs.h: Update all entries according to real
8299 move latencies from Agner Fog's manual and chip documentation.
8301 2017-10-23 Jakub Jelinek <jakub@redhat.com>
8304 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
8305 * config/i386/constraints.md (Wf): New constraint.
8306 * config/i386/i386.md (UNSPEC_SBB): New unspec.
8307 (cmp<dwi>_doubleword): Removed.
8308 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
8309 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
8310 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
8311 expand with cmp<dwi>_doubleword. For LTU and GEU use
8312 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
8314 * common.opt (gcolumn-info): Enable by default.
8315 * doc/invoke.texi (gcolumn-info): Document new default.
8317 2017-10-23 Richard Biener <rguenther@suse.de>
8319 PR tree-optimization/82672
8320 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
8321 Fold the stmt if we propagated into it.
8323 2017-10-23 Richard Biener <rguenther@suse.de>
8325 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
8326 (bitmap_remove_expr_from_set): ... this. All callers call this
8327 for non-constant values.
8328 (bitmap_set_subtract): Rename to...
8329 (bitmap_set_subtract_expressions): ... this. Adjust and
8331 (bitmap_set_contains_value): Remove superfluous check.
8332 (bitmap_set_replace_value): Inline into single caller ...
8333 (bitmap_value_replace_in_set): ... here and simplify.
8334 (dependent_clean): Merge into ...
8335 (clean): ... this using an overload. Adjust.
8336 (prune_clobbered_mems): Adjust.
8337 (compute_antic_aux): Likewise.
8338 (compute_partial_antic_aux): Likewise.
8340 2017-10-23 Richard Biener <rguenther@suse.de>
8342 PR tree-optimization/82129
8344 2017-08-01 Richard Biener <rguenther@suse.de>
8346 PR tree-optimization/81181
8347 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8348 (compute_antic): ... end of iteration here.
8350 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8352 * target.def (starting_frame_offset): New hook.
8353 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
8354 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
8355 * doc/tm.texi.in: Regenerate.
8356 * hooks.h (hook_hwi_void_0): Declare.
8357 * hooks.c (hook_hwi_void_0): New function.
8358 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
8359 STARTING_FRAME_OFFSET.
8360 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
8361 * reload1.c (reload): Likewise.
8362 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
8363 instead of STARTING_FRAME_OFFSET.
8364 * function.c (try_fit_stack_local): Likewise.
8365 (assign_stack_local_1): Likewise
8366 (instantiate_virtual_regs): Likewise.
8367 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
8368 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
8369 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
8370 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
8371 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
8372 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
8373 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
8374 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
8375 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
8376 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
8377 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
8378 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
8379 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
8380 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
8381 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
8382 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
8383 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
8384 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
8385 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
8386 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
8387 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
8388 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
8389 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
8390 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
8391 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
8392 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
8393 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
8394 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
8395 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
8396 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
8397 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
8398 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
8399 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
8400 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
8401 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
8402 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
8403 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
8404 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
8405 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
8406 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
8407 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
8408 * config/avr/avr.c (avr_starting_frame_offset): Make static and
8409 return a HOST_WIDE_INT.
8410 (avr_builtin_setjmp_frame_value): Use it instead of
8411 STARTING_FRAME_OFFSET.
8412 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8413 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
8414 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
8416 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8417 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
8418 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
8419 (TARGET_CONSTANT_ALIGNMENT): Redefine.
8420 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
8421 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
8422 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8423 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
8424 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
8425 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8426 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
8427 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
8429 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8430 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
8431 * config/mips/mips.c (mips_compute_frame_info): Refer to
8432 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8433 (mips_starting_frame_offset): New function.
8434 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8435 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
8436 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
8437 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
8438 and return a HOST_WIDE_INT.
8439 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8440 (mmix_initial_elimination_offset): Refer to
8441 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
8442 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
8443 * config/pa/pa.c (pa_starting_frame_offset): New function.
8444 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
8445 (pa_expand_prologue): Likewise.
8446 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8447 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
8448 !FRAME_GROWS_DOWNWARD handling to...
8449 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8450 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
8451 !FRAME_GROWS_DOWNWARD handling to...
8452 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8453 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
8454 !FRAME_GROWS_DOWNWARD handling to...
8455 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8456 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
8458 (rs6000_starting_frame_offset): New function.
8459 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
8460 !FRAME_GROWS_DOWNWARD handling to...
8461 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8462 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
8463 !FRAME_GROWS_DOWNWARD handling to...
8464 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8465 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
8466 !FRAME_GROWS_DOWNWARD handling to...
8467 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
8468 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
8469 (rs6000_starting_frame_offset): New function.
8470 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
8471 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
8472 * config/vax/vax.c (vax_starting_frame_offset): New function.
8473 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
8474 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8475 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
8476 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
8477 (TARGET_STARTING_FRAME_OFFSET): Redefine.
8478 * system.h (STARTING_FRAME_OFFSET): Poison.
8480 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8482 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8483 SCALAR_TYPE_MODE instead of TYPE_MODE.
8485 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8486 Alan Hayward <alan.hayward@arm.com>
8487 David Sherwood <david.sherwood@arm.com>
8489 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
8491 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8492 Alan Hayward <alan.hayward@arm.com>
8493 David Sherwood <david.sherwood@arm.com>
8495 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
8497 2017-10-23 Richard Biener <rguenther@suse.de>
8499 PR tree-optimization/82129
8500 * tree-ssa-pre.c (bitmap_set_and): Remove.
8501 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
8502 canonicalizing expressions in the set to those with lowest
8503 ID rather than taking that from the first edge.
8505 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
8507 * combine.c (rtx_equal_for_field_assignment_p): Use
8508 byte_lowpart_offset.
8510 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8511 Alan Hayward <alan.hayward@arm.com>
8512 David Sherwood <david.sherwood@arm.com>
8514 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
8515 to a SUBREG_PROMOTED_VAR.
8517 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8518 Alan Hayward <alan.hayward@arm.com>
8519 David Sherwood <david.sherwood@arm.com>
8521 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
8522 (expand_debug_source_expr): Likewise.
8523 * combine.c (combine_simplify_rtx): Likewise.
8524 * cse.c (fold_rtx): Likewise.
8525 * optabs.c (expand_float): Likewise.
8526 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8527 (simplify_binary_operation_1): Likewise.
8529 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8530 Alan Hayward <alan.hayward@arm.com>
8531 David Sherwood <david.sherwood@arm.com>
8533 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
8534 (record_promoted_value): Likewise.
8535 * expr.c (expand_expr_real_2): Likewise.
8536 * ree.c (update_reg_equal_equiv_notes): Likewise.
8537 (combine_set_extension): Likewise.
8538 * rtlanal.c (low_bitmask_len): Likewise.
8539 * simplify-rtx.c (neg_const_int): Likewise.
8540 (simplify_binary_operation_1): Likewise.
8542 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8543 Alan Hayward <alan.hayward@arm.com>
8544 David Sherwood <david.sherwood@arm.com>
8546 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
8547 * regcprop.c (maybe_mode_change): Likewise.
8548 * reload1.c (alter_reg): Likewise.
8550 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8552 * inchash.h (inchash::hash::add_wide_int): New function.
8553 * lto-streamer-out.c (hash_tree): Use it.
8555 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
8557 * inchash.h (inchash::hash::add_wide_int): Rename to...
8558 (inchash::hash::add_hwi): ...this.
8559 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
8560 (polymorphic_call_target_hasher::hash): Likewise.
8561 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
8562 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
8563 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
8564 * lto-streamer-out.c (hash_tree): Likewise.
8565 * optc-save-gen.awk: Likewise.
8566 * tree.c (add_expr): Likewise.
8568 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8571 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
8572 for ordered inequality comparisons even with TARGET_IEEE_FP.
8574 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
8577 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
8578 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
8579 Expand with cmp<dwi>_doubleword.
8581 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8583 * extend.texi: Add x86 specific to 'nocf_check' attribute.
8584 List CET intrinsics.
8585 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
8586 specific to -fcf-protection option.
8588 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8590 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
8591 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
8592 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
8593 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
8594 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
8595 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
8596 (extra_objs): Add cet.o for Linux/x86 targets.
8597 (tmake_file): Add i386/t-cet for Linux/x86 targets.
8598 * config/i386/cet.c: New file.
8599 * config/i386/cetintrin.h: Likewise.
8600 * config/i386/t-cet: Likewise.
8601 * config/i386/cpuid.h (bit_SHSTK): New.
8602 (bit_IBT): Likewise.
8603 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
8604 pass IBT and SHSTK bits.
8605 * config/i386/i386-builtin-types.def
8606 (VOID_FTYPE_UNSIGNED_PVOID): New.
8607 (VOID_FTYPE_UINT64_PVOID): Likewise.
8608 * config/i386/i386-builtin.def: Add CET intrinsics.
8609 * config/i386/i386-c.c (ix86_target_macros_internal): Add
8610 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
8611 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
8612 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
8614 * config/i386/i386.c (rest_of_insert_endbranch): New.
8615 (pass_data_insert_endbranch): Likewise.
8616 (pass_insert_endbranch): Likewise.
8617 (make_pass_insert_endbranch): Likewise.
8618 (ix86_notrack_prefixed_insn_p): Likewise.
8619 (ix86_target_string): Add -mibt, -mshstk flags.
8620 (ix86_option_override_internal): Add flag_cf_protection
8622 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
8623 (ix86_print_operand): Add 'notrack' prefix output.
8624 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
8625 (ix86_expand_builtin): Expand CET intrinsics.
8626 (x86_output_mi_thunk): Add 'endbranch' instruction.
8627 * config/i386/i386.h (TARGET_IBT): New.
8628 (TARGET_IBT_P): Likewise.
8629 (TARGET_SHSTK): Likewise.
8630 (TARGET_SHSTK_P): Likewise.
8631 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
8632 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
8633 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
8634 (builtin_setjmp_setup): New pattern.
8635 (builtin_longjmp): Likewise.
8636 (rdssp<mode>): Likewise.
8637 (incssp<mode>): Likewise.
8638 (saveprevssp): Likewise.
8639 (rstorssp): Likewise.
8640 (wrss<mode>): Likewise.
8641 (wruss<mode>): Likewise.
8642 (setssbsy): Likewise.
8643 (clrssbsy): Likewise.
8644 (nop_endbr): Likewise.
8645 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
8647 * config/i386/immintrin.h: Include <cetintrin.h>.
8648 * config/i386/linux-common.h
8649 (file_end_indicate_exec_stack_and_cet): New prototype.
8650 (TARGET_ASM_FILE_END): New.
8652 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8654 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
8655 latencies instead of having separate table; make difference between
8656 integer and float costs.
8657 * i386.h (processor_costs): Remove scalar_stmt_cost,
8658 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
8659 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
8661 * x86-tune-costs.h: Remove entries which has been removed in
8662 procesor_costs from all tables; make cond_taken_branch_cost
8663 and cond_not_taken_branch_cost COST_N_INSNS based.
8665 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
8667 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
8669 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8671 * config/i386/i386.md (isa): Remove fma_avx512f.
8672 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
8673 <avx512>_fmadd_<mode>_mask3<round_name>,
8674 <avx512>_fmsub_<mode>_mask<round_name>,
8675 <avx512>_fmsub_<mode>_mask3<round_name>,
8676 <avx512>_fnmadd_<mode>_mask<round_name>,
8677 <avx512>_fnmadd_<mode>_mask3<round_name>,
8678 <avx512>_fnmsub_<mode>_mask<round_name>,
8679 <avx512>_fnmsub_<mode>_mask3<round_name>,
8680 <avx512>_fmaddsub_<mode>_mask<round_name>,
8681 <avx512>_fmaddsub_<mode>_mask3<round_name>,
8682 <avx512>_fmsubadd_<mode>_mask<round_name>,
8683 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
8684 (*vec_widen_umult_even_v16si<mask_name>,
8685 *vec_widen_smult_even_v16si<mask_name>): Likewise.
8686 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
8688 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8690 * extend.texi: Add 'nocf_check' documentation.
8691 * gimple.texi: Add second parameter to
8692 gimple_build_call_from_tree.
8693 * invoke.texi: Add -fcf-protection documentation.
8694 * rtl.texi: Add REG_CALL_NOTRACK documenation.
8696 2017-10-20 Richard Biener <rguenther@suse.de>
8698 PR tree-optimization/82473
8699 * tree-vect-loop.c (vectorizable_reduction): Properly get at
8700 the largest input type.
8702 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
8704 * c-attribs.c (handle_nocf_check_attribute): New function.
8705 (c_common_attribute_table): Add 'nocf_check' handling.
8706 * gimple-parser.c: Add second argument NULL to
8707 gimple_build_call_from_tree.
8708 * attrib.c (comp_type_attributes): Check nocf_check attribute.
8709 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
8711 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
8712 * common.opt: Add fcf-protection flag.
8713 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
8714 * flag-types.h: Add enum cf_protection_level.
8715 * gimple.c (gimple_build_call_from_tree): Add second parameter.
8716 Add 'nocf_check' attribute propagation to gimple call.
8717 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
8718 (gimple_build_call_from_tree): Update prototype.
8719 (gimple_call_nocf_check_p): New function.
8720 (gimple_call_set_nocf_check): Likewise.
8721 * gimplify.c: Add second argument to gimple_build_call_from_tree.
8722 * ipa-icf.c: Add nocf_check attribute in statement hash.
8723 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
8724 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
8725 * toplev.c (process_options): Add flag_cf_protection handling.
8727 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8729 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
8731 2017-10-20 Richard Biener <rguenther@suse.de>
8733 PR tree-optimization/82603
8734 * tree-if-conv.c (predicate_mem_writes): Make sure to only
8735 remove false predicated stores.
8737 2017-10-20 Richard Biener <rguenther@suse.de>
8739 * graphite-isl-ast-to-gimple.c
8740 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
8741 Remove return value and simplify, dump copied stmt after lhs
8743 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
8744 Reduce dump verbosity.
8745 (gsi_insert_earliest): Likewise.
8746 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
8747 * graphite.c (print_global_statistics): Adjust dumping.
8748 (print_graphite_scop_statistics): Likewise.
8749 (print_graphite_statistics): Do not dump loops here.
8750 (graphite_transform_loops): But here.
8752 2017-10-20 Nicolas Roche <roche@adacore.com>
8754 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
8755 * configure: Regenerate.
8757 2017-10-20 Jakub Jelinek <jakub@redhat.com>
8760 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
8761 functions when optimizing replace GIMPLE_RETURN stmts with
8762 calls to __builtin_unreachable ().
8765 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
8766 for -fsanitize=thread link of executables.
8767 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
8768 link of executables.
8771 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
8773 (<shift_insn><mode>3<mask_name>): Change the last of the 3
8774 define_insns for logical vector shifts to use VI248_AVX512BW
8775 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
8776 condition, useless isa and prefix attributes. Change the first
8777 2 of these define_insns to ...
8778 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
8779 define_insn for avx512vl.
8780 (<shift_insn><mode>3): ... and this, new define_insn without
8781 masking for non-avx512vl.
8784 * config/i386/sse.md (*andnot<mode>3,
8785 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
8786 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
8787 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
8788 not applied use empty suffix even for TARGET_AVX512VL.
8789 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
8790 is applied, supply evex,evex or evex,evex,evex instead of just
8793 2017-10-20 Julia Koval <julia.koval@intel.com>
8795 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
8796 (OPTION_MASK_ISA_GFNI_UNSET): New.
8797 (ix86_handle_option): Handle OPT_mgfni.
8798 * config/i386/cpuid.h (bit_GFNI): New.
8799 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
8800 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
8801 * config/i386/i386.c (ix86_target_string): Add -mgfni.
8802 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
8803 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
8804 * config/i386/i386.opt: Add mgfni.
8806 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
8808 * config/msp430/msp430.c (msp430_option_override): Disable
8809 -fdelete-null-pointer-checks.
8810 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
8812 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8814 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
8815 of x87 and SSE instructions.
8817 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
8819 * asan.c (create_cond_insert_point): Do not update edge count.
8820 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
8821 (afdo_propagate_circuit): Likewise.
8822 (afdo_calculate_branch_prob): Likewise.
8823 (afdo_annotate_cfg): Likewise.
8824 * basic-block.h (struct edge_def): Remove count.
8825 (edge_def::count): New accessor.
8826 * bb-reorder.c (rotate_loop): Update.
8827 (find_traces_1_round): Update.
8828 (connect_traces): Update.
8829 (sanitize_hot_paths): Update.
8830 * cfg.c (unchecked_make_edge): Update.
8831 (make_single_succ_edge): Update.
8832 (check_bb_profile): Update.
8833 (dump_edge_info): Update.
8834 (update_bb_profile_for_threading): Update.
8835 (scale_bbs_frequencies_int): Update.
8836 (scale_bbs_frequencies_gcov_type): Update.
8837 (scale_bbs_frequencies_profile_count): Update.
8838 (scale_bbs_frequencies): Update.
8839 * cfganal.c (connect_infinite_loops_to_exit): Update.
8840 * cfgbuild.c (compute_outgoing_frequencies): Update.
8841 (find_many_sub_basic_blocks): Update.
8842 * cfgcleanup.c (try_forward_edges): Update.
8843 (try_crossjump_to_edge): Update
8844 * cfgexpand.c (expand_gimple_cond): Update
8845 (expand_gimple_tailcall): Update
8846 (construct_exit_block): Update
8847 * cfghooks.c (verify_flow_info): Update
8848 (redirect_edge_succ_nodup): Update
8849 (split_edge): Update
8850 (make_forwarder_block): Update
8851 (duplicate_block): Update
8852 (account_profile_record): Update
8853 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
8854 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
8855 * cfgloopmanip.c (scale_loop_profile): Update.
8857 (lv_adjust_loop_entry_edge): Update.
8858 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8859 (force_nonfallthru_and_redirect): Update.
8860 (purge_dead_edges): Update.
8861 (rtl_flow_call_edges_add): Update.
8862 * cgraphunit.c (init_lowered_empty_function): Update.
8863 (cgraph_node::expand_thunk): Update.
8864 * gimple-pretty-print.c (dump_probability): Update.
8865 (dump_edge_probability): Update.
8866 * gimple-ssa-isolate-paths.c (isolate_path): Update.
8867 * haifa-sched.c (sched_create_recovery_edges): Update.
8868 * hsa-gen.c (convert_switch_statements): Update.
8869 * ifcvt.c (dead_or_predicable): Update.
8870 * ipa-inline-transform.c (inline_transform): Update.
8871 * ipa-split.c (split_function): Update.
8872 * ipa-utils.c (ipa_merge_profiles): Update.
8873 * loop-doloop.c (add_test): Update.
8874 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
8875 * lto-streamer-in.c (input_cfg): Update.
8876 (input_function): Update.
8877 * lto-streamer-out.c (output_cfg): Update.
8878 * modulo-sched.c (sms_schedule): Update.
8879 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
8880 * predict.c (maybe_hot_edge_p): Update.
8881 (unlikely_executed_edge_p): Update.
8882 (probably_never_executed_edge_p): Update.
8883 (dump_prediction): Update.
8884 (drop_profile): Update.
8885 (propagate_unlikely_bbs_forward): Update.
8886 (determine_unlikely_bbs): Update.
8887 (force_edge_cold): Update.
8888 * profile.c (compute_branch_probabilities): Update.
8889 * reg-stack.c (better_edge): Update.
8890 * shrink-wrap.c (handle_simple_exit): Update.
8891 * tracer.c (better_p): Update.
8892 * trans-mem.c (expand_transaction): Update.
8893 (split_bb_make_tm_edge): Update.
8894 * tree-call-cdce.c: Update.
8895 * tree-cfg.c (gimple_find_sub_bbs): Update.
8896 (gimple_split_edge): Update.
8897 (gimple_duplicate_sese_region): Update.
8898 (gimple_duplicate_sese_tail): Update.
8899 (gimple_flow_call_edges_add): Update.
8900 (insert_cond_bb): Update.
8901 (execute_fixup_cfg): Update.
8902 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
8903 * tree-complex.c (expand_complex_div_wide): Update.
8904 * tree-eh.c (lower_resx): Update.
8905 (unsplit_eh): Update.
8906 (cleanup_empty_eh_move_lp): Update.
8907 * tree-inline.c (copy_edges_for_bb): Update.
8908 (freqs_to_counts): Update.
8909 (copy_cfg_body): Update.
8910 * tree-ssa-dce.c (remove_dead_stmt): Update.
8911 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8912 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8913 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8914 (unloop_loops): Update.
8915 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8916 * tree-ssa-loop-split.c (connect_loops): Update.
8917 (split_loop): Update.
8918 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
8919 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8920 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8921 * tree-ssa-reassoc.c (branch_fixup): Update.
8922 * tree-ssa-tail-merge.c (replace_block_by): Update.
8923 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8924 (compute_path_counts): Update.
8925 (update_profile): Update.
8926 (recompute_probabilities): Update.
8927 (update_joiner_offpath_counts): Update.
8928 (estimated_freqs_path): Update.
8929 (freqs_to_counts_path): Update.
8930 (clear_counts_path): Update.
8931 (ssa_fix_duplicate_block_edges): Update.
8932 (duplicate_thread_path): Update.
8933 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8934 (case_bit_test_cmp): Update.
8935 (collect_switch_conv_info): Update.
8936 (gen_inbound_check): Update.
8937 (do_jump_if_equal): Update.
8938 (emit_cmp_and_jump_insns): Update.
8939 * tree-tailcall.c (decrease_profile): Update.
8940 (eliminate_tail_call): Update.
8941 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8942 (vect_do_peeling): Update.
8943 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8944 * ubsan.c (ubsan_expand_null_ifn): Update.
8945 (ubsan_expand_ptr_ifn): Update.
8946 * value-prof.c (gimple_divmod_fixed_value): Update.
8947 (gimple_mod_pow2): Update.
8948 (gimple_mod_subtract): Update.
8949 (gimple_ic): Update.
8950 (gimple_stringop_fixed_value): Update.
8952 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8955 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
8957 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
8959 PR rtl-optimization/82395
8960 * ira-color.c (allocno_priority_compare_func): Fix comparison step
8961 based on non_spilled_static_chain_regno_p.
8963 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
8965 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
8966 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
8967 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
8969 2017-10-19 Martin Sebor <msebor@redhat.com>
8971 PR tree-optimization/82596
8972 * tree.c (array_at_struct_end_p): Handle STRING_CST.
8974 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
8976 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
8977 (get_mem_refs_of_builtin_call): Likewise.
8978 * builtins.c (expand_builtin_apply): Adjust call to
8979 allocate_dynamic_stack_space.
8980 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
8981 the third argument to allocate_dynamic_stack_space, otherwise -1.
8982 (expand_builtin): Deal with all alloca variants.
8983 (is_inexpensive_builtin): Likewise.
8984 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
8985 * calls.c (special_function_p): Deal with all alloca variants.
8986 (initialize_argument_information): Adjust call to
8987 allocate_dynamic_stack_space.
8988 (expand_call): Likewise.
8989 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
8990 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
8991 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
8992 use it for the stack usage computation.
8993 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
8994 * function.c (gimplify_parameters): Call build_alloca_call_expr.
8995 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
8996 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
8997 (in_loop_p): Remove first argument and useless check.
8998 (pass_walloca::execute): Remove useless test and adjust call to above.
8999 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
9000 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
9001 (gimplify_call_expr): Deal with all alloca variants.
9002 * hsa-gen.c (gen_hsa_alloca): Likewise.
9003 (gen_hsa_insns_for_call): Likewise.
9004 * ipa-pure-const.c (special_builtin_state): Likewise.
9005 * tree-chkp.c (chkp_build_returned_bound): Likewise.
9006 * tree-object-size.c (alloc_object_size): Likewise.
9007 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
9008 (call_may_clobber_ref_p_1): Likewise.
9009 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
9010 (ccp_fold_stmt): Likewise.
9011 (optimize_stack_restore): Likewise.
9012 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9013 (mark_all_reaching_defs_necessary_1): Likewise.
9014 (propagate_necessity): Likewise.
9015 (eliminate_unnecessary_stmts): Likewise.
9016 * tree.c (build_common_builtin_nodes): Build
9017 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
9018 (build_alloca_call_expr): New function.
9019 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
9020 (CASE_BUILT_IN_ALLOCA): Likewise.
9021 (build_alloca_call_expr): Declare.
9022 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
9024 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
9027 * dwarf2out.c (new_die_raw): New static inline function.
9028 (new_die): Use it to create the DIE.
9029 (add_AT_external_die_ref): Likewise.
9030 (clone_die): Likewise.
9031 (clone_as_declaration): Likewise.
9032 (dwarf2out_vms_debug_main_pointer): Likewise.
9033 (base_type_die): Likewise. Remove early return for corner cases.
9034 Do not call add_pubtype on the DIE here.
9035 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
9036 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
9037 typedefs for base types with DW_AT_endianity. Make sure a DIE with
9038 native order exists for base types, attach the DIE manually and call
9039 add_pubtype on it. Do not equate a reverse order DIE to the type.
9041 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
9043 * config/arm/arm.c (align_ok_ldrd_strd): New function.
9044 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
9046 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
9048 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9050 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
9051 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
9052 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
9053 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
9054 * opts.c (sanitizer_opts): Add builtin.
9055 * ubsan.c (instrument_builtin): New function.
9056 (pass_ubsan::execute): Call it.
9057 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
9058 * doc/invoke.texi: Document -fsanitize=builtin.
9060 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
9061 builtins, store max (log2 (align), 0) into uchar field instead of
9062 align into uptr field.
9063 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
9064 store uchar 0 field instead of uptr 0 field.
9065 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
9066 instead of passing one address of struct with 2 locations pass
9067 two addresses of structs with 1 location each.
9068 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
9069 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
9070 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
9071 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
9072 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
9073 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
9074 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
9075 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
9077 2017-10-19 Martin Liska <mliska@suse.cz>
9080 * file-find.c (remove_prefix): Remove.
9081 * file-find.h (remove_prefix): Likewise.
9082 * gcc-ar.c: Remove smartness of lookup.
9084 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
9086 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
9087 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
9088 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
9090 2017-10-19 Jakub Jelinek <jakub@redhat.com>
9093 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
9094 (setcc + and to xor + setcc): New peephole2.
9096 2017-10-19 Tom de Vries <tom@codesourcery.com>
9098 * doc/sourcebuild.texi (Test Directives, Variants of
9099 dg-require-support): Add dg-require-stack-size.
9101 2017-10-19 Martin Liska <mliska@suse.cz>
9104 * gimplify.c (gimplify_decl_expr): Do not instrument variables
9105 that have a large alignment.
9106 (gimplify_target_expr): Likewise.
9108 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
9110 PR rtl-optimization/82602
9111 * ira.c (rtx_moveable_p): Return false for volatile asm.
9113 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
9116 * config/i386/i386-modes.def (CCGZ): New CC mode.
9117 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
9118 * config/i386/predicates.md (ix86_comparison_operator):
9120 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
9121 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
9122 with double-word subtraction.
9123 (put_condition_code): Handle CCGZmode.
9125 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
9127 * wide-int.cc (debug (const wide_int &)): New.
9128 (debug (const wide_int *)): New.
9129 (debug (const widest_int &)): New.
9130 (debug (const widest_int *)): New.
9132 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
9135 * lra-constraints.c (curr_insn_transform): Use non-input operand
9136 instead of output one for matched reload.
9138 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9140 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
9141 (tree-ssa-loop-ivopts.h): New header file.
9142 (struct builtin_info): New fields.
9143 (classify_builtin_1): Compute and record base and offset parts for
9144 memset builtin partition by calling strip_offset.
9145 (offset_cmp, fuse_memset_builtins): New functions.
9146 (finalize_partitions): Fuse adjacent memset partitions by calling
9148 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
9149 Expose the interface.
9150 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
9152 2017-10-18 Bin Cheng <bin.cheng@arm.com>
9154 PR tree-optimization/82574
9155 * tree-loop-distribution.c (find_single_drs): New parameter. Check
9156 that data reference must be executed exactly once per iteration
9157 against the outermost loop in nest.
9158 (classify_partition): Update call to above function.
9160 2017-10-18 Richard Biener <rguenther@suse.de>
9162 PR tree-optimization/82591
9163 * graphite.c (graphite_transform_loops): Move code gen message
9165 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9166 Here. Handle scop_to_isl_ast failing.
9167 (scop_to_isl_ast): Limit the number of ISL operations.
9169 2017-10-18 Richard Biener <rguenther@suse.de>
9171 * graphite-isl-ast-to-gimple.c
9172 (translate_isl_ast_to_gimple::set_rename): Simplify.
9173 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
9174 (graphite_copy_stmts_from_block): ... here.
9175 (copy_bb_and_scalar_dependences): Simplify.
9176 (add_parameters_to_ivs_params): Canonicalize.
9177 (generate_entry_out_of_ssa_copies): Simplify.
9178 * graphite-sese-to-poly.c (extract_affine_name): Simplify
9179 by passing in ISL dimension.
9180 (parameter_index_in_region_1): Rename to ...
9181 (parameter_index_in_region): ... this.
9182 (extract_affine): Adjust assert, pass down parameter index.
9183 (add_param_constraints): Use range-info when available.
9184 (build_scop_context): Adjust.
9185 * sese.c (new_sese_info): Adjust.
9186 (free_sese_info): Likewise.
9187 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
9188 Remove unused typedefs.
9189 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
9191 2017-10-18 Martin Liska <mliska@suse.cz>
9193 * combine.c (simplify_compare_const): Add gcc_fallthrough.
9195 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9197 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
9198 (s390_sched_init): Do not reset s390_sched_state if we entered the
9199 current basic block via a fallthru edge and all others are unlikely.
9201 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
9203 * config/s390/s390.c (NUM_SIDES): New variable.
9204 (LONGRUNNING_THRESHOLD): New variable.
9205 (LATENCY_FACTOR): New variable.
9206 (s390_sched_score): Decrease score for long-running instructions on
9208 (s390_sched_variable_issue): Perform bookkeeping for long-running
9211 2017-10-18 Richard Biener <rguenther@suse.de>
9213 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
9214 Simplify with removal of the parameter rename map.
9215 (set_rename): Likewise.
9216 (should_copy_to_new_region): Likewise.
9217 (graphite_copy_stmts_from_block): Likewise.
9218 (copy_bb_and_scalar_dependences): Remove initialization of
9219 unused copied_bb_map.
9221 (copy_internal_parameters): Likewise.
9222 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
9223 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
9224 Use INTEGRAL_TYPE_P.
9225 (parameter_index_in_region_1): Rename to ...
9226 (assign_parameter_index_in_region): ... this. Assert we have
9227 a parameter we handle.
9228 (scan_tree_for_params): Adjust.
9229 * sese.h (parameter_rename_map_t): Remove.
9230 (struct sese_info_t): Remove unused parameter_rename_map and
9231 copied_bb_map members.
9232 * sese.c (new_sese_info): Adjust.
9233 (free_sese_info): Likewise.
9235 2017-10-18 Martin Liska <mliska@suse.cz>
9238 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
9239 on an abnormal edge.
9241 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
9243 * doc/invoke.texi (ffunction-sections and fdata-sections):
9246 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9248 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
9249 the use statement can throw internally.
9251 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
9253 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
9254 any RTX present on the RHS of a SET.
9255 * compare-elim.c (try_eliminate_compare): Restore comment.
9257 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9259 * langhooks.h (struct lang_hooks): Document that tree_size langhook
9260 may be also called on tcc_type nodes.
9261 * langhooks.c (lhd_tree_size): Likewise.
9263 2017-10-17 David Malcolm <dmalcolm@redhat.com>
9265 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
9266 format_warning_at_substring.
9267 (maybe_warn): Convert source_range * param to a location_t. Pass
9268 UNKNOWN_LOCATION rather than NULL to fmtwarn.
9269 (format_directive): Remove code to extract source_ranges and
9270 source_range * in favor of just a location_t.
9271 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
9273 * substring-locations.c (format_warning_va): Convert
9274 source_range * param to a location_t.
9275 (format_warning_at_substring): Likewise.
9276 * substring-locations.h (format_warning_va): Likewise.
9277 (format_warning_at_substring): Likewise.
9279 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
9281 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
9283 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
9284 and scatter/gather ops.
9286 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
9287 vec_gather_load and vec_scatter_store.
9288 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
9289 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
9290 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
9291 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
9292 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
9293 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
9295 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
9297 * reg-stack.c (compare_for_stack_reg): Add bool argument.
9298 Detect FTST instruction and handle its register pops. Only pop
9299 second operand if can_pop_second_op is true.
9300 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
9301 set can_pop_second_op to false in the compare_for_stack_reg call.
9303 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
9304 output_fp_compare for stack register operands.
9305 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
9306 instructions here. Do not emit stack register pops here. Assert
9307 that FCOMPP pops next to top stack register. Rewrite function.
9309 2017-10-17 Nathan Sidwell <nathan@acm.org>
9312 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
9313 use DECL_ASSEMBLER_NAME_RAW.
9316 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
9319 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
9320 Wilco Dijkstra <wilco.dijkstra@arm.com>
9322 * builtins.c (expand_builtin_update_setjmp_buf): Add a
9323 converstion to Pmode from the buf_addr.
9325 2017-10-17 Richard Biener <rguenther@suse.de>
9327 * graphite-dependences.c (scop_get_reads_and_writes): Change
9328 output parameters to references.
9330 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
9332 PR 71026/tree-optimization
9333 * fold-const.c (distribute_real_division): Removed.
9334 (fold_binary_loc): Remove calls to distribute_real_divison.
9336 2017-10-17 Richard Biener <rguenther@suse.de>
9338 * graphite-scop-detection.c
9339 (scop_detection::stmt_has_simple_data_refs_p): Always use
9340 the full nest as region.
9341 (try_generate_gimple_bb): Likewise.
9342 * sese.c (scalar_evolution_in_region): Simplify now that
9343 SCEV can handle instantiation in regions.
9344 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
9345 in the non-loop part of a function if requested.
9347 2017-10-17 Richard Biener <rguenther@suse.de>
9349 PR tree-optimization/82563
9350 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
9352 (graphite_regenerate_ast_isl): Call it.
9353 * graphite-scop-detection.c (build_scops): Remove entry edge split.
9355 2017-10-17 Jakub Jelinek <jakub@redhat.com>
9357 PR tree-optimization/82549
9358 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
9359 Formatting fixes. Instead of calling make_bit_field_ref with negative
9362 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
9364 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
9365 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
9366 _mm_maskz_reduce_ss): New.
9367 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
9368 __builtin_ia32_reducess_mask): Ditto..
9369 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
9370 * config/i386/sse.md (reduces<mode>): Renamed to ...
9371 (reduces<mode><mask_scalar_name>): ... this.
9372 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
9374 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
9375 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
9377 2017-10-16 David Malcolm <dmalcolm@redhat.com>
9379 * Makefile.in (OBJS): Add unique-ptr-tests.o.
9380 * selftest-run-tests.c (selftest::run_tests): Call
9381 selftest::unique_ptr_tests_cc_tests.
9382 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
9383 * unique-ptr-tests.cc: New file.
9385 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
9388 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9390 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9391 (make_hard_regno_born, make_hard_regno_dead): Update
9392 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
9394 2017-10-16 Jeff Law <law@redhat.com>
9396 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9398 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
9400 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
9402 2017-10-16 Olivier Hainque <hainque@adacore.com>
9404 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
9405 with_cpu if we were configured for an e500v2 target cpu name.
9407 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
9409 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
9410 * doc/invoke.texi: Document +nodsp as a valid extension for
9413 2017-10-16 Martin Liska <mliska@suse.cz>
9415 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
9416 (test_set_range): Likewise.
9417 (test_range_functions): Rename to ...
9418 (test_bit_in_range): ... this.
9419 (sbitmap_c_tests): Add new test.
9421 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9423 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
9425 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
9426 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
9428 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9430 * config/aarch64/aarch64-builtins.c
9431 (aarch64_types_quadopu_lane_qualifiers): New.
9432 (TYPES_QUADOPU_LANE): New.
9433 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
9434 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
9435 (aarch64_<sur>dot_laneq<vsi2qi>): New.
9436 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
9437 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
9438 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
9439 (Vdottype, DOTPROD): New.
9440 (sur): Add SDOT and UDOT.
9442 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9444 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
9445 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
9446 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
9448 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
9449 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
9450 Enable TARGET_DOTPROD.
9451 (cortex-a75.cortex-a55): Likewise.
9452 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
9454 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9456 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
9457 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
9459 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
9461 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
9462 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
9463 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
9464 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
9465 * config/arm/types.md (neon_dot, neon_dot_q): New.
9466 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
9468 2017-10-16 Tamar Christina <tamar.christina@arm.com>
9470 * config/arm/arm.h (TARGET_DOTPROD): New.
9471 * config/arm/arm.c (arm_arch_dotprod): New.
9472 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
9473 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
9474 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
9475 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
9476 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
9477 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
9478 * doc/invoke.texi (armv8.2-a): Document dotprod
9480 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9482 * i386.c (ix86_vec_cost): New function.
9483 (ix86_rtx_costs): Handle vector operations better.
9484 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
9485 * x86-tune-costs.h: Add new costs to all tables.
9487 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
9489 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
9491 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
9492 divsd, sqrtss and sqrtsd
9493 * x86-tune-costs.h: Add new entries to all costs.
9494 (znver1_cost): Fix to match real instruction latencies.
9496 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9497 Michael Collison <michael.collison@arm.com>
9499 * compare-elim.c: Include emit-rtl.h.
9500 (can_merge_compare_into_arith): New function.
9501 (try_validate_parallel): Likewise.
9502 (try_merge_compare): Likewise.
9503 (try_eliminate_compare): Call the above when no previous clobber
9505 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
9508 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9512 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
9513 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
9517 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
9518 to be any operand_equal_p operands. For & (B - 1) require
9519 B to be power of 2. Recognize
9520 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
9522 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9525 * optabs.c (expand_memory_blockage): Fix call of
9526 targetm.have_memory_blockage.
9528 2017-10-14 Jakub Jelinek <jakub@redhat.com>
9531 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
9532 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
9533 objects to extra_objs instead of overwriting it.
9535 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9537 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
9538 Use any_fp_register_operand as operand[3] predicate. Simplify
9539 equality test for operands[2] and operands[4] memory location.
9540 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
9541 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
9542 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
9543 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
9544 any_fp_register_operand as operand[1] predicate. Simplify
9545 equality test for operands[0] and operands[3] memory location.
9546 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
9547 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
9548 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
9550 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
9552 * target-insns.def: Add memory_blockage.
9553 * optabs.c (expand_memory_blockage): New function.
9554 (expand_asm_memory_barrier): Rename ...
9555 (expand_asm_memory_blockage): ... to this.
9556 (expand_mem_thread_fence): Call expand_memory_blockage
9557 instead of expand_asm_memory_barrier.
9558 (expand_mem_singnal_fence): Ditto.
9559 (expand_atomic_load): Ditto.
9560 (expand_atomic_store): Ditto.
9561 * doc/md.texi (Standard Pattern Names For Generation):
9562 Document memory_blockage instruction pattern.
9564 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
9566 * config/rl78/rl78.c (rl78_emit_libcall): New function.
9567 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
9568 * config/rl78/rl78.md: New define_expand "adddi3".
9570 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9572 * cfghooks.c (verify_flow_info): Disable check that all probabilities
9575 2017-10-13 Jeff Law <law@redhat.com>
9577 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
9579 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9582 * internal-fn.c (expand_mul_overflow): If both operands have
9583 the same highpart of -1 or 0 and the topmost bit of lowpart
9584 is different, overflow is if res <= 0 rather than res < 0.
9586 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
9588 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
9589 TARGET_P9_VECTOR code for unaligned_load case.
9591 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
9593 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
9595 2017-10-13 Nathan Sidwell <nathan@acm.org>
9597 * tree-core.h (tree_contains_struct): Make bool.
9598 * tree.c (tree_contains_struct): Likewise.
9599 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
9600 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
9601 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
9602 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
9603 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
9605 2017-10-13 Richard Biener <rguenther@suse.de>
9607 * graphite-isl-ast-to-gimple.c
9608 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
9609 parameters and dominance check.
9610 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
9611 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
9612 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
9613 Do not update SSA form here or do intermediate IL verification.
9614 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
9615 (graphite_initialize): Remove check on the number of loops in
9616 the function and inline into graphite_transform_loops.
9617 (graphite_finalize): Inline into graphite_transform_loops.
9618 (graphite_transform_loops): Perform SSA update and IL verification
9620 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
9622 2017-10-13 Richard Biener <rguenther@suse.de>
9624 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
9625 graphite_expression_type_precision): Avoid global constructor
9627 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
9628 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
9629 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
9630 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
9631 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
9632 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
9634 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
9637 * config/i386/i386.h (ix86_red_zone_size): New.
9638 * config/i386/i386.md (push peephole2s): Replace
9639 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
9641 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9642 Alan Hayward <alan.hayward@arm.com>
9643 David Sherwood <david.sherwood@arm.com>
9645 * combine.c (can_change_dest_mode): Reject changes in
9646 REGMODE_NATURAL_SIZE.
9648 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9649 Alan Hayward <alan.hayward@arm.com>
9650 David Sherwood <david.sherwood@arm.com>
9652 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
9653 (expand_debug_source_expr): Likewise.
9654 * combine.c (combine_simplify_rtx): Likewise.
9655 * cse.c (fold_rtx): Likewise.
9656 * fwprop.c (canonicalize_address): Likewise.
9657 * targhooks.c (default_shift_truncation_mask): Likewise.
9659 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9660 Alan Hayward <alan.hayward@arm.com>
9661 David Sherwood <david.sherwood@arm.com>
9663 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
9664 (widened_mode): Likewise.
9665 (expand_unop): Likewise.
9666 * ree.c (transform_ifelse): Likewise.
9667 (merge_def_and_ext): Likewise.
9668 (combine_reaching_defs): Likewise.
9669 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9671 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9672 Alan Hayward <alan.hayward@arm.com>
9673 David Sherwood <david.sherwood@arm.com>
9675 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
9676 * combine.c (gen_lowpart_for_combine): Likewise.
9677 * dwarf2out.c (rtl_for_decl_location): Likewise.
9678 * final.c (alter_subreg): Likewise.
9679 * rtlhooks.c (gen_lowpart_general): Likewise.
9680 (gen_lowpart_if_possible): Likewise.
9682 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
9683 Alan Hayward <alan.hayward@arm.com>
9684 David Sherwood <david.sherwood@arm.com>
9686 * calls.c (expand_call): Use subreg_lowpart_offset.
9687 * cse.c (cse_insn): Likewise.
9688 * regcprop.c (copy_value): Likewise.
9689 (copyprop_hardreg_forward_1): Likewise.
9691 2017-10-13 Jakub Jelinek <jakub@redhat.com>
9694 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
9695 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
9696 =Q constraints to +Q and into insn condition add check
9697 that operands[0] and operands[1] are equal.
9698 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
9699 =Q constraints to +Q and into insn condition add check
9700 that operands[0] is equal to either operands[1] or operands[2].
9703 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
9704 instead of handling MINUS_EXPR twice (once for each argument),
9705 canonicalize operand order and handle just once, use rtype where
9706 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
9709 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
9710 any values of __C while still being pattern recognizable as a simple
9713 2017-10-13 Richard Biener <rguenther@suse.de>
9715 PR tree-optimization/82451
9717 2017-10-02 Richard Biener <rguenther@suse.de>
9719 PR tree-optimization/82355
9720 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
9721 a mapping for the enclosing loop but avoid generating one for
9723 (copy_bb_and_scalar_dependences): Remove premature codegen
9724 error on PHIs in blocks duplicated into multiple places.
9725 * graphite-scop-detection.c
9726 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
9727 in the region use it as loop and nest to analyze the DR in.
9728 (try_generate_gimple_bb): Likewise.
9729 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
9730 (add_loop_constraints): For blocks in a loop not in the region
9731 create a dimension with a single iteration.
9732 * sese.h (gbb_loop_at_index): Remove assert.
9734 * cfgloop.c (loop_preheader_edge): For the loop tree root
9735 return the single successor of the entry block.
9736 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
9737 Reset the SCEV hashtable and niters.
9738 * graphite-scop-detection.c
9739 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
9740 assert that we only have POLYNOMIAL_CHREC that vary in loops
9741 contained in the region.
9742 (scop_detection::graphite_can_represent_expr): Adjust.
9743 (scop_detection::stmt_has_simple_data_refs_p): For loops
9744 not in the region set loop to NULL. The nest is now the
9745 entry edge to the region.
9746 (try_generate_gimple_bb): Likewise.
9747 * sese.c (scalar_evolution_in_region): Adjust for
9748 instantiate_scev change.
9749 * tree-data-ref.h (graphite_find_data_references_in_stmt):
9750 Make nest parameter the edge into the region.
9751 (create_data_ref): Likewise.
9752 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
9753 entry edge into a region and adjust instantiate_scev calls.
9754 (create_data_ref): Likewise.
9755 (graphite_find_data_references_in_stmt): Likewise.
9756 (find_data_references_in_stmt): Pass the loop preheader edge
9757 from the nest argument.
9758 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
9759 parameter the edge into the region.
9760 (instantiate_parameters): Use the loop preheader edge as entry.
9761 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
9763 (get_instantiated_value_entry): Make instantiate_below parameter
9764 the edge into the region.
9765 (instantiate_scev_name): Likewise. Adjust dominance checks,
9766 when we cannot use loop-based instantiation instantiate by
9767 walking use-def chains.
9768 (instantiate_scev_poly): Adjust.
9769 (instantiate_scev_binary): Likewise.
9770 (instantiate_scev_convert): Likewise.
9771 (instantiate_scev_not): Likewise.
9772 (instantiate_array_ref): Remove.
9773 (instantiate_scev_3): Likewise.
9774 (instantiate_scev_2): Likewise.
9775 (instantiate_scev_1): Likewise.
9776 (instantiate_scev_r): Do not blindly handle N-operand trees.
9777 Do not instantiate array-refs. Handle all constants and invariants.
9778 (instantiate_scev): Make instantiate_below parameter
9779 the edge into the region.
9780 (resolve_mixers): Use the loop preheader edge for the region
9781 parameter to instantiate_scev_r.
9782 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
9784 2017-10-13 Richard Biener <rguenther@suse.de>
9786 PR tree-optimization/82525
9787 * graphite-isl-ast-to-gimple.c
9788 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
9790 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
9791 Fail code generation when we cannot represent the isl integer.
9792 (binary_op_to_tree): Elide modulo operations that are no-ops
9793 in the type we code generate. Remove now superfluous code
9796 2017-10-13 Richard Biener <rguenther@suse.de>
9798 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
9799 (scop_detection::harmful_loop_in_region): Remove premature
9800 IV type restriction.
9801 (scop_detection::graphite_can_represent_scev): We can handle
9802 pointer IVs just fine.
9804 2017-10-13 Alan Modra <amodra@gmail.com>
9806 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
9807 "Clobbers and Scratch Registers". Add paragraph on
9808 alternative to clobbers for scratch registers and OpenBLAS
9811 2017-10-13 Alan Modra <amodra@gmail.com>
9813 * doc/extend.texi (Clobbers): Correct vax example. Delete old
9814 example of a memory input for a string of known length. Move
9815 commentary out of table. Add a number of new examples
9816 covering array memory inputs.
9818 2017-10-12 Martin Liska <mliska@suse.cz>
9820 PR tree-optimization/82493
9821 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
9822 (test_range_functions): New function.
9823 (sbitmap_c_tests): Likewise.
9824 * selftest-run-tests.c (selftest::run_tests): Run new tests.
9825 * selftest.h (sbitmap_c_tests): New function.
9827 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
9829 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
9831 * config/rs6000/amo.h: Fix spacing issue.
9833 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9836 * config/i386/i386.md (*ashl<mode>3_mask_1,
9837 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
9838 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
9841 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9843 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
9844 (profile_probability): Set max_probability
9845 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
9847 * profile-count.c (profile_probability::differs_from_p): Do not
9848 rely on max_probaiblity == 10000
9850 2017-10-12 Jeff Law <law@redhat.com>
9852 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
9855 2017-10-12 Martin Sebor <msebor@redhat.com>
9859 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
9860 (handle_alias_pairs): Call it.
9861 * common.opt (-Wattribute-alias): New option.
9862 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
9863 * doc/invoke.texi (-Wattribute-alias): Document.
9865 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
9868 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
9870 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
9872 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
9873 (make_hard_regno_born, make_hard_regno_dead): Update
9874 bb_killed_pseudos and bb_gen_pseudos.
9876 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
9878 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
9880 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
9882 * config/alpha/alpha.c (alpha_split_conditional_move):
9883 Use std::swap instead of manually swapping.
9884 (alpha_stdarg_optimize_hook): Ditto.
9885 (alpha_canonicalize_comparison): Ditto.
9887 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9889 * tree-loop-distribution.c (struct builtin_info): New struct.
9890 (struct partition): Refactor fields into struct builtin_info.
9891 (partition_free): Free struct builtin_info.
9892 (build_size_arg_loc, build_addr_arg_loc): Delete.
9893 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
9894 information from struct builtin_info.
9895 (find_single_drs): New function refactored from classify_partition.
9896 Also moved builtin validity checks to this function.
9897 (compute_access_range, alloc_builtin): New functions.
9898 (classify_builtin_st, classify_builtin_ldst): New functions.
9899 (classify_partition): Refactor code into functions find_single_drs,
9900 classify_builtin_st and classify_builtin_ldst.
9901 (distribute_loop): Don't do runtime alias check when distributing
9903 (find_seed_stmts_for_distribution): New function.
9904 (pass_loop_distribution::execute): Refactor code finding seed
9905 stmts into above function. Support distribution for the innermost
9906 two-level loop nest. Adjust dump information.
9908 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9910 * tree-loop-distribution.c: Adjust the general comment.
9911 (NUM_PARTITION_THRESHOLD): New macro.
9912 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
9913 (classify_partition): Skip builtin pattern of loop nest's inner loop.
9914 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
9915 in call to build_partition_graph.
9916 (finalize_partitions): New parameter. Make loop distribution more
9917 conservative by fusing more partitions.
9918 (distribute_loop): Don't do runtime alias check in case of loop nest
9920 (find_seed_stmts_for_distribution): New function.
9921 (prepare_perfect_loop_nest): New function.
9922 (pass_loop_distribution::execute): Refactor code finding seed stmts
9923 and loop nest into above functions. Support loop nest distribution.
9924 Adjust dump information accordingly.
9926 2017-10-12 Bin Cheng <bin.cheng@arm.com>
9928 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
9929 and set PTYPE_SEQUENTIAL for merged partition.
9931 2017-10-12 Richard Biener <rguenther@suse.de>
9933 PR tree-optimization/69728
9935 2017-09-19 Richard Biener <rguenther@suse.de>
9937 PR tree-optimization/69728
9938 * graphite-sese-to-poly.c (schedule_error): New global.
9939 (add_loop_schedule): Handle empty domain by failing the
9941 (build_original_schedule): Handle schedule_error.
9943 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
9944 domain by returning an unchanged schedule.
9946 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9948 * genrecog.c (validate_pattern): For VEC_SELECT verify that
9949 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
9951 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
9953 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
9956 2017-10-12 Jakub Jelinek <jakub@redhat.com>
9959 * expr.c (store_field): Don't optimize away bitsize == 0 store
9960 from CALL_EXPR with addressable return type.
9962 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9964 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
9965 * config/rs6000/rs6000.md (sel): Delete mode attribute.
9966 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
9967 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
9968 TARGET_ISEL instead of TARGET_ISEL<sel>.
9970 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
9972 * config/rs6000/rs6000.c
9973 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
9975 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
9977 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
9979 (reg_or_cint_operand): Fix comment.
9980 (reg_or_zero_operand): New predicate.
9981 * config/rs6000/rs6000-protos.h (output_isel): Delete.
9982 * config/rs6000/rs6000.c (output_isel): Delete.
9983 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
9984 instead of reg_or_cint_operand. Output instruction directly (not via
9986 (isel_unsigned_<mode>): Ditto.
9987 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
9988 gpc_reg_operand. Add an instruction alternative for this. Output
9989 instruction directly.
9990 (*isel_reversed_unsigned_<mode>): Ditto.
9992 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
9994 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
9995 (TARGET_CANONICALIZE_COMPARISON): Define.
9997 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
10000 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
10001 Check whether the dest is REG before adding REG_EQUIV note.
10003 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
10006 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
10008 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
10009 (make_hard_regno_born, make_hard_regno_dead): Update
10010 bb_killed_pseudos and bb_gen_pseudos.
10012 2017-10-11 Nathan Sidwell <nathan@acm.org>
10014 * incpath.h (enum incpath_kind): Name enum, prefix values.
10015 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
10016 * incpath.c (heads, tails): Use INC_MAX.
10017 (add_env_var_paths, add_standard_paths): Use incpath_kind.
10018 (merge_include_chains, split_quote_chain,
10019 register_include_chains): Update incpath_kind names.
10020 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
10021 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
10023 (add_framework_path, darwin_register_objc_includes): Likewise.
10024 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
10026 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10028 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
10029 Do not use float_operator operator predicate.
10030 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
10031 * config/i386/predicates.md (float_operator): Remove predicate.
10033 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
10035 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
10036 (*jccxf_i387): Ditto.
10037 (*jcc<mode>_i387): Ditto.
10038 (*jccu<mode>_i387): Ditto.
10039 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
10040 (*jcc_*_i387 splitters): Remove.
10041 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
10042 * config/i386/i386.c (ix86_split_fp_branch): Remove.
10043 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
10046 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10048 * profile-count.h (slow_safe_scale_64bit): New function.
10049 (safe_scale_64bit): New inline.
10050 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
10051 * profile-count.c: Include wide-int.h
10052 (slow_safe_scale_64bit): New.
10054 2017-10-11 Nathan Sidwell <nathan@acm.org>
10056 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
10057 HAS_DECL_ASSEMBLER_NAME_P.
10058 * gimple-expr.c (gimple_decl_printable_name: Check
10059 HAS_DECL_ASSEMBLER_NAME_P too.
10060 * ipa-utils.h (type_in_anonymous_namespace_p): Check
10061 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
10062 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
10063 * passes.c (rest_of_decl_compilation): Check
10064 HAS_DECL_ASSEMBLER_NAME_P too.
10065 * recog.c (verify_changes): Likewise.
10066 * tree-pretty-print.c (dump_decl_name): Likewise.
10067 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
10069 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
10070 (SET_DECL_ASSEMBLER_NAME): Use it.
10071 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
10072 (COPY_DECL_ASSEMBLER_NAME): Likewise.
10073 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
10075 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
10077 * config.gcc (i386, x86_64): Add extra objects.
10078 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
10079 (ix86_min_insn_size): Declare.
10080 (ix86_issue_rate): Declare.
10081 (ix86_adjust_cost): Declare.
10082 (ia32_multipass_dfa_lookahead): Declare.
10083 (ix86_macro_fusion_p): Declare.
10084 (ix86_macro_fusion_pair_p): Declare.
10085 (ix86_bd_has_dispatch): Declare.
10086 (ix86_bd_do_dispatch): Declare.
10087 (ix86_core2i7_init_hooks): Declare.
10088 (ix86_atom_sched_reorder): Declare.
10089 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
10090 (COSTS_N_BYTES): Move to x86-tune-costs.h.
10091 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
10092 (rip_relative_addr_p): Rename to ...
10093 (ix86_rip_relative_addr_p): ... this one; export.
10094 (memory_address_length): Update.
10095 (ix86_issue_rate): Move to x86-tune-sched.c.
10096 (ix86_flags_dependent): Move to x86-tune-sched.c.
10097 (ix86_agi_dependent): Move to x86-tune-sched.c.
10098 (exact_dependency_1): Move to x86-tune-sched.c.
10099 (exact_store_load_dependency): Move to x86-tune-sched.c.
10100 (ix86_adjust_cost): Move to x86-tune-sched.c.
10101 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
10102 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
10103 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
10104 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
10105 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
10106 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
10107 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
10108 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
10109 (min_insn_size): Rename to ...
10110 (ix86_min_insn_size): ... this one; export.
10111 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
10112 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
10113 (core2i7_first_cycle_multipass_backtrack): Move to
10114 x86-tune-sched-core.c.
10115 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
10116 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
10117 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
10118 (ix86_avoid_jump_mispredicts): Update.
10119 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
10120 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
10121 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
10122 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
10123 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
10124 (MAX_INSN): Move to ix86-tune-sched-bd.c.
10125 (MAX_IMM): Move to ix86-tune-sched-bd.c.
10126 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
10127 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
10128 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
10129 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
10130 (MAX_STORE): Move to ix86-tune-sched-bd.c.
10131 (BIG): Move to ix86-tune-sched-bd.c.
10132 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
10133 (enum insn_path): Move to ix86-tune-sched-bd.c.
10134 (get_mem_group): Move to ix86-tune-sched-bd.c.
10135 (is_cmp): Move to ix86-tune-sched-bd.c.
10136 (dispatch_violation): Move to ix86-tune-sched-bd.c.
10137 (is_branch): Move to ix86-tune-sched-bd.c.
10138 (is_prefetch): Move to ix86-tune-sched-bd.c.
10139 (init_window): Move to ix86-tune-sched-bd.c.
10140 (allocate_window): Move to ix86-tune-sched-bd.c.
10141 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
10142 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
10143 (process_end_window): Move to ix86-tune-sched-bd.c.
10144 (allocate_next_window): Move to ix86-tune-sched-bd.c.
10145 (find_constant): Move to ix86-tune-sched-bd.c.
10146 (get_num_immediates): Move to ix86-tune-sched-bd.c.
10147 (has_immediate): Move to ix86-tune-sched-bd.c.
10148 (get_insn_path): Move to ix86-tune-sched-bd.c.
10149 (get_insn_group): Move to ix86-tune-sched-bd.c.
10150 (count_num_restricted): Move to ix86-tune-sched-bd.c.
10151 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
10152 (add_insn_window): Move to ix86-tune-sched-bd.c.
10153 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
10154 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
10155 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
10156 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
10157 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
10158 (do_dispatch): Move to ix86-tune-sched-bd.c.
10159 (has_dispatch): Move to ix86-tune-sched-bd.c.
10160 * config/i386/t-i386: Add new object files.
10161 * config/i386/x86-tune-costs.h: New file.
10162 * config/i386/x86-tune-sched-atom.c: New file.
10163 * config/i386/x86-tune-sched-bd.c: New file.
10164 * config/i386/x86-tune-sched-core.c: New file.
10165 * config/i386/x86-tune-sched.c: New file.
10167 2017-10-11 Liu Hao <lh_mouse@126.com>
10169 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
10170 the generic version below instead.
10171 (should_colorize): Recognize Windows consoles as terminals
10173 * pretty-print.c [__MINGW32__] (write_all): New function.
10174 [__MINGW32__] (find_esc_head): Likewise.
10175 [__MINGW32__] (find_esc_terminator): Likewise.
10176 [__MINGW32__] (eat_esc_sequence): Likewise.
10177 [__MINGW32__] (mingw_ansi_fputs): New function that handles
10179 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
10182 2017-10-11 Richard Biener <rguenther@suse.de>
10184 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
10185 Properly call analyze_scalar_evolution with the loop of the stmt.
10187 2017-10-11 Richard Biener <rguenther@suse.de>
10189 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
10190 * tree-core.h (tree_base): Add chrec_var union member.
10191 * tree.h (CHREC_VAR): Remove.
10192 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
10193 * tree-chrec.h (build_polynomial_chrec): Adjust.
10194 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
10195 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
10197 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
10199 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
10200 * match.pd: ... here.
10201 ((T) X == (T) Y): Relax condition.
10203 2017-10-11 Bin Cheng <bin.cheng@arm.com>
10205 PR tree-optimization/82472
10206 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
10208 (break_alias_scc_partitions): Update postorder number.
10210 2017-10-11 Martin Liska <mliska@suse.cz>
10213 * opts.c (parse_no_sanitize_attribute): Do not use error_value
10215 * opts.h (parse_no_sanitize_attribute): Remove last argument.
10217 2017-10-11 Martin Liska <mliska@suse.cz>
10219 * print-rtl.c (print_insn): Move declaration of idbuf
10220 to same scope as name.
10222 2017-10-11 Martin Liska <mliska@suse.cz>
10227 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10228 volatile arguments.
10230 2017-10-11 Martin Liska <mliska@suse.cz>
10233 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
10234 volatile arguments.
10236 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
10238 * config.gcc (default_gnu_indirect_function): Default to yes for
10239 arm*-*-linux* with glibc.
10241 2017-10-11 Richard Biener <rguenther@suse.de>
10243 * tree-scalar-evolution.c (get_scalar_evolution): Handle
10244 default-defs and types we do not want to analyze.
10245 (interpret_loop_phi): Replace unreachable code with an assert.
10246 (compute_scalar_evolution_in_loop): Remove and inline ...
10247 (analyze_scalar_evolution_1): ... here, replacing condition with
10248 what makes the intent clearer. Remove handling of cases
10249 get_scalar_evolution now handles.
10251 2017-10-10 Jim Wilson <wilson@tuliptree.org>
10253 PR rtl-optimization/81434
10254 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
10255 comment for main loop. In sched_group_found if, also add checks for
10256 pass and min_cost_group.
10258 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
10260 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
10261 (rs6000_insn_cost): New function.
10262 * config/rs6000/rs6000.md (cost): New attribute.
10264 2017-10-10 Jakub Jelinek <jakub@redhat.com>
10265 H.J. Lu <hongjiu.lu@intel.com>
10269 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
10270 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
10271 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
10272 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
10273 to OPTION_MASK_ISA_AVX512VL - builtins that have both
10274 OPTION_MASK_ISA_MMX and some other bit set require both
10275 mmx and the ISAs without the mmx bit.
10276 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
10277 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
10278 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
10279 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
10280 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
10281 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
10282 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
10283 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
10284 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
10285 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
10286 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
10287 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
10288 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
10289 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
10290 Add OPTION_MASK_ISA_MMX.
10292 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
10294 * config.gcc (armv7*-*-freebsd*): New target.
10295 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
10297 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
10299 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
10300 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
10303 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
10305 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
10306 a template parameter.
10307 (WIDE_INT_REF_FOR): Update accordingly.
10308 * tree.h (wi::int_traits <const_tree>): Delete.
10309 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
10310 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
10311 (wi::tree_to_wide_ref): New typedef.
10312 (wi::to_wide): New function.
10313 * calls.c (get_size_range): Use wi::to_wide when operating on
10314 trees as wide_ints.
10315 * cgraph.c (cgraph_node::create_thunk): Likewise.
10316 * config/i386/i386.c (ix86_data_alignment): Likewise.
10317 (ix86_local_alignment): Likewise.
10318 * dbxout.c (stabstr_O): Likewise.
10319 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
10320 * expr.c (const_vector_from_tree): Likewise.
10321 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
10322 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
10323 (fold_negate_expr_1, int_const_binop_1, const_binop)
10324 (fold_convert_const_int_from_real, optimize_bit_field_compare)
10325 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
10326 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
10327 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
10328 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
10329 (fold_not_const, round_up_loc): Likewise.
10330 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
10331 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
10332 (alloca_call_type): Likewise.
10333 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
10334 * godump.c (go_output_typedef): Likewise.
10335 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
10336 * internal-fn.c (get_min_precision): Likewise.
10337 * ipa-cp.c (ipcp_store_vr_results): Likewise.
10338 * ipa-polymorphic-call.c
10339 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
10340 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
10341 (ipa_modify_call_arguments): Likewise.
10342 * match.pd: Likewise.
10343 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
10344 * print-tree.c (print_node_brief, print_node): Likewise.
10345 * stmt.c (expand_case): Likewise.
10346 * stor-layout.c (layout_type): Likewise.
10347 * tree-affine.c (tree_to_aff_combination): Likewise.
10348 * tree-cfg.c (group_case_labels_stmt): Likewise.
10349 * tree-data-ref.c (dr_analyze_indices): Likewise.
10350 (prune_runtime_alias_test_list): Likewise.
10351 * tree-dump.c (dequeue_and_dump): Likewise.
10352 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
10353 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
10354 * tree-pretty-print.c (dump_generic_node): Likewise.
10355 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
10356 (simple_iv_with_niters): Likewise.
10357 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
10358 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
10359 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
10360 * tree-ssa-loop-niter.c (split_to_var_and_offset)
10361 (refine_value_range_using_guard, number_of_iterations_ne_max)
10362 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
10363 (get_cst_init_from_scev, record_nonwrapping_iv)
10364 (scev_var_range_cant_overflow): Likewise.
10365 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
10366 * tree-ssa-pre.c (compute_avail): Likewise.
10367 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
10368 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
10369 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
10370 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
10371 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
10372 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
10373 (dump_case_nodes, try_switch_expansion): Likewise.
10374 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
10375 (vect_do_peeling): Likewise.
10376 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
10377 * tree-vect-stmts.c (vectorizable_load): Likewise.
10378 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
10379 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
10380 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
10381 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
10382 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
10383 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
10384 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
10385 (evrp_dom_walker::before_dom_children): Likewise.
10386 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
10387 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
10388 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
10389 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
10390 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
10391 (get_range_pos_neg): Likewise.
10392 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
10393 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
10394 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
10395 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
10396 * config/avr/avr.c (avr_fold_builtin): Likewise.
10397 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
10398 * config/msp430/msp430.c (msp430_attr): Likewise.
10399 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
10400 * config/powerpcspe/powerpcspe-c.c
10401 (altivec_resolve_overloaded_builtin): Likewise.
10402 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
10403 (rs6000_expand_ternop_builtin): Likewise.
10404 * config/rs6000/rs6000-c.c
10405 (altivec_resolve_overloaded_builtin): Likewise.
10406 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
10407 (rs6000_expand_ternop_builtin): Likewise.
10408 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
10410 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10412 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
10413 when copying loop nest with only one inner loop.
10415 2017-10-10 Richard Biener <rguenther@suse.de>
10417 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
10418 blocks if SCEV is active.
10419 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
10421 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
10422 (scev_initialize): Assert we are not yet initialized.
10424 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10426 * tree-loop-distribution.c (generate_loops_for_partition): Remove
10427 inner loop's exit stmt by making it always exit the loop, otherwise
10428 we would generate an infinite empty loop.
10430 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10432 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
10433 renaming variables in new preheader if it's deleted.
10435 2017-10-10 Bin Cheng <bin.cheng@arm.com>
10437 * tree-loop-distribution.c (struct partition): Remove unused field
10438 loops of the structure.
10439 (partition_alloc, partition_free): Ditto.
10440 (build_rdg_partition_for_vertex): Ditto.
10442 2017-10-09 Jeff Law <law@redhat.com>
10444 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
10445 return type to match prototype and documentation.
10447 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10449 * config/rs6000/rs6000.c (processor_costs): Move to ...
10450 * config/rs6000/rs6000.h: ... here.
10451 (rs6000_cost): Declare.
10453 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
10455 * except.c (setjmp_fn): New global variable.
10456 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
10457 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
10458 if DONT_USE_BUILTIN_SETJMP is defined.
10460 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10462 * target.def (insn_cost): New hook.
10463 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
10464 * doc/tm.texi: Regenerate.
10465 * rtlanal.c (insn_cost): Use the new hook.
10467 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10469 * combine.c (combine_validate_cost): Compute the new insn_cost,
10470 not just pattern_cost.
10471 (try_combine): Adjust comment.
10473 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
10475 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
10477 * combine.c (uid_insn_cost): Adjust comment.
10478 (combine_validate_cost): Adjust comment. Use pattern_cost instead
10480 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
10481 * dse.c (find_shift_sequence): Ditto.
10482 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
10483 (bb_valid_for_noce_process_p): Use pattern_cost.
10484 * rtl.h (insn_rtx_cost): Delete.
10485 (pattern_cost): New prototype.
10486 (insn_cost): New prototype.
10487 * rtlanal.c (insn_rtx_cost): Rename to...
10488 (pattern_cost): ... this.
10491 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
10493 * config/i386/i386.md (*jcc_2): Remove insn pattern.
10494 (*jcc<mode>_0_r_i387): Ditto.
10495 (*jccxf_r_i387): Ditto.
10496 (*jcc<mode>_r_i387): Ditto.
10497 (*jccu<mode>_r_i387): Ditto.
10498 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
10499 (*jcc): Rename from *jcc_1.
10501 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10503 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
10504 deferred rescans after the lvx/stvx recombination pre-pass.
10506 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
10508 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
10509 memory operation instruction support.
10510 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
10511 (rs6000-ibm-aix[789]*): Likewise.
10512 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
10513 Document new functions.
10515 2017-10-09 Richard Biener <rguenther@suse.de>
10517 PR tree-optimization/82397
10518 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
10519 equality only for semantically equal trees.
10521 2017-10-09 Richard Biener <rguenther@suse.de>
10523 PR tree-optimization/82449
10524 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
10525 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
10526 allow constant addresses.
10527 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
10530 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10532 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
10535 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10538 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
10541 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10544 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
10546 2017-10-09 Jakub Jelinek <jakub@redhat.com>
10549 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
10550 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
10552 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
10554 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
10555 (WI_BINARY_PREDICATE_RESULT): Likewise.
10556 (wi::binary_traits::operator_result): New type.
10557 (wi::binary_traits::predicate_result): Likewise.
10558 (generic_wide_int::operator~, unary generic_wide_int::operator-)
10559 (generic_wide_int::operator==, generic_wide_int::operator!=)
10560 (generic_wide_int::operator&, generic_wide_int::and_not)
10561 (generic_wide_int::operator|, generic_wide_int::or_not)
10562 (generic_wide_int::operator^, generic_wide_int::operator+
10563 (binary generic_wide_int::operator-, generic_wide_int::operator*):
10565 (operator~, unary operator-, operator==, operator!=, operator&)
10566 (operator|, operator^, operator+, binary operator-, operator*): New
10568 * expr.c (get_inner_reference): Use wi::bit_and_not.
10569 * fold-const.c (fold_binary_loc): Likewise.
10570 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
10571 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
10572 (bit_value_binop): Likewise.
10573 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
10574 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
10575 (extract_range_from_binary_expr_1): Likewise.
10576 (masked_increment): Likewise.
10577 (simplify_bit_ops_using_ranges): Likewise.
10579 2017-10-09 Martin Jambor <mjambor@suse.cz>
10582 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
10583 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
10584 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
10586 (hsa_fixup_mov_insn_type): New function.
10587 (hsa_op_with_type::get_in_type): Use it.
10588 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
10589 immediates in an assert.
10590 (hsa_op_with_type::extend_int_to_32bit): New method.
10591 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
10592 types. Convert to dest type if necessary.
10593 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
10594 (reg_for_gimple_ssa): Pass false as min32int to
10595 hsa_type_for_scalar_tree_type.
10596 (gen_hsa_addr): Fixup type when creating addresable temporary.
10597 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
10598 (gen_hsa_unary_operation): Extend operands and convert to dest type if
10599 necessary. Call hsa_fixup_mov_insn_type.
10600 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
10601 extend operands and convert to dest type if necessary.
10602 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
10603 to dest type if necessary.
10604 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
10605 if conversion nt necessary and size matches.
10606 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
10607 to dest type if necessary.
10608 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
10609 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
10611 (gen_hsa_clrsb): Likewise.
10612 (gen_hsa_ffs): Likewise.
10613 (gen_hsa_divmod): Extend operands and convert to dest type if
10615 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
10617 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10619 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
10620 Remove empty default arguments. Use a brace block as output
10622 (conditional return): Ditto.
10624 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
10625 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
10627 (group_ending_nop): Ditto.
10628 (doloop_end): Ditto.
10629 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
10630 (splitters for those): Ditto.
10632 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
10634 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
10635 a conditional jump (and the compare for it) so that pc_rtx is the
10637 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
10638 for the deleted and renamed ctr<mode>_internal[234] patterns.
10639 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
10640 Delete second conditional return pattern.
10641 (ctr<mode>_internal2): Delete this second bdnz pattern.
10642 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
10643 (ctr<mode>_internal4): Delete this second bdz pattern.
10645 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10647 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
10648 (always_initialized_rtx_for_ssa_name_p): New predicate.
10649 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
10650 (finish_out_of_ssa): Free new field of SA.
10651 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
10652 * tree-ssa-coalesce.c: Include tree-ssa.h.
10653 (get_parm_default_def_partitions): Remove extern keyword.
10654 (get_undefined_value_partitions): New function.
10655 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
10656 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
10657 uninitialized bits.
10658 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
10660 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
10662 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
10664 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10666 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
10667 for targets that preffer 128bit.
10669 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
10671 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
10673 2017-10-08 Olivier Hainque <hainque@adacore.com>
10675 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
10676 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
10678 (thumb_set_return_address): Likewise.
10680 2017-10-08 Olivier Hainque <hainque@adacore.com>
10682 * common/config/arm/arm-common.c (arm_except_unwind_info):
10683 Handle DWARF2_UNWIND_INFO.
10685 2017-10-07 Michael Collison <michael.collison@arm.com>
10687 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
10690 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
10692 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
10693 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
10694 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
10695 defined, force the creation of a new block for a dispatch label.
10697 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
10699 * invoke.texi (Wsuggest-attribute=cold): Document.
10700 * common.opt (Wsuggest-attribute=cold): New
10701 * ipa-pure-const.c (warn_function_cold): New function.
10702 * predict.c (compute_function_frequency): Use it.
10703 * predict.h (warn_function_cold): Declare.
10705 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
10707 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
10710 2017-10-06 Martin Liska <mliska@suse.cz>
10712 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
10713 keyword for member functions.
10714 (struct sanopt_tree_couple): New struct.
10715 (struct sanopt_tree_couple_hash): New function.
10716 (struct sanopt_ctx): Add new hash_map.
10717 (has_dominating_ubsan_ptr_check): New function.
10718 (record_ubsan_ptr_check_stmt): Likewise.
10719 (maybe_optimize_ubsan_ptr_ifn): Likewise.
10720 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
10721 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
10723 2017-10-06 Sudakshina Das <sudi.das@arm.com>
10726 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
10727 aarch64_simd_valid_immediate on CONST_VECTORs.
10728 (aarch64_reg_or_bic_imm): Likewise.
10730 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
10732 PR rtl-optimization/82396
10733 * haifa-sched.c (ready_sort_real): Disable qsort checking.
10735 2017-10-06 Sebastian Pop <sebpop@gmail.com>
10737 * graphite-dependences.c (scop_get_reads): Move code to...
10738 (scop_get_must_writes): Move code to...
10739 (scop_get_may_writes): Move code to...
10740 (scop_get_reads_and_writes): ... here.
10741 (scop_get_dependences): Call scop_get_reads_and_writes.
10743 2017-10-06 Jakub Jelinek <jakub@redhat.com>
10745 PR tree-optimization/82434
10746 * fold-const.h (can_native_encode_type_p,
10747 can_native_encode_string_p): Remove.
10748 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
10749 don't encode anything, just return what would be otherwise returned.
10750 (native_encode_fixed, native_encode_complex, native_encode_vector):
10752 (native_encode_string): Likewise. Inline by hand
10753 can_native_encode_string_p.
10754 (can_native_encode_type_p): Remove.
10755 (can_native_encode_string_p): Remove.
10756 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
10757 STRING_CSTs using can_native_encode_string_p, test all
10758 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
10759 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
10760 argument from native_encode_expr.
10761 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
10762 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
10765 2017-10-06 Richard Biener <rguenther@suse.de>
10767 PR tree-optimization/82397
10768 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
10769 operand_equal_p but rely on data_ref_compare_tree for detecting
10771 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
10772 to match up with dr_group_sort_cmp.
10774 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10777 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
10779 * config/s390/s390-builtin-types.def: Regenerate.
10781 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10784 * config/s390/s390-builtin-types.def: Regenerate.
10785 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
10786 Change flag from B_VXE to B_VX.
10787 (s390_vec_min_dbl): Remove B_VXE flag.
10789 2017-10-06 Richard Biener <rguenther@suse.de>
10791 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
10792 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
10793 translate_isl_ast_to_gimple::is_valid_rename,
10794 translate_isl_ast_to_gimple::get_rename,
10795 translate_isl_ast_to_gimple::get_def_bb_for_const,
10796 translate_isl_ast_to_gimple::get_new_name,
10797 translate_isl_ast_to_gimple::collect_all_ssa_names,
10798 translate_isl_ast_to_gimple::copy_loop_phi_args,
10799 translate_isl_ast_to_gimple::collect_all_ssa_names,
10800 translate_isl_ast_to_gimple::copy_loop_phi_args,
10801 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
10802 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
10803 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
10804 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
10805 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
10806 translate_isl_ast_to_gimple::copy_cond_phi_args,
10807 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
10808 translate_isl_ast_to_gimple::edge_for_new_close_phis,
10809 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
10810 translate_isl_ast_to_gimple::rename_uses,
10811 translate_isl_ast_to_gimple::rename_all_uses): Remove.
10812 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
10813 (set_rename_for_each_def): Likewise.
10814 (graphite_copy_stmts_from_block): Handle debug stmt resetting
10815 here. Handle rewriting SCEV analyzable uses here.
10816 (copy_bb_and_scalar_dependences): Generate code for PHI
10818 (graphite_regenerate_ast_isl): Adjust.
10819 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
10820 (add_write, add_read): New functions.
10821 (build_cross_bb_scalars_def): Use it and simplify.
10822 (build_cross_bb_scalars_use): Likewise.
10823 (graphite_find_cross_bb_scalar_vars): Inline into...
10824 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
10825 simulating out-of-SSA. Compute liveout and add dependencies.
10826 (build_scops): Force an empty entry block.
10827 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
10829 (sese_build_liveouts): Declare.
10830 (sese_trivially_empty_bb_p): Likewise.
10831 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
10832 compute liveout and debug_liveout.
10833 (sese_bad_liveouts_use): Remove.
10834 (sese_reset_debug_liveouts_bb): Likewise.
10835 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
10836 (sese_build_liveouts): Build liveout and debug_liveout and store
10838 (new_sese_info): Adjust.
10839 (free_sese_info): Likewise.
10840 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
10841 do not build liveout here.
10842 (move_sese_in_condition): Adjust region entry.
10843 (scev_analyzable_p): Match up with chrec_apply requirements.
10844 (sese_trivially_empty_bb_p): New.
10845 * tree-into-ssa.c (get_reaching_def): Properly support generating
10846 default-defs for incremental rewrite of anonymous names.
10848 2017-10-06 Richard Biener <rguenther@suse.de>
10850 * graphite-sese-to-poly.c (extract_affine): For casts increasing
10851 precision do not perform modulo reduction.
10853 2017-10-06 Richard Biener <rguenther@suse.de>
10855 PR tree-optimization/82436
10856 * tree-vect-slp.c (vect_supported_load_permutation_p): More
10857 conservatively choose the vectorization factor when checking
10858 whether we can perform the required load permutation.
10859 (vect_transform_slp_perm_load): Assert when we may not fail.
10861 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
10863 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
10864 message for incompatible -msdata=* and -mcall-* options.
10866 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10868 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
10869 rate for post-reload scheduling.
10871 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10873 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
10875 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10877 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
10878 to improve monte carlo in scimark.
10880 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
10882 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
10883 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
10884 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
10885 pentium4_cost, nocona_cost): Set reassociation width to 1.
10886 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
10887 width to 2 for fp operations and 1 otherwise.
10888 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
10890 (atom_cost): Set reassociation width to 2.
10891 (slm_cost, generic_cost): Set fp reassociation width
10892 to 2 and 1 otherwise.
10893 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
10894 (core_cost): Set fp reassociation width to 4 and vector to 2.
10895 (ix86_reassociation_width): Rewrite using cost table; special case
10896 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
10897 and TARGET_AVX128_OPTIMAL.
10898 * config/i386/i386.h (processor_costs): Add
10899 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
10900 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
10901 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
10902 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
10903 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
10904 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
10906 2017-10-05 Nathan Sidwell <nathan@acm.org>
10908 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
10910 2017-10-05 Tamar Christina <tamar.christina@arm.com>
10912 * config/arm/arm.c (arm_test_fpu_data): New.
10913 (arm_run_selftests): Call arm_test_fpu_data.
10915 2017-10-04 Nathan Sidwell <nathan@acm.org>
10917 * toplev.c (toplev::main): Remove excess parens on pretty_printer
10919 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
10921 2017-10-04 Sudakshina Das <sudi.das@arm.com>
10923 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
10924 check type for aarch64_simd_valid_immediate.
10925 (aarch64_output_simd_mov_immediate): Update prototype.
10926 (aarch64_simd_valid_immediate): Update prototype.
10927 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
10928 support for ORR-immediate.
10929 (and<mode>3): modified pattern to add support for BIC-immediate.
10930 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
10931 now checks for valid immediate for BIC and ORR based on new enum
10933 (aarch64_output_simd_mov_immediate): Function now used to output
10934 BIC/ORR imm as well based on new enum argument.
10935 * config/aarch64/constraints.md (Do): New vector immediate constraint.
10937 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
10938 (aarch64_reg_or_bic_imm): Likewise.
10940 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10942 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
10943 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
10945 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10949 PR rtl-optimization/82396
10950 * haifa-sched.c (autopref_multipass_init): Simplify
10952 (autopref_rank_data): Simplify sort order.
10953 * sched-int.h (autopref_multipass_data_): Remove
10954 multi_mem_insn_p, min_offset and max_offset.
10956 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10958 * doc/sourcebuild.texi: Document vect_peeling_profitable.
10960 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10962 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
10963 vect_doubleint_cvt.
10965 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10967 * doc/sourcebuild.texi: Document vect_long_mult.
10969 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
10971 PR tree-optimization/82413
10972 * fold-const.c (build_range_check): Use widest_int when comparing
10973 the maximum ETYPE value with HIGH.
10975 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
10977 PR rtl-optimization/82396
10978 * haifa-sched.c (autopref_multipass_init): Simplify
10980 (autopref_rank_data): Simplify sort order.
10981 * sched-int.h (autopref_multipass_data_): Remove
10982 multi_mem_insn_p, min_offset and max_offset.
10984 2017-10-04 Jakub Jelinek <jakub@redhat.com>
10986 PR tree-optimization/82381
10987 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
10988 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
10991 PR tree-optimization/82374
10992 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
10993 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
10994 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
10995 current_function_decl to the new decl.
10997 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
10999 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
11000 helper macro for IEEE float128 hardware built-in functions.
11001 (SQRTF128_ODD): Add built-in functions with the round-to-odd
11003 (TRUNCF128_ODD): Likewise.
11004 (ADDF128_ODD): Likewise.
11005 (SUBF128_ODD): Likewise.
11006 (MULF128_ODD): Likewise.
11007 (DIVF128_ODD): Likewise.
11008 (FMAF128_ODD): Likewise.
11009 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
11010 UNSPEC_TRUNC_ROUND_TO_ODD.
11011 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
11012 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
11013 floating point round to odd instructions.
11014 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
11015 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
11016 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
11017 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
11018 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
11019 (trunc<mode>sf2_hw): Change the truncate with round to odd
11020 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
11021 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
11022 to odd hardware instructions.
11023 (sub<mode>3_odd): Likewise.
11024 (mul<mode>3_odd): Likewise.
11025 (div<mode>3_odd): Likewise.
11026 (sqrt<mode>2_odd): Likewise.
11027 (fma<mode>4_odd): Likewise.
11028 (fms<mode>4_odd): Likewise.
11029 (nfma<mode>4_odd): Likewise.
11030 (nfms<mode>4_odd): Likewise.
11031 (trunc<mode>df2_odd): Change the truncate with round to odd
11032 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
11034 * doc/extend.texi (PowerPC built-in functions): Update documentation
11035 for existing IEEE float128-bit built-in functions. Add built-in
11036 functions that generate the IEEE 128-bit floating point round to
11039 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
11041 PR rtl-optimization/77729
11042 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
11043 to (X&(C1&~C2))|C2 transformations.
11045 2017-10-03 Martin Jambor <mjambor@suse.cz>
11047 PR tree-optimization/82363
11048 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
11049 mismatch, mark lacc written regardless of racc.
11051 2017-10-03 Jakub Jelinek <jakub@redhat.com>
11053 PR tree-optimization/82381
11054 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
11055 stmt_to_insert nor wheather SSA_NAMEs are default defs.
11056 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
11057 fallthrough into reassoc_stmt_dominates_stmt_p.
11060 * combine.c (combine_instructions): Don't combine in unreachable
11063 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
11066 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
11067 function to not use the have_cpu variable. Do not set cpu_index,
11068 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
11069 or the default cpu.
11070 (rs6000_valid_attribute_p): Remove duplicate initializations of
11071 old_optimize and func_optimize.
11072 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
11073 (rs6000_activate_target_options): Make global.
11074 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
11077 2017-10-02 Jakub Jelinek <jakub@redhat.com>
11079 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
11080 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
11081 Set *poffset to 0 and *psize and *pmax_size to -1 if
11082 *poffset + *psize overflows in HOST_WIDE_INT.
11084 PR tree-optimization/82387
11085 PR tree-optimization/82388
11086 PR tree-optimization/82389
11087 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
11088 instead of live_bytes non-NULL.
11090 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
11093 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
11096 2017-10-02 Richard Biener <rguenther@suse.de>
11098 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
11099 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
11100 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
11102 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11104 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
11105 requested precision matches the type's.
11106 * calls.c (alloc_max_size): Calculate the new candidate size as
11107 a widest_int and use wi::to_widest when comparing it with the
11108 current candidate size.
11109 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
11110 zero rather than integer_zero_node.
11111 * match.pd: Check for a no-op conversion before using wi::add
11112 rather than after. Use tree_to_uhwi when summing small shift
11113 counts into an unsigned int.
11115 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
11116 Alan Hayward <alan.hayward@arm.com>
11117 David Sherwood <david.sherwood@arm.com>
11120 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
11121 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
11122 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
11123 POINTER_AND_FP_REGS.
11125 2017-10-02 Richard Biener <rguenther@suse.de>
11127 PR tree-optimization/82355
11128 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
11129 a mapping for the enclosing loop but avoid generating one for
11130 the loop tree root.
11131 (copy_bb_and_scalar_dependences): Remove premature codegen
11132 error on PHIs in blocks duplicated into multiple places.
11133 * graphite-scop-detection.c
11134 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
11135 in the region use it as loop and nest to analyze the DR in.
11136 (try_generate_gimple_bb): Likewise.
11137 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
11138 (add_loop_constraints): For blocks in a loop not in the region
11139 create a dimension with a single iteration.
11140 * sese.h (gbb_loop_at_index): Remove assert.
11142 2017-10-01 Kevin Buettner <kevinb@redhat.com>
11144 * omp-expand.c (adjust_context_scope): New function.
11145 (expand_parallel_call): Call adjust_context_scope.
11147 2017-10-01 Jeff Law <law@redhat.com>
11149 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
11150 dom_opt_dom_walker class with direct access to private members.
11151 Add comments. Call test_for_singularity.
11152 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
11153 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
11154 m_dummy_cond anymore.
11155 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
11157 (pass_dominator:execute): Build the dummy_cond here and pass it
11158 to the dom_opt_dom_walker ctor.
11159 (test_for_singularity): New function.
11161 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
11162 Maya Rashish <coypu@sdf.org>
11164 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
11165 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
11166 (alpha*-*-netbsd*) Use nbsd_tm_file.
11167 (arm*-*-netbsdelf*) Likewise.
11168 (i[34567]86-*-netbsdelf*) Likewise.
11169 (x86_64-*-netbsd*) Likewise.
11170 (mips*-*-netbsd*) Likewise.
11171 (powerpc-*-netbsd*) Likewise.
11172 (sh*-*-netbsd*) Likewise.
11173 (sparc-*-netbsdelf*) Likewise.
11174 (sparc64-*-netbsd*) Likewise.
11175 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
11177 (vax-*-netbsdelf*) Likewise.
11178 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
11179 (UINT_FAST8_TYPE) Likewise.
11180 (INT_FAST16_TYPE) Check CHAR_FAST16.
11181 (UINT_FAST16_TYPE) Likewise.
11183 2017-09-30 Jakub Jelinek <jakub@redhat.com>
11186 * config/i386/i386.md
11187 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
11188 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
11189 *divmodsi4_zext_2): New define_insn_and_split.
11190 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
11191 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
11192 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
11193 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
11194 New define_insn_and_split.
11195 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
11196 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
11197 operands[1] having DImode when mode is SImode.
11199 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
11200 always SImode for DIV and MOD in REG_EQUAL notes.
11202 2017-09-29 Yury Gribov <tetra2005@gmail.com>
11204 PR middle-end/82319
11205 * match.pd: Fix handling of NaNs in pattern.
11207 2017-09-29 Jeff Law <law@redhat.com>
11209 * sbitmap.c (bitmap_bit_in_range_p): New function.
11210 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
11211 * tree-ssa-dse.c (live_bytes_read): New function.
11212 (dse_classify_store): Ignore reads of dead bytes.
11214 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
11215 typos and whitespace errors.
11216 * config/i386/predicates.md (address_no_seg_operand): Likewise.
11217 * config/s390/s390.c (s390_emit_prologue): Likewise.
11219 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11222 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
11223 with a symbol for LRA.
11225 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
11227 PR rtl-optimization/82338
11228 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
11230 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
11232 * genmodes.c (calc_wider_mode): Suppress qsort macro.
11233 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
11234 (qsort_chk): Declare.
11235 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
11236 (qsort_chk): New function.
11238 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11240 PR tree-optimization/82337
11241 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
11242 phi definition if the PHI result appears in an abnormal PHI.
11243 (find_basis_for_base_expr): Don't record a basis if the LHS of the
11244 basis appears in an abnormal PHI.
11246 2017-09-29 Richard Biener <rguenther@suse.de>
11248 * graphite-isl-ast-to-gimple.c
11249 (translate_isl_ast_to_gimple::set_codegen_error): New function.
11250 (binary_op_to_tree): Use it.
11251 (get_rename_from_scev): Likewise.
11252 (copy_loop_phi_nodes): Likewise.
11253 (copy_bb_and_scalar_dependences): Likewise.
11254 (translate_pending_phi_nodes): Likewise.
11256 2017-09-29 Jakub Jelinek <jakub@redhat.com>
11259 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
11260 for movabsq $(i32 << shift), r64.
11262 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
11264 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
11265 index when encoding %esp as %rsp to avoid 0x67 prefix.
11267 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
11269 * config/i386/i386.md (*movsf_internal, *movdf_internal):
11270 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
11272 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
11274 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
11275 Extensions with more than 16 double VFP registers.
11276 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
11277 to_clear_mask and all code related to it. Replace the remaining
11278 entry by a sbitmap and adapt code accordingly.
11280 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
11282 * brig-builtins.def: Change pure attributes to const.
11284 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11286 * config.gcc (default_gnu_indirect_function): Default to yes for
11287 sparc*-*-linux* with glibc.
11289 2017-09-28 Joseph Myers <joseph@codesourcery.com>
11291 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
11292 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
11293 when creating .init_array and .fini_array sections with priority
11296 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
11299 * config/aarch64/aarch64.c
11300 (aarch64_builtin_support_vector_misalignment): Always return false
11301 when misalignment is unknown.
11303 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
11305 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
11306 this function to return false if the definition used by the swap
11307 instruction is artificial, or if the memory address from which the
11308 constant value is loaded is not represented by a base address held
11309 in a register or if the base address register is a frame or stack
11310 pointer. Additionally, return false if the base address of the
11311 loaded constant is a SYMBOL_REF but is not considered to be a
11313 (replace_swapped_load_constant): New function.
11314 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
11315 loaded constant vector with a load of a swapped constant vector.
11317 2017-09-27 Carl Love <cel@us.ibm.com>
11319 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
11320 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
11321 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
11324 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
11326 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
11327 first, always call autopref_rank_data otherwise.
11329 2017-09-27 Richard Biener <rguenther@suse.de>
11331 * graphite-scop-detection.c (find_scop_parameters): Move
11332 loop bound handling ...
11333 (gather_bbs::before_dom_children): ... here, avoiding the need
11334 to build scop_info->loop_nest.
11335 (record_loop_in_sese): Remove.
11336 * sese.h (sese_info_t::loop_nest): Remove.
11337 * sese.c (new_sese_info): Do not allocate loop_nest.
11338 (free_sese_info): Do not free loop_nest.
11340 2017-09-27 Jakub Jelinek <jakub@redhat.com>
11343 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
11344 lhs from calls if the lhs has addressable type.
11346 2017-09-27 Richard Biener <rguenther@suse.de>
11348 * graphite.h (scop::max_alias_set): New member.
11349 * graphite-scop-detection.c: Remove references to non-existing
11350 --param in comments.
11351 (build_alias_sets): Record the maximum alias set used for drs.
11352 (build_scops): Support zero as unlimited for
11353 --param graphite-max-arrays-per-scop.
11354 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
11355 and inline into ...
11356 (build_poly_sr_1): ... here. Compute alias set based on the
11357 maximum alias set used for drs rather than
11358 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
11360 2017-09-27 Richard Biener <rguenther@suse.de>
11362 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
11363 --param loop-block-tile-size=0 to disable tiling.
11365 2017-09-27 Richard Biener <rguenther@suse.de>
11367 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
11368 (graphite-max-nb-scop-params): Document special value zero.
11369 * domwalk.h (dom_walker::STOP): New symbolical constant.
11370 (dom_walker::dom_walker): Add optional parameter for bb to
11372 (dom_walker::~dom_walker): Declare.
11373 (dom_walker::before_dom_children): Document STOP return value.
11374 (dom_walker::m_user_bb_to_rpo): New member.
11375 (dom_walker::m_bb_to_rpo): Likewise.
11376 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
11377 mapping here if not provided by the user.
11378 (dom_walker::~dom_walker): Free bb to RPO mapping if not
11379 provided by the user.
11380 (dom_walker::STOP): Define.
11381 (dom_walker::walk): Do not compute bb to RPO mapping here.
11382 Support STOP return value from before_dom_children to stop
11384 * graphite-optimize-isl.c (optimize_isl): If the schedule
11385 is the same still generate code if -fgraphite-identity
11386 or -floop-parallelize-all are given.
11387 * graphite-scop-detection.c: Include cfganal.h.
11388 (gather_bbs::gather_bbs): Get and pass through bb to RPO
11390 (gather_bbs::before_dom_children): Return STOP for BBs
11392 (build_scops): Compute bb to RPO mapping and pass it to
11393 the domwalk. Treat --param graphite-max-nb-scop-params=0
11394 as not limiting the number of params.
11395 * graphite.c (graphite_initialize): Remove limit on the
11396 number of basic-blocks in a function.
11397 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
11398 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
11399 default value of 10.
11401 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11403 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
11404 Adjust code to eliminate needing to do the shift right 32-bits
11405 operation after XSCVDPSPN.
11407 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
11409 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
11410 ((X / Y) != 0 -> X >= Y): Likewise.
11412 2017-09-26 Carl Love <cel@us.ibm.com>
11414 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
11415 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
11416 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
11417 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
11418 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
11419 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
11420 definitions and overloading.
11421 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
11422 statement for P9V_BUILTIN_XST_LEN_R.
11423 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
11424 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
11425 define_expand and define_insn for the instructions and builtins.
11426 * doc/extend.texi: Update the built-in documentation file for the new
11427 built-in functions.
11428 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
11429 define_insn for the instructions
11431 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
11434 * gcc/config/netbsd-protos.h: New file.
11435 * gcc/config/netbsd.c: New file.
11436 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
11437 * gcc/config/t-netbsd: New file.
11438 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
11439 (tmake_file) Add t-netbsd.
11440 (extra_objs) Add netbsd.o.
11442 2017-09-26 Janus Weil <janus@gcc.gnu.org>
11446 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
11448 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11450 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
11451 sign extension from a vector register to a GPR by doing a 32-bit
11452 direct move and then an EXTSW.
11453 (extendsi<mode>2 splitter): Likewise.
11454 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
11455 right or vector extract after doing XSCVDPSPN. Use
11456 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
11458 (movdi_from_sf_zero_ext): Likewise.
11459 (reload_gpr_from_vsxsf): Likewise.
11460 (p8_mfvsrd_4_disf): Delete, no longer used.
11461 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
11462 then needing to move the SFmode to a GPR to use the XSCVDPSP
11463 instruction instead of FRSP and XSCVDPSPN.
11464 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
11465 it is adjacent to the other XSCVSPDP insns.
11466 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
11467 SFmode to be in traditional Altivec registers.
11468 (vsx_xscvdpspn): Eliminate useless alternative constraint.
11469 (vsx_xscvspdpn): Likewise.
11470 (vsx_xscvspdpn_scalar): Likewise.
11472 2017-09-26 Martin Jambor <mjambor@suse.cz>
11474 * tree-sra.c (compare_access_positions): Put integral types first,
11475 stabilize sorting of integral types, remove conditions putting
11476 non-full-precision integers last.
11477 (sort_and_splice_var_accesses): Disable scalarization if a
11478 non-integert would be represented by a non-full-precision integer.
11480 2017-09-26 Joseph Myers <joseph@codesourcery.com>
11482 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
11483 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
11484 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
11485 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
11486 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
11487 conditionals inside the function instead of around it. Call
11488 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
11489 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
11491 2017-09-26 Richard Biener <rguenther@suse.de>
11493 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
11495 (scop_detection::build_scop_breadth): ... this. Removed.
11496 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
11497 (scop_detection::harmful_stmt_in_bb): Likewise.
11498 (scop_detection::graphite_can_represent_stmt): Likewise.
11499 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
11500 (scop_detection::can_represent_loop): Remove recursion, fold in ...
11501 (scop_detection::can_represent_loop_1): ... this. Removed.
11502 (scop_detection::harmful_loop_in_region): Simplify after inlining
11503 the above and remove more quadraticness.
11504 (build_scops): Adjust.
11505 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
11508 2017-09-26 Jakub Jelinek <jakub@redhat.com>
11511 * config/i386/i386.c (ix86_print_operand_address_as): Only test
11512 REGNO (base) == SP_REG if base is a REG.
11514 PR middle-end/35691
11515 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
11516 if it is different SSA_NAME.
11517 (optimize_range_tests_cmp_bitwise): New function.
11518 (optimize_range_tests): Call it.
11520 2017-09-26 Richard Biener <rguenther@suse.de>
11522 PR tree-optimization/82321
11523 * graphite.c (canonicalize_loop_closed_ssa): Properly check
11524 for the def being inside the loop.
11526 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11528 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
11530 * config/s390/s390-builtins.def: Fix constraint on op4.
11532 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11534 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
11535 independent expanders.
11536 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
11537 ("vec_ordered", "vec_unordered"): New expanders.
11539 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11541 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
11544 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11546 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
11547 vec_unpacks_lo_v16qi.
11548 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
11550 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11552 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
11553 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
11554 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
11556 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11558 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
11560 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
11562 ("*vec_slb<mode>"): New insn pattern.
11563 ("vec_shr_<mode>"): New expander.
11564 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
11565 and force the shift count operand to V16QImode.
11566 ("vec_srb<mode>"): Set shift count mode to V16QI.
11568 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11570 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
11571 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
11572 ("vec_widen_smult_hi_<mode>"): New expander definitions.
11574 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
11577 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
11579 2017-09-26 Richard Biener <rguenther@suse.de>
11581 PR tree-optimization/82320
11582 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
11585 2017-09-25 Jeff Law <law@redhat.com>
11587 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
11588 prototype for new argument.
11589 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
11590 mostly extracted from rs6000_emit_allocate_stack.
11591 (rs6000_emit_probe_stack_range_stack_clash): New function.
11592 (rs6000_emit_allocate_stack): Call
11593 rs6000_emit_probe_stack_range_stack_clash as needed.
11594 (rs6000_emit_probe_stack_range): Add additional argument
11595 to call to gen_probe_stack_range{si,di}.
11596 (output_probe_stack_range): New.
11597 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
11598 (output_probe_stack_range_stack_clash): New.
11599 (rs6000_emit_prologue): Emit notes into dump file as requested.
11600 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
11601 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
11602 Add additional operand and pass it to output_probe_stack_range.
11604 2017-09-25 Bin Cheng <bin.cheng@arm.com>
11606 PR tree-optimization/82163
11607 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
11608 (checking_verify_loop_closed_ssa): New parameter.
11609 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
11610 (check_loop_closed_ssa_stmt): Delete.
11611 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
11612 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
11613 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
11616 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
11618 * brig-builtins.def: Treat HSAIL barrier builtins as
11619 setjmp/longjump style functions.
11621 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11623 * target.def (constant_alignment): New hook.
11624 * defaults.h (CONSTANT_ALIGNMENT): Delete.
11625 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
11626 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
11627 * doc/tm.texi: Regenerate.
11628 * targhooks.h (default_constant_alignment): Declare.
11629 (constant_alignment_word_strings): Likewise.
11630 * targhooks.c (default_constant_alignment): New function.
11631 (constant_alignment_word_strings): Likewise.
11632 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
11633 instead of CONSTANT_ALIGNMENT.
11634 * varasm.c (align_variable, get_variable_align, build_constant_desc)
11635 (force_const_mem): Likewise.
11636 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
11637 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
11638 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
11639 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11640 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
11642 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
11643 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11644 constant_alignment_word_strings.
11645 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
11646 (CONSTANT_ALIGNMENT): Likewise.
11647 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11648 (arm_constant_alignment): New function.
11649 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
11650 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11651 constant_alignment_word_strings.
11652 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
11653 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11654 constant_alignment_word_strings.
11655 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
11656 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11657 (cris_constant_alignment): New function.
11658 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
11659 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11660 (epiphany_constant_alignment): New function.
11661 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
11662 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11663 constant_alignment_word_strings.
11664 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
11665 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11666 constant_alignment_word_strings.
11667 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
11668 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11669 constant_alignment_word_strings.
11670 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
11671 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
11672 * config/i386/i386.c (ix86_constant_alignment): Make static.
11673 Use the same interface as the target hook.
11674 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11675 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
11676 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11677 constant_alignment_word_strings.
11678 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
11679 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
11680 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11681 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
11682 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11683 constant_alignment_word_strings.
11684 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
11685 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11686 constant_alignment_word_strings.
11687 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
11688 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11689 constant_alignment_word_strings.
11690 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
11691 * config/microblaze/microblaze.c (microblaze_constant_alignment):
11693 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11694 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
11695 * config/mips/mips.c (mips_constant_alignment): New function.
11696 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11697 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
11698 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
11699 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11700 (mmix_constant_alignment): Make static. Use the same interface
11701 as the target hook.
11702 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
11703 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11704 constant_alignment_word_strings.
11705 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
11706 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11707 constant_alignment_word_strings.
11708 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
11709 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11710 constant_alignment_word_strings.
11711 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
11712 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11713 (rs6000_constant_alignment): New function.
11714 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
11715 * config/riscv/riscv.c (riscv_constant_alignment): New function.
11716 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11717 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
11718 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11719 (rs6000_constant_alignment): New function.
11720 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
11721 * config/s390/s390.c (s390_constant_alignment): New function.
11722 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11723 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
11724 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11725 constant_alignment_word_strings.
11726 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
11727 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11728 (sparc_constant_alignment): New function.
11729 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
11730 * config/spu/spu.c (spu_constant_alignment): New function.
11731 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11732 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
11733 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11734 constant_alignment_word_strings.
11735 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
11736 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11737 constant_alignment_word_strings.
11738 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
11739 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
11740 constant_alignment_word_strings.
11741 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
11742 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11743 (visium_constant_alignment): New function.
11744 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
11745 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
11746 (xtensa_constant_alignment): New function.
11747 * system.h (CONSTANT_ALIGNMENT): Poison.
11749 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
11751 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11752 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
11753 (rs6000_builtin_valid_without_lhs): New helper function.
11754 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11755 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
11757 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
11759 * target.h (vec_perm_indices): Use unsigned short rather than
11761 (auto_vec_perm_indices): Likewise.
11762 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
11763 Use unsigned int rather than unsigned char.
11764 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
11766 2017-09-25 Richard Biener <rguenther@suse.de>
11768 * cfgloop.h (sort_sibling_loops): Declare.
11769 * cfgloop.c (sort_sibling_loops_cmp): New helper.
11770 (sort_sibling_loops): New function sorting the sibling loop list
11772 * graphite.c (graphite_transform_loops): Sort sibling loops.
11774 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
11776 * target.def (vec_perm_const_ok): Change sel parameter to
11778 * optabs-query.c (can_vec_perm_p): Update accordingly.
11779 * doc/tm.texi: Regenerate.
11780 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
11781 auto_vec_perm_indices and remove separate nelt field.
11782 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
11783 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
11784 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
11785 (aarch64_expand_vec_perm_const): Update accordingly.
11786 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
11787 to vec_perm_indices.
11788 * config/arm/arm.c (expand_vec_perm_d): Change perm to
11789 auto_vec_perm_indices and remove separate nelt field.
11790 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
11791 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
11792 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
11794 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
11795 to vec_perm_indices.
11796 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
11797 sel to vec_perm_indices.
11798 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
11799 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
11800 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
11802 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
11805 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
11808 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
11809 on the FUNCTION_DECL function context if it has a DIE that is a
11812 2017-09-25 Richard Biener <rguenther@suse.de>
11814 PR tree-optimization/82285
11815 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
11818 2017-09-25 Tom de Vries <tom@codesourcery.com>
11822 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
11825 2017-09-25 Richard Biener <rguenther@suse.de>
11827 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
11828 ISL errors other than isl_error_quota happen. Dump if the
11829 schedule is the same.
11830 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
11831 errors instead of aborting inside ISL.
11833 2017-09-25 Iain Sandoe <iain@codesourcery.com>
11836 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
11837 of libgcc_eh for m64.
11838 * config/i386/darwin64.h: Likewise.
11840 2017-09-25 Richard Biener <rguenther@suse.de>
11842 PR middle-end/82144
11843 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
11844 attribute for incomplete types nor twice for complete ones.
11846 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
11849 * config/i386/i386.c (ix86_print_operand_address_as): Encode
11850 %esp as %rsp to avoid 0x67 prefix if there is no index or base
11853 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
11856 * config/i386/i386.opt (mprefer-avx256): Use
11857 ix86_target_flags variable.
11858 * config/i386/i386.c (ix86_target_string): Move
11859 -mprefer-avx256 to flag2_opts.
11861 2017-09-22 Jakub Jelinek <jakub@redhat.com>
11863 PR middle-end/35691
11864 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
11865 and x != -1 | y != -1 into (x & y) != -1.
11867 2017-09-22 Steve Ellcey <sellcey@cavium.com>
11869 * config.gcc: Add new case statement to set
11870 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
11871 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
11872 s390x-*-linux* case statements. Added aarch64 to the list of
11873 supported architectures.
11875 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11877 PR tree-optimization/82289
11878 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
11879 STMT_VINFO_RELEVANT_P.
11881 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11882 Alan Hayward <alan.hayward@arm.com>
11883 David Sherwood <david.sherwood@arm.com>
11885 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
11886 for VR_RANGE only; don't allow VR_ANTI_RANGE.
11887 (extract_range_from_binary_expr_1): Don't call
11888 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
11890 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11891 Alan Hayward <alan.hayward@arm.com>
11892 David Sherwood <david.sherwood@arm.com>
11894 * target.def (preferred_vector_alignment): New hook.
11895 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
11897 * doc/tm.texi: Regenerate.
11898 * targhooks.h (default_preferred_vector_alignment): Declare.
11899 * targhooks.c (default_preferred_vector_alignment): New function.
11900 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
11902 (DR_TARGET_ALIGNMENT): New macro.
11903 (aligned_access_p): Update commentary.
11904 (vect_known_alignment_in_bytes): New function.
11905 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
11907 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
11908 Calculate the misalignment based on the target alignment rather than
11910 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
11911 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
11912 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
11913 the target alignment, rather than masking the element misalignment
11914 with the number of elements in a vector. Also use the target
11915 alignment when calculating the maximum number of peels.
11916 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
11917 instead of TYPE_ALIGN_UNIT.
11918 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
11919 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
11920 (vect_create_addr_base_for_vector_ref): Update call accordingly.
11921 (vect_create_data_ref_ptr): Likewise.
11922 (vect_setup_realignment): Realign by ANDing with
11923 -DR_TARGET_MISALIGNMENT.
11924 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
11925 the number of peels based on DR_TARGET_ALIGNMENT.
11926 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
11927 with the guaranteed alignment boundary when deciding whether
11929 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
11930 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
11931 (ensure_base_align): Remove stmt_info parameter. Get the
11932 target base alignment from DR_TARGET_ALIGNMENT.
11933 (vectorizable_store): Update call accordingly. Interpret
11934 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
11936 (vectorizable_load): Likewise.
11938 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
11939 Alan Hayward <alan.hayward@arm.com>
11940 David Sherwood <david.sherwood@arm.com>
11942 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
11943 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
11944 (vect_enhance_data_refs_alignment): Likewise.
11946 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11948 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
11949 error. Only quit immediately if parsing is complete.
11950 (BEGIN): Initialize fatal_err and parse_done.
11951 (begin fpu, end fpu): Check number of arguments.
11952 (begin arch, end arch): Likewise.
11953 (begin cpu, end cpu): Likewise.
11954 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
11955 (optalias): Likewise.
11957 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
11959 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
11960 * config/arm/arm-isa.h: Delete. Move definitions to ...
11961 * arm-cpus.in: ... here. Use new feature and fgroup values.
11962 * config/arm/arm.c (arm_option_override): Use lower case for feature
11964 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
11965 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
11966 * config/arm/parsecpu.awk (END): Add new command 'isa'.
11968 (print_isa_bits_for): New function.
11969 (gen_isa): New function.
11970 (gen_comm_data): Use print_isa_bits_for.
11971 (define feature): New keyword.
11972 (define fgroup): New keyword.
11973 * config/arm/t-arm (TM_H): Remove.
11974 (GTM_H): Add arm-isa.h.
11975 (arm-isa.h): Add rule to generate file.
11976 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
11977 case for feature bit names.
11979 2017-09-22 Richard Biener <rguenther@suse.de>
11981 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
11983 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
11984 print of no dependency loops ...
11985 * graphite.c (graphite_transform_loops): ... here.
11986 (canonicalize_loop_closed_ssa_form): Work from inner to outer
11988 (same_close_phi_node, remove_duplicate_close_phi,
11989 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
11990 (canonicalize_loop_closed_ssa): ... here and simplify.
11991 * graphite-optimize-isl.c: Include tree-vectorizer.h.
11992 (optimize_isl): Use dump_printf_loc to tell when we stopped
11993 optimizing because of an ISL timeout.
11995 2017-09-22 Richard Biener <rguenther@suse.de>
11997 PR tree-optimization/82291
11998 * tree-if-conv.c (predicate_mem_writes): Make sure to
11999 remove writes in blocks predicated with false.
12001 2017-09-22 Richard Biener <rguenther@suse.de>
12003 * sese.c: Include cfganal.h.
12004 (if_region_set_false_region): Remove.
12005 (create_if_region_on_edge): Likewise.
12006 (move_sese_in_condition): Re-implement without destroying
12009 2017-09-22 Richard Biener <rguenther@suse.de>
12011 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
12012 Verify both BBs contain loop PHI nodes before dispatching to
12013 copy_loop_phi_args.
12014 (graphite_regenerate_ast_isl): Do not recompute dominators,
12015 do not verify three times. Restructure for clarity.
12016 * graphite-scop-detection.c (same_close_phi_node,
12017 remove_duplicate_close_phi, make_close_phi_nodes_unique,
12018 defined_in_loop_p, canonicalize_loop_closed_ssa,
12019 canonicalize_loop_closed_ssa_form): Simplify, remove excess
12020 checking and SSA rewrite, move to ...
12021 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
12022 (graphite_initialize): Do not pass in ctx, do not reset the
12023 SCEV cache, compute only dominators.
12024 (graphite_transform_loops): Allocate ISL ctx after
12025 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
12026 Maintain post-dominators only around build_scops.
12027 * sese.c (if_region_set_false_region): Make static. Free
12028 and recompute dominators.
12029 (move_sese_in_condition): Assert we don't get called with
12030 post-dominators computed.
12031 * sese.h (if_region_set_false_region): Remove.
12033 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
12035 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
12036 mode attribute for TARGET_AVX512VL.
12038 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
12040 * config/i386/i386.opt (mprefer-avx256): New option.
12041 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
12043 (ix86_preferred_simd_mode): Return 256-bit AVX modes
12044 for TARGET_PREFER_AVX256.
12045 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
12047 2017-09-21 Jeff Law <law@redhat.com>
12049 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
12050 Fix dump output if the only stack space is for pushed registers.
12052 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12054 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
12057 2017-09-21 Martin Sebor <msebor@redhat.com>
12060 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
12061 code (in C++) or code that triggers warnings.
12063 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
12065 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
12067 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
12069 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
12070 * sched-rgn.c: Ditto.
12071 * sel-sched-ir.c: Ditto.
12073 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
12075 * toplev.h (set_random_seed): Adjust return type.
12076 * toplev.c (init_local_tick): Move eager initialization of random_seed
12077 to get_random_seed. Adjust comment.
12078 (init_random_seed): Inline to get_random_seed, delete.
12079 (get_random_seed): Initialize random_seed lazily.
12080 (set_random_seed): Do not return previous value.
12081 (print_switch_value): Do not call get_random_seed.
12083 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
12085 * cgraph.c (delete_function_version): New, broken out from...
12086 (cgraph_node::delete_function_version): ...here. Rename to
12087 cgraph_node::delete_function_version_by_decl. Update all uses.
12088 (cgraph_node::remove): Call delete_function_version.
12090 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12093 * tree-inline.c (expand_call_inline): Emit clobber stmts for
12094 VAR_DECLs to which addressable non-volatile parameters are mapped
12095 and for id->retvar after the return value assignment. Clear
12096 id->retval and id->retbnd after inlining.
12098 2017-09-21 Richard Biener <rguenther@suse.de>
12100 PR tree-optimization/82276
12101 PR tree-optimization/82244
12102 * tree-vrp.c (build_assert_expr_for): Set
12103 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
12105 (remove_range_assertions): Revert earlier change.
12107 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
12110 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
12112 2017-09-21 Richard Biener <rguenther@suse.de>
12114 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12115 Restore valid IL after code generation errors.
12116 * graphite.c (graphite_transform_loops): Diagnose code
12117 generation issues as MSG_MISSED_OPTIMIZATION and continue
12118 with processing SCOPs.
12120 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12121 Alan Hayward <alan.hayward@arm.com>
12122 David Sherwood <david.sherwood@arm.com>
12124 * calls.c (compute_argument_addresses): Use simplify_gen_binary
12125 rather than choosing between plus_constant and gen_rtx_<CODE>.
12126 * expr.c (emit_push_insn): Likewise.
12127 (expand_expr_real_2): Likewise.
12129 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12130 Alan Hayward <alan.hayward@arm.com>
12131 David Sherwood <david.sherwood@arm.com>
12133 * loop-unroll.c (split_iv): Call copy_rtx on the step.
12135 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12136 Alan Hayward <alan.hayward@arm.com>
12137 David Sherwood <david.sherwood@arm.com>
12139 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
12140 calling tree_to_uhwi.
12142 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12143 Alan Hayward <alan.hayward@arm.com>
12144 David Sherwood <david.sherwood@arm.com>
12146 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
12147 INTEGER_CST rather than a negative test for ADDR_EXPR.
12149 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
12150 Alan Hayward <alan.hayward@arm.com>
12151 David Sherwood <david.sherwood@arm.com>
12153 * tree-vrp.c (extract_range_from_binary_expr_1): Check
12154 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
12156 2017-09-21 Richard Biener <rguenther@suse.de>
12158 PR tree-optimization/71351
12159 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
12160 graphite_create_new_loop_guard): Remove, fold remaining parts
12162 (translate_isl_ast_node_for): ... here and simplify.
12164 2017-09-21 Jakub Jelinek <jakub@redhat.com>
12167 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
12168 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
12169 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
12170 alternative always use QI mode, for -Os imov (=R,R) alternative
12171 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
12174 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12175 Jeff Law <law@redhat.com>
12177 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
12178 (allocate_stack_space): New function, partially extracted from
12179 s390_emit_prologue.
12180 (s390_emit_prologue): Track offset to most recent stack probe.
12181 Code to allocate space moved into allocate_stack_space.
12182 Dump actions when no stack is allocated.
12183 (s390_prologue_plus_offset): New function.
12184 (s390_emit_stack_probe): Likewise.
12186 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
12188 * common.opt (Wa, Wl, Wp, g, gz=): Add
12190 (gno-column-info): Remove.
12191 (gcolumn-info): Drop RejectNegative.
12192 (gno-): New prefix.
12193 (gno-record-gcc-switches): Remove.
12194 (grecord-gcc-switches): Drop RejectNegative.
12195 (gno-split-dwarf): Remove.
12196 (gsplit-dwarf): Drop RejectNegative.
12197 (gno-strict-dwarf): Remove.
12198 (gstrict-dwarf): Drop RejectNegative.
12199 * config/darwin.opt (gfull, gused): Add RejectNegative.
12200 * dwarf2out.c (gen_producer_string): Drop
12201 gno-record-gcc-switches handler.
12202 * optc-gen.awk: Add g to prefixes with negative forms.
12203 * opts-common.c (remapping_prefix_p): New.
12204 (find_opt): Check it.
12205 (generate_canonical_option): Test g prefix.
12206 (option_map): Add -gno- mapping.
12207 (add_misspelling_candidates): Check remapping_prefix_p.
12209 2017-09-20 Jeff Law <law@redhat.com>
12211 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
12212 thinko in stack clash protection support.
12214 * explow.c (compute_stack_clash_protection_loop_data): Use
12215 CONST_INT_P instead of explicit test. Verify object is a
12216 CONST_INT_P before looking at INTVAL.
12217 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
12218 instead of explicit test.
12220 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
12223 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
12224 address instead of to r1 and r11.
12226 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
12228 * config.gcc: Support "knm".
12229 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
12230 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12232 * config/i386/i386.c (m_KNM): Define.
12233 (processor_target_table): Add "knm".
12235 (ix86_option_override_internal): Add "knm".
12236 (ix86_issue_rate): Add PROCESSOR_KNM.
12237 (ix86_adjust_cost): Ditto.
12238 (ia32_multipass_dfa_lookahead): Ditto.
12239 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
12240 (fold_builtin_cpu): Add M_INTEL_KNM.
12241 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
12242 (processor_type): Add PROCESSOR_KNM.
12243 * config/i386/x86-tune.def: Add m_KNM.
12244 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
12246 2017-09-20 Richard Biener <rguenther@suse.de>
12248 PR tree-optimization/80213
12249 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
12250 are allowed in empty BBs as well.
12251 (canonicalize_loop_closed_ssa): Also look for other complex
12253 (scop_detection::get_sese): Include the loop-closed PHI block
12255 (scop_detection::merge_sese): Remove code adding extra blocks.
12256 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
12257 (build_scops): Assert the final returned scop is invalid.
12259 2017-09-20 Richard Biener <rguenther@suse.de>
12261 PR tree-optimization/82264
12262 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
12264 (vn_phi_lookup): Likewise.
12265 (vn_phi_insert): Likewise.
12267 2017-09-20 Jakub Jelinek <jakub@redhat.com>
12269 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
12270 that fits into uhwi or shwi, add DW_AT_const_value regardless
12271 of early_dwarf without going through RTL, using add_AT_unsigned
12274 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
12275 (DEBUG_LTO_ABBREV_SECTION): Likewise.
12276 (DEBUG_LTO_MACINFO_SECTION): Likewise.
12277 (DEBUG_MACRO_SECTION): Likewise.
12278 (DEBUG_LTO_MACRO_SECTION): Likewise.
12279 (DEBUG_STR_DWO_SECTION): Likewise.
12280 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
12281 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
12282 (DEBUG_LTO_DWO_LINE_SECTION): Define.
12283 (DEBUG_LTO_LINE_STR_SECTION): Define.
12284 (init_sections_and_labels): Initialize debug_line_str_section
12285 variable. Initialize debug_loc_section for -gdwarf-5 to
12286 DEBUG_LOCLISTS_SECTION. Formatting fixes.
12288 2017-09-20 Richard Biener <rguenther@suse.de>
12290 * graphite-sese-to-poly.c (extract_affine): Properly handle
12291 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
12293 2017-09-20 Richard Biener <rguenther@suse.de>
12295 PR tree-optimization/81373
12296 * graphite-scop-detection.c (build_cross_bb_scalars_def):
12297 Force SESE live-out defs to be handled even if they are
12300 2017-09-19 Jeff Law <law@redhat.com>
12302 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
12303 nothing for stack adjustments with REG_STACK_CHECK.
12304 * sched-deps.c (parse_add_or_inc): Reject insns with
12305 REG_STACK_CHECK from dependency breaking.
12306 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
12307 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
12308 * reg-notes.def (STACK_CHECK): New note.
12310 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
12311 (ix86_expand_prologue): Dump stack clash info as needed.
12312 Call ix86_adjust_stack_and_probe_stack_clash as needed.
12314 * function.c (dump_stack_clash_frame_info): New function.
12315 * function.h (dump_stack_clash_frame_info): Prototype.
12316 (enum stack_clash_probes): New enum.
12318 * config/alpha/alpha.c (alpha_expand_prologue): Also check
12319 flag_stack_clash_protection.
12320 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
12321 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
12322 (arm_frame_pointer_required): Likewise.
12323 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
12324 (ia64_expand_prologue): Likewise.
12325 * config/mips/mips.c (mips_expand_prologue): Likewise.
12326 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
12327 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12328 (sparc_flat_expand_prologue): Likewise.
12329 * config/spu/spu.c (spu_expand_prologue): Likewise.
12331 * explow.c: Include "params.h".
12332 (anti_adjust_stack_and_probe_stack_clash): New function.
12333 (get_stack_check_protect): Likewise.
12334 (compute_stack_clash_protection_loop_data): Likewise.
12335 (emit_stack_clash_protection_loop_start): Likewise.
12336 (emit_stack_clash_protection_loop_end): Likewise.
12337 (allocate_dynamic_stack_space): Use get_stack_check_protect.
12338 Use anti_adjust_stack_and_probe_stack_clash.
12339 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
12340 (emit_stack_clash_protection_loop_start): Likewise.
12341 (emit_stack_clash_protection_loop_end): Likewise.
12342 * rtl.h (get_stack_check_protect): Prototype.
12343 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
12344 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
12345 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
12347 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
12349 * doc/tm.texi: Rebuilt.
12350 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
12351 get_stack_check_protect.
12352 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
12353 * config/arm/arm.c (arm_expand_prologue): Likewise.
12354 (arm_frame_pointer_required): Likewise.
12355 * config/i386/i386.c (ix86_expand_prologue): Likewise.
12356 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
12357 * config/mips/mips.c (mips_expand_prologue): Likewise.
12358 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
12359 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
12360 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
12361 (sparc_flat_expand_prologue): Likewise.
12363 * common.opt (-fstack-clash-protection): New option.
12364 * flag-types.h (enum stack_check_type): Note difference between
12365 -fstack-check= and -fstack-clash-protection.
12366 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
12367 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
12368 * toplev.c (process_options): Issue warnings/errors for cases
12369 not handled with -fstack-clash-protection.
12370 * doc/invoke.texi (-fstack-clash-protection): Document new option.
12371 (-fstack-check): Note additional problem with -fstack-check=generic.
12372 Note that -fstack-check is primarily for Ada and refer users
12373 to -fstack-clash-protection for stack-clash-protection.
12374 Document new params for stack clash protection.
12376 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
12378 * config/i386/i386.c (ix86_split_long_move): Do not handle
12379 address used for LEA in a special way.
12381 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
12383 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
12385 2017-09-19 Martin Sebor <msebor@redhat.com>
12388 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
12389 of incompatible types.
12391 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
12393 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12394 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
12395 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12396 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
12398 2017-09-19 Richard Biener <rguenther@suse.de>
12400 PR tree-optimization/82244
12401 * tree-vrp.c (remove_range_assertions): Do not propagate
12402 a constant to abnormals but replace the assert with a copy.
12404 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
12406 PR rtl-optimization/57878
12407 PR rtl-optimization/68988
12408 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
12409 avoidance test involving non_reload_pseudos. Move frequency test
12410 below the general fragmentation avoidance test.
12412 2017-09-19 Richard Biener <rguenther@suse.de>
12414 PR tree-optimization/69728
12415 * graphite-sese-to-poly.c (schedule_error): New global.
12416 (add_loop_schedule): Handle empty domain by failing the
12418 (build_original_schedule): Handle schedule_error.
12420 2017-09-19 Richard Biener <rguenther@suse.de>
12422 * graphite-scop-detection.c (scop_detection::can_represent_loop):
12423 Do not iterate to sibling loops but only to siblings of inner
12426 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
12429 * config/m68k/m68k.md (moveq feeding equality comparison): Check
12430 that the registers are different.
12432 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
12434 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
12435 to processor_model and "amdfam17h" to arch_names_table.
12436 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
12438 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12441 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
12443 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12444 Alan Hayward <alan.hayward@arm.com>
12445 David Sherwood <david.sherwood@arm.com>
12447 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
12449 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
12451 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
12452 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
12453 vect_schedule_slp_instance.
12454 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
12455 Update call to vect_slp_analyze_node_operations. Simplify return
12457 (vect_slp_analyze_bb_1): Update call accordingly.
12458 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
12459 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
12460 (vect_schedule_slp): Update call accordingly.
12462 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12463 Alan Hayward <alan.hayward@arm.com>
12464 David Sherwood <david.sherwood@arm.com>
12466 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
12467 with types that aren't in fact scalar.
12469 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12471 * tree-vect-slp.c (vect_record_max_nunits): New function,
12473 (vect_build_slp_tree_1): ...here.
12474 (vect_build_slp_tree_2): Call it for phis too.
12476 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12478 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
12479 to vect_get_vec_def_for_operand when getting the mask operand.
12481 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12482 Alan Hayward <alan.hayward@arm.com>
12483 David Sherwood <david.sherwood@arm.com>
12485 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
12488 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12489 Alan Hayward <alan.hayward@arm.com>
12490 David Sherwood <david.sherwood@arm.com>
12492 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
12493 calculation for vector booleans.
12495 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
12496 Alan Hayward <alan.hayward@arm.com>
12497 David Sherwood <david.sherwood@arm.com>
12499 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
12501 (vect_transform_stmt): ...here.
12502 (vect_analyze_stmt): Use it instead of calling
12503 vectorizable_live_operation directly.
12505 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
12507 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
12508 non-SIMT targets in acc vector loops.
12510 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
12512 * configure.ac: Add arc and check if assembler supports gdwarf2.
12513 * configure: Regenerate.
12515 2017-09-18 Richard Biener <rguenther@suse.de>
12517 PR tree-optimization/82220
12518 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
12519 epilogue niters from the min_profitable_iters compute.
12521 2017-09-18 Jakub Jelinek <jakub@redhat.com>
12524 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
12525 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
12526 (ix86_init_pic_reg): Revert 2017-09-01 changes.
12528 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
12531 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
12532 switching to a new text section.
12534 2017-09-18 Richard Biener <rguenther@suse.de>
12536 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
12538 (build_alias_set): Reject aliases with no access function.
12540 2017-09-18 Richard Biener <rguenther@suse.de>
12542 PR tree-optimization/79622
12543 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
12545 (build_cross_bb_scalars_use): Likewise.
12547 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
12549 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
12551 2017-09-18 Alan Modra <amodra@gmail.com>
12554 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
12555 stack_pointer_rtx for count 0. Update comments. Break up
12556 large rtl expression.
12558 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
12560 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
12561 Increase to 20 bytes.
12562 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
12563 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
12564 or avx version of the stub.
12566 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
12569 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
12570 compute the minimum stack alignment. Also update preferred stack
12571 boundary for leaf functions.
12573 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
12575 PR tree-optimization/82228
12576 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
12579 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12581 * common/config/nds32/nds32-common.c
12582 (nds32_option_optimization_table): Refine formatting.
12583 (nds32_option_optimization_table): Use -fsched-pressure and
12584 -fomit-frame-pointer for specific optimization level.
12586 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
12588 * config/nds32/nds32.c: Refine formatting and comments.
12589 * config/nds32/nds32.h: Likewise.
12590 * config/nds32/nds32.md: Likewise.
12591 * config/nds32/nds32-cost.c: Likewise.
12592 * config/nds32/nds32-isr.c: Likewise.
12593 * config/nds32/nds32-md-auxiliary.c: Likewise.
12594 * config/nds32/nds32-multiple.md: Likewise.
12595 * config/nds32/nds32-predicates.c: Likewise.
12597 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
12598 Jakub Jelinek <jakub@redhat.com>
12600 Add support for -std=c++2a.
12601 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
12603 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
12605 2017-09-15 Steve Ellcey <sellcey@cavium.com>
12608 * doc/extend.texi (Common Function Attributes): Add
12609 references to ARM, AArch64, and S/390 specific attributes.
12610 (Function Specific Option Pragmas): Add AArch64 and S/390
12611 to list of back ends that support the target pragma.
12613 2017-09-15 Nathan Sidwell <nathan@acm.org>
12615 * doc/standards.texi: Fix C++17 description. Update URLs for
12618 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
12620 * common.opt (Wcast-align=strict): New warning option.
12621 * doc/invoke.texi: Document -Wcast-align=strict.
12623 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
12625 * cgraph.h (cgraph_thunk_info): Add comments.
12626 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
12627 assert for VIRTUAL_* arguments stricter.
12629 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
12631 PR tree-optimization/71026
12632 * match.pd: Move RDIV patterns from fold-const.c
12633 * fold-const.c (distribute_real_division): Removed.
12634 (fold_binary_loc): Remove calls to distribute_real_divison.
12636 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12638 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
12639 c++1z and gnu++1z as deprecated. Change other references to
12640 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
12641 Change -Wc++1z-compat to -Wc++17-compat.
12642 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
12643 * dwarf2out.c (highest_c_language): Handle C++17.
12644 (gen_compile_unit_die): Likewise.
12646 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12648 PR rtl-optimization/82192
12649 * combine.c (make_extraction): Don't look through non-paradoxical
12650 SUBREGs or TRUNCATE if pos + len is or might be bigger than
12653 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12654 Alan Hayward <alan.hayward@arm.com>
12655 David Sherwood <david.sherwood@arm.com>
12657 * target.def (function_arg_offset): New hook.
12658 * targhooks.h (default_function_arg_offset): Declare.
12659 * targhooks.c (default_function_arg_offset): New function.
12660 * function.c (locate_and_pad_parm): Use
12661 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
12662 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
12663 (TARGET_FUNCTION_ARG_OFFSET): ...this.
12664 * doc/tm.texi: Regenerate.
12665 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
12666 * config/spu/spu.c (spu_function_arg_offset): New function.
12667 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
12668 * system.h (FUNCTION_ARG_OFFSET): Poison.
12670 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12671 Alan Hayard <alan.hayward@arm.com>
12672 David Sherwood <david.sherwood@arm.com>
12674 * target.def (truly_noop_truncation): New hook.
12675 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
12676 than TRULY_NOOP_TRUNCATION.
12677 * hooks.h (hook_bool_uint_uint_true): Declare.
12678 * hooks.c (hook_bool_uint_uint_true): New function.
12679 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
12680 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
12681 * doc/tm.texi: Regenerate.
12682 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
12683 rather than TRULY_NOOP_TRUNCATION in comments.
12684 (simplify_comparison): Likewise.
12685 (record_truncated_value): Likewise.
12686 * expmed.c (extract_bit_field_1): Likewise.
12687 (extract_split_bit_field): Likewise.
12688 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
12689 instead of TRULY_NOOP_TRUNCATION.
12690 * function.c (assign_parm_setup_block): Likewise.
12691 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
12692 * rtlhooks.c: Include target.h.
12693 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
12694 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
12695 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
12696 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
12697 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
12698 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
12699 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
12700 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
12701 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
12702 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
12703 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
12704 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
12705 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
12706 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
12707 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
12708 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
12709 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
12710 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
12711 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
12712 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
12713 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
12714 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
12715 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
12716 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
12717 * config/mips/mips.c (mips_truly_noop_truncation): New function.
12718 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12719 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12720 rather than TRULY_NOOP_TRUNCATION in comments.
12721 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
12722 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
12723 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
12724 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
12725 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
12726 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
12727 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
12728 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
12729 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
12730 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
12731 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
12732 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12733 rather than TRULY_NOOP_TRUNCATION in comments.
12734 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
12735 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
12736 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
12737 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
12738 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
12739 TRULY_NOOP_TRUNCATION condition.
12740 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
12741 (TRULY_NOOP_TRUNCATION): Delete.
12742 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
12743 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
12744 * config/spu/spu.c (spu_truly_noop_truncation): New function.
12745 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12746 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
12747 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
12748 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
12749 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
12750 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
12751 rather than TRULY_NOOP_TRUNCATION in comments.
12752 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
12753 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
12754 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
12755 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
12756 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
12757 * system.h (TRULY_NOOP_TRUNCATION): Poison.
12759 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
12762 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
12763 (*cmp_ior): Likewise.
12764 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
12765 (*ior_scc_scc_cmp): Likewise.
12766 (*and_scc_scc): Likewise.
12767 (*and_scc_scc_cmp): Likewise.
12769 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
12770 Alan Hayard <alan.hayward@arm.com>
12771 David Sherwood <david.sherwood@arm.com>
12773 * target.def (can_change_mode_class): New hook.
12774 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
12775 (hard_regno_nregs): Likewise.
12776 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
12777 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
12778 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
12779 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
12780 (LOAD_EXTEND_OP): Update accordingly.
12781 * doc/tm.texi: Regenerate.
12782 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
12783 CANNOT_CHANGE_MODE_CLASS.
12784 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
12785 (REG_CAN_CHANGE_MODE_P): ...this new macro.
12786 * combine.c (simplify_set): Update accordingly.
12787 * emit-rtl.c (validate_subreg): Likewise.
12788 * recog.c (general_operand): Likewise.
12789 * regcprop.c (mode_change_ok): Likewise.
12790 * reload1.c (choose_reload_regs): Likewise.
12791 (inherit_piecemeal_p): Likewise.
12792 * rtlanal.c (simplify_subreg_regno): Likewise.
12793 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
12794 instead of CANNOT_CHANGE_MODE_CLASS.
12795 (reload_cse_simplify_operands): Likewise.
12796 * reload.c (push_reload): Use targetm.can_change_mode_class
12797 instead of CANNOT_CHANGE_MODE_CLASS.
12798 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
12799 REG_CANNOT_CHANGE_MODE_P.
12800 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12801 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
12802 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12803 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12804 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12805 (arm_can_change_mode_class): New function.
12806 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12807 than CANNOT_CHANGE_MODE_CLASS in comments.
12808 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12809 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
12810 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
12811 (ix86_can_change_mode_class): ...this new function, inverting the
12812 sense of the return value.
12813 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12814 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12815 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12816 (ia64_can_change_mode_class): New function.
12817 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12818 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
12819 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
12820 (m32c_can_change_mode_class): ...this new function, inverting the
12821 sense of the return value.
12822 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12823 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12824 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
12825 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
12826 (mips_can_change_mode_class): ...this new function, inverting the
12827 sense of the return value.
12828 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12829 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12830 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12831 (msp430_can_change_mode_class): New function.
12832 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12833 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
12834 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12835 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12836 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12837 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
12838 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12839 (pa_cannot_change_mode_class): Replace with...
12840 (pa_can_change_mode_class): ...this new function, inverting the
12841 sense of the return value.
12842 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
12843 than CANNOT_CHANGE_MODE_CLASS in comments.
12844 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12845 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
12846 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12847 (pdp11_cannot_change_mode_class): Replace with...
12848 (pdp11_can_change_mode_class): ...this new function, inverting the
12849 sense of the return value.
12850 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12851 * config/powerpcspe/powerpcspe-protos.h
12852 (rs6000_cannot_change_mode_class_ptr): Delete.
12853 * config/powerpcspe/powerpcspe.c
12854 (rs6000_cannot_change_mode_class_ptr): Delete.
12855 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12856 (rs6000_option_override_internal): Assign to
12857 targetm.can_change_mode_class instead of
12858 rs6000_cannot_change_mode_class_ptr.
12859 (rs6000_cannot_change_mode_class): Replace with...
12860 (rs6000_can_change_mode_class): ...this new function, inverting the
12861 sense of the return value.
12862 (rs6000_debug_cannot_change_mode_class): Replace with...
12863 (rs6000_debug_can_change_mode_class): ...this new function.
12864 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12865 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
12866 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12867 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12868 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
12870 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
12871 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12872 (rs6000_option_override_internal): Assign to
12873 targetm.can_change_mode_class instead of
12874 rs6000_cannot_change_mode_class_ptr.
12875 (rs6000_cannot_change_mode_class): Replace with...
12876 (rs6000_can_change_mode_class): ...this new function, inverting the
12877 sense of the return value.
12878 (rs6000_debug_cannot_change_mode_class): Replace with...
12879 (rs6000_debug_can_change_mode_class): ...this new function.
12880 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12881 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
12882 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
12883 (s390_can_change_mode_class): ...this new function, inverting the
12884 sense of the return value.
12885 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12886 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12887 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
12888 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12889 (sh_cannot_change_mode_class): Replace with...
12890 (sh_can_change_mode_class): ...this new function, inverting the
12891 sense of the return value.
12892 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12893 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12894 (sparc_can_change_mode_class): New function.
12895 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12896 * config/spu/spu.c (spu_can_change_mode_class): New function.
12897 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12898 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
12899 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
12900 (visium_can_change_mode_class): New function.
12901 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
12903 2017-09-15 Richard Biener <rguenther@suse.de>
12905 PR tree-optimization/82217
12906 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
12907 but not undefined case.
12909 2017-09-15 Jakub Jelinek <jakub@redhat.com>
12912 * postreload.c (reload_cse_simplify_operands): Skip
12913 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
12915 2017-09-15 Richard Biener <rguenther@suse.de>
12917 PR tree-optimization/68823
12918 * graphite-scop-detection.c (build_alias_set): If we have a
12919 possible dependence check whether we can handle them by just
12920 looking at the DRs DR_ACCESS_FNs.
12921 (build_scops): If build_alias_set fails, fail the SCOP.
12923 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
12925 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
12926 to support float128 built-in functions that require the ISA 3.0
12928 (BU_FLOAT128_3_HW): Likewise.
12929 (SQRTF128): Add support for the IEEE 128-bit square root and fma
12930 built-in functions.
12931 (FMAF128): Likewise.
12933 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
12934 support for built-in functions that need the ISA 3.0 IEEE 128-bit
12935 floating point instructions.
12936 (rs6000_invalid_builtin): Likewise.
12937 (rs6000_builtin_mask_names): Likewise.
12938 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
12939 (RS6000_BTM_FLOAT128_HW): Likewise.
12940 (RS6000_BTM_COMMON): Likewise.
12941 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
12943 * doc/extend.texi (RS/6000 built-in functions): Document the
12944 IEEE 128-bit floating point square root and fused multiply-add
12945 built-in functions.
12947 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
12949 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
12950 reg (r2) isn't in the set of registers defined in the prologue.
12952 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12953 Alan Hayward <alan.hayward@arm.com>
12954 David Sherwood <david.sherwood@arm.com>
12956 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
12957 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
12958 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
12959 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
12960 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
12962 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
12963 max_vectorization_factor.
12964 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
12966 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12967 Alan Hayward <alan.hayward@arm.com>
12968 David Sherwood <david.sherwood@arm.com>
12970 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
12971 (vect_worthwhile_without_simd_p): Declare.
12972 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
12973 (vectorizable_reduction): Use it.
12974 * tree-vect-stmts.c (vectorizable_shift): Likewise.
12975 (vectorizable_operation): Likewise.
12977 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
12978 Alan Hayward <alan.hayward@arm.com>
12979 David Sherwood <david.sherwood@arm.com>
12981 * tree-vectorizer.h (vect_get_num_copies): New function.
12982 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
12983 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12984 (vectorizable_induction): Likewise.
12985 (vectorizable_live_operation): Likewise.
12986 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12987 (vectorizable_bswap): Likewise.
12988 (vectorizable_call): Likewise.
12989 (vectorizable_conversion): Likewise.
12990 (vectorizable_assignment): Likewise.
12991 (vectorizable_shift): Likewise.
12992 (vectorizable_operation): Likewise.
12993 (vectorizable_store): Likewise.
12994 (vectorizable_load): Likewise.
12995 (vectorizable_condition): Likewise.
12996 (vectorizable_comparison): Likewise.
12997 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
12999 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13000 Alan Hayward <alan.hayward@arm.com>
13001 David Sherwood <david.sherwood@arm.com>
13003 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
13004 of vect_init_vector.
13006 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13007 Alan Hayward <alan.hayward@arm.com>
13008 David Sherwood <david.sherwood@arm.com>
13010 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
13011 an inline wrapper that provides a location.
13012 (gimple_build_vector): Likewise.
13013 * gimple-fold.c (gimple_build_vector_from_val): New function.
13014 (gimple_build_vector): Likewise.
13015 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
13016 functions to build the initial value. Always return a gimple value.
13017 (get_initial_defs_for_reduction): Likewise. Only compute
13019 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
13020 vect_init_vector on the results from get_initial_def(s)_for_reduction.
13021 (vectorizable_induction): Use gimple_build_vector rather than
13024 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13025 Alan Hayward <alan.hayward@arm.com>
13026 David Sherwood <david.sherwood@arm.com>
13028 * target.h (vec_perm_indices): New typedef.
13029 (auto_vec_perm_indices): Likewise.
13030 * optabs-query.h: Include target.h
13031 (can_vec_perm_p): Take a vec_perm_indices *.
13032 * optabs-query.c (can_vec_perm_p): Likewise.
13033 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
13034 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13035 * tree-vect-generic.c (lower_vec_perm): Likewise.
13036 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
13037 (vect_grouped_load_supported): Likewise.
13038 (vect_shift_permute_load_chain): Likewise.
13039 (vect_permute_store_chain): Use auto_vec_perm_indices.
13040 (vect_permute_load_chain): Likewise.
13041 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
13042 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
13043 Update uses of can_vec_perm_p.
13044 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
13045 mode with a number of elements. Take a vec_perm_indices *.
13046 (vect_create_epilog_for_reduction): Update accordingly.
13047 Use auto_vec_perm_indices.
13048 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
13049 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
13050 (vect_transform_slp_perm_load): Likewise.
13051 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
13052 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
13053 (vect_gen_perm_mask_checked): Likewise.
13054 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
13055 (vect_gen_perm_mask_checked): Likewise.
13056 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
13057 (vectorizable_store): Likewise.
13058 (vectorizable_load): Likewise.
13059 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
13060 (vectorizable_bswap): Likewise.
13062 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13063 Alan Hayward <alan.hayward@arm.com>
13064 David Sherwood <david.sherwood@arm.com>
13066 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
13067 * tree.c (build_vector): Likewise.
13068 (build_vector_from_ctor): Update accordingly.
13069 (build_vector_from_val): Likewise.
13070 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
13071 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
13072 * tree-vect-generic.c (add_rshift): Likewise.
13073 (expand_vector_divmod): Likewise.
13074 (optimize_vector_constructor): Likewise.
13075 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
13076 (vect_transform_slp_perm_load): Likewise.
13077 (vect_schedule_slp_instance): Likewise.
13078 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
13079 (vectorizable_call): Likewise.
13080 (vect_gen_perm_mask_any): Likewise. Add elements in order.
13081 * expmed.c (make_tree): Likewise.
13082 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
13083 a vector passed to build_vector.
13084 (fold_convert_const): Likewise.
13085 (exact_inverse): Likewise.
13086 (fold_ternary_loc): Likewise.
13087 (fold_relational_const): Likewise.
13088 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
13089 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
13090 (const_unop): Likewise. Store the reduction accumulator in a
13091 variable rather than an array.
13092 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
13093 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
13094 the new vector, rather than constructing it after the input arrays.
13095 (native_interpret_vector): Use auto_vec<tree> when building
13096 a vector passed to build_vector. Add elements in order.
13097 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
13098 auto_vec<tree> when building a vector passed to build_vector.
13099 (vect_create_epilog_for_reduction): Likewise.
13100 (vectorizable_induction): Likewise.
13101 (get_initial_def_for_reduction): Likewise. Fix indentation of
13103 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
13105 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
13106 passed to build_vector.
13108 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
13109 Alan Hayward <alan.hayward@arm.com>
13110 David Sherwood <david.sherwood@arm.com>
13112 * tree-core.h (tree_base::u): Add an "nelts" field.
13113 (tree_vector): Use VECTOR_CST_NELTS as the length.
13114 * tree.c (tree_size): Likewise.
13115 (make_vector): Initialize VECTOR_CST_NELTS.
13116 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
13117 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
13118 TYPE_VECTOR_SUBPARTS.
13119 * expr.c (const_vector_mask_from_tree): Consistently use "units"
13120 as the number of units, setting it from VECTOR_CST_NELTS.
13121 (const_vector_from_tree): Likewise.
13122 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
13123 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
13124 (fold_negate_expr_1): Likewise.
13125 (fold_convert_const): Likewise.
13126 (const_binop): Likewise. Differentiate the number of output and
13128 (const_unop): Likewise.
13129 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
13130 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
13131 in cases that did the opposite.
13133 2017-09-14 Richard Biener <rguenther@suse.de>
13135 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
13138 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
13140 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
13142 2017-09-14 Jakub Jelinek <jakub@redhat.com>
13145 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
13146 if and where to split a bb, except for splitting before debug insn
13147 sequences followed by non-label real insn. Delete debug insns
13148 in between basic blocks.
13150 * combine.c (make_compound_operation_int): Formatting fixes.
13152 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
13153 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13154 * config/netbsd.h (LINK_EH_SPEC): Likewise.
13155 * config/sol2.h (LINK_EH_SPEC): Likewise.
13156 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13157 * config/s390/linux.h (LINK_SPEC): Likewise.
13158 * config/freebsd.h (LINK_EH_SPEC): Likewise.
13159 * config/openbsd.h (LINK_EH_SPEC): Likewise.
13160 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13161 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
13162 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
13163 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13164 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
13165 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
13167 2017-09-13 Jakub Jelinek <jakub@redhat.com>
13169 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
13171 (ENDFILE_LINUX_SPEC): Likewise.
13172 (LINK_EH_SPEC): Likewise.
13173 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
13174 (LINK_OS_LINUX_SPEC32): Likewise.
13175 (LINK_OS_LINUX_SPEC64): Likewise.
13176 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
13177 (LINK_OS_LINUX_SPEC): Likewise.
13179 2017-09-13 Martin Liska <mliska@suse.cz>
13181 PR middle-end/82154
13182 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
13183 CASE_HIGH is NULL_TREE.
13185 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13186 Alan Hayward <alan.hayward@arm.com>
13187 David Sherwood <david.sherwood@arm.com>
13189 * target.def (secondary_memory_needed): New hook.
13190 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
13191 instead of SECONDARY_MEMORY_NEEDED.
13192 (secondary_memory_needed_mode): Likewise.
13193 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
13194 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
13195 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
13196 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
13197 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
13198 * doc/tm.texi: Regenerate.
13199 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
13200 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
13201 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13202 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
13203 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
13204 * config/i386/i386.c (inline_secondary_memory_needed): Put the
13205 mode argument first and change the reg_class arguments to reg_class_t.
13206 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
13207 Make static. Update the call to inline_secondary_memory_needed.
13208 (ix86_register_move_cost): Update the call to
13209 inline_secondary_memory_needed.
13210 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13211 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
13213 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
13214 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13216 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
13217 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
13218 * config/mips/mips.c (mips_secondary_memory_needed): Make static
13219 and match hook interface. Add comment from mips.h.
13220 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13221 * config/mmix/mmix.md (truncdfsf2): Refer to
13222 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13224 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
13225 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
13226 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13227 (pa_secondary_memory_needed): New function.
13228 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
13229 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
13230 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13231 (pdp11_secondary_memory_needed): Make static and match hook interface.
13232 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
13233 * config/powerpcspe/powerpcspe-protos.h
13234 (rs6000_secondary_memory_needed_ptr): Delete.
13235 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
13237 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13238 (rs6000_option_override_internal): Assign to
13239 targetm.secondary_memory_needed rather than
13240 rs6000_secondary_memory_needed_ptr.
13241 (rs6000_secondary_memory_needed): Match hook interface.
13242 (rs6000_debug_secondary_memory_needed): Likewise.
13243 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
13244 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
13245 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
13246 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13247 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
13248 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
13250 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
13251 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13252 (rs6000_option_override_internal): Assign to
13253 targetm.secondary_memory_needed rather than
13254 rs6000_secondary_memory_needed_ptr.
13255 (rs6000_secondary_memory_needed): Match hook interface.
13256 (rs6000_debug_secondary_memory_needed): Likewise.
13257 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
13258 * config/s390/s390.c (s390_secondary_memory_needed): New function.
13259 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13260 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
13261 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
13262 (sparc_secondary_memory_needed): New function.
13263 * lra-constraints.c (check_and_process_move): Refer to
13264 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
13266 (curr_insn_transform): Likewise.
13267 (process_alt_operands): Use targetm.secondary_memory_needed
13268 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13269 (check_secondary_memory_needed_p): Likewise.
13270 (choose_split_class): Likewise.
13271 * reload.c: Unconditionally include code that was previously
13272 conditional on SECONDARY_MEMORY_NEEDED.
13273 (push_secondary_reload): Use targetm.secondary_memory_needed
13274 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13275 (push_reload): Likewise.
13276 * reload1.c: Unconditionally include code that was previously
13277 conditional on SECONDARY_MEMORY_NEEDED.
13278 (choose_reload_regs): Use targetm.secondary_memory_needed
13279 instead of TARGET_SECONDARY_MEMORY_NEEDED.
13280 (gen_reload): Likewise.
13281 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
13283 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
13284 Alan Hayward <alan.hayward@arm.com>
13285 David Sherwood <david.sherwood@arm.com>
13287 * target.def (secondary_memory_needed_mode): New hook:
13288 * targhooks.c (default_secondary_memory_needed_mode): Declare.
13289 * targhooks.h (default_secondary_memory_needed_mode): New function.
13290 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
13291 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
13292 * doc/tm.texi: Regenerate.
13293 * lra-constraints.c (check_and_process_move): Use
13294 targetm.secondary_memory_needed_mode instead of
13295 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
13296 (curr_insn_transform): Likewise.
13297 * reload.c (get_secondary_mem): Likewise.
13298 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13299 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
13301 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13302 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13303 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
13304 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13305 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
13307 * config/powerpcspe/powerpcspe-protos.h
13308 (rs6000_secondary_memory_needed_mode): Delete.
13309 * config/powerpcspe/powerpcspe.c
13310 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13311 (rs6000_secondary_memory_needed_mode): Make static.
13312 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13313 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
13315 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13317 (rs6000_secondary_memory_needed_mode): Make static.
13318 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13319 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
13320 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
13321 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
13322 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
13324 (sparc_secondary_memory_needed_mode): New function.
13325 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
13327 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
13329 * config/aarch64/constraints.md (Umq): New constraint.
13330 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
13332 (mov<mode>): Update condition.
13334 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13336 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
13337 when bitposition is the same.
13339 2017-09-13 Richard Biener <rguenther@suse.de>
13341 * dwarf2out.c (output_die_symbol): Remove.
13342 (output_die): Do not output a DIEs symbol.
13344 2017-09-13 Richard Biener <rguenther@suse.de>
13346 PR middle-end/82128
13347 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
13348 default-def to avoid breaking iterator update with the weird
13349 interaction with cgraph_update_edges_for_call_stmt_node.
13351 2017-09-13 Richard Biener <rguenther@suse.de>
13353 * tree-cfg.c (verify_gimple_assign_binary): Add verification
13354 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
13355 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
13356 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
13358 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
13360 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
13361 Disable pc relative literal load irrespective of
13362 TARGET_FIX_ERR_A53_84341 for default.
13364 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
13366 * config/sparc/sparc.c (output_return): Output the source location of
13367 the insn in the delay slot, if any.
13368 (output_sibcall): Likewise.
13370 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
13373 * common.opt (-static-pie): New alias.
13374 (shared): Negate static-pie.
13375 (-no-pie): Update help text.
13377 (static-pie): New option.
13378 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
13379 -static-pie support.
13380 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
13381 (LINK_EH_SPEC): Likewise.
13382 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
13383 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13384 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
13385 * gcc.c (LINK_COMMAND_SPEC): Likewise.
13386 (init_gcc_specs): Likewise.
13387 (init_spec): Likewise.
13388 (display_help): Update help message for -pie.
13389 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
13392 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
13394 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
13395 (movdi_aarch64): Likewise.
13396 (movti_aarch64): Likewise.
13398 2017-09-12 Simon Wright <simon@pushface.org>
13401 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
13402 calculation of the minor version, always output as 0.
13404 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13407 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
13408 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
13409 on it early, rather than manual conversion late. For
13410 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
13411 instead of performing manual conversion.
13413 2017-09-12 Carl Love <cel@us.ibm.com>
13415 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
13416 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
13418 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
13419 VMULOSW): Add definitions.
13420 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13421 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
13422 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
13423 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
13424 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
13426 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13428 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
13430 (movti_aarch64): Likewise.
13431 (movdf_aarch64): Likewise.
13432 (movtf_aarch64): Likewise.
13433 (load_pairdi): Likewise.
13434 (store_pairdi): Likewise.
13435 (load_pairdf): Likewise.
13436 (store_pairdf): Likewise.
13437 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
13438 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
13439 (ldr_got_small_<mode>): Likewise.
13440 (ldr_got_small_28k_<mode>): Likewise.
13441 (ldr_got_tiny): Likewise.
13442 * config/aarch64/iterators.md (ldst_sz): New.
13443 (ldpstp_sz): Likewise.
13444 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
13446 (thunderx_load): Split load_8 to load_16.
13447 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
13449 (thunderx2t99_storepair_basic): Split store_8 to store_16.
13450 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
13451 (xgene1_store_pair): Split store_8 to store_16.
13452 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
13453 (falkor_st_0_st_sd): Split store_8 to store_16.
13455 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
13457 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
13458 and store1/2/3/4 to store_4/8/12/16.
13459 * config/aarch64/aarch64.md: Update for rename.
13460 * config/arm/arm.md: Likewise.: Likewise.
13461 * config/arm/arm.c: Likewise.
13462 * config/arm/thumb1.md: Likewise.
13463 * config/arm/thumb2.md: Likewise.
13464 * config/arm/vfp.md: Likewise.
13465 * config/arm/arm-generic.md: Likewise.
13466 * config/arm/arm1020e.md: Likewise.
13467 * config/arm/arm1026ejs.md: Likewise.
13468 * config/arm/arm1136jfs.md: Likewise.
13469 * config/arm/arm926ejs.md: Likewise.
13470 * config/arm/cortex-a15.md: Likewise.
13471 * config/arm/cortex-a17.md: Likewise.
13472 * config/arm/cortex-a5.md: Likewise.
13473 * config/arm/cortex-a53.md: Likewise.
13474 * config/arm/cortex-a57.md: Likewise.
13475 * config/arm/cortex-a7.md: Likewise.
13476 * config/arm/cortex-a8.md: Likewise.
13477 * config/arm/cortex-a9.md: Likewise.
13478 * config/arm/cortex-m4.md: Likewise.
13479 * config/arm/cortex-m7.md: Likewise.
13480 * config/arm/cortex-r4.md: Likewise.
13481 * config/arm/exynos-m1.md: Likewise.
13482 * config/arm/fa526.md: Likewise.
13483 * config/arm/fa606te.md: Likewise.
13484 * config/arm/fa626te.md: Likewise.
13485 * config/arm/fa726te.md: Likewise.
13486 * config/arm/fmp626.md: Likewise.
13487 * config/arm/iwmmxt.md: Likewise.
13488 * config/arm/ldmstm.md: Likewise.
13489 * config/arm/marvell-pj4.md: Likewise.
13490 * config/arm/xgene1.md: Likewise.
13491 * config/aarch64/thunderx.md: Likewise.
13492 * config/aarch64/thunderx2t99.md: Likewise.
13493 * config/aarch64/falkor.md: Likewise.
13495 2017-09-12 Martin Liska <mliska@suse.cz>
13497 * attribs.c (private_lookup_attribute): New function.
13498 * attribs.h (private_lookup_attribute): Declared here.
13499 (lookup_attribute): Called from this place.
13501 2017-09-12 Richard Biener <rguenther@suse.de>
13503 PR tree-optimization/82157
13504 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
13505 stmts with side-effects.
13507 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13508 Alan Hayward <alan.hayward@arm.com>
13509 David Sherwood <david.sherwood@arm.com>
13511 * target.def (hard_regno_nregs): New hook.
13512 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
13513 * targhooks.h (default_hard_regno_nregs): Declare.
13514 * targhooks.c (default_hard_regno_nregs): New function.
13515 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
13516 (TARGET_HARD_REGNO_NREGS): ...this hook.
13517 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
13518 (CLASS_MAX_NREGS): Likewise.
13519 * doc/tm.texi: Regenerate.
13520 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
13521 instead of HARD_REGNO_NREGS.
13522 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
13523 HARD_REGNO_NREGS in the comment.
13524 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
13525 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
13526 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
13527 Return an unsigned int.
13528 (TARGET_HARD_REGNO_NREGS): Redefine.
13529 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
13530 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
13531 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13532 (arc_hard_regno_nregs): New function.
13533 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
13534 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
13535 (arm_hard_regno_nregs): New function.
13536 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
13537 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
13538 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
13539 (TARGET_HARD_REGNO_NREGS): Redefine.
13540 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
13541 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
13542 (HARD_REGNO_NREGS): Delete.
13543 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
13544 (cr16_hard_regno_nregs): New function.
13545 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
13546 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
13547 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
13548 (cris_hard_regno_nregs): New function.
13549 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
13550 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
13551 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13552 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
13553 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
13554 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
13555 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
13556 (frv_hard_regno_nregs): Make static. Take and return an
13558 (frv_class_max_nregs): Remove outdated copy of documentation.
13559 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
13560 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
13561 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
13562 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
13563 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
13564 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
13565 (TARGET_HARD_REGNO_NREGS): Redefine.
13566 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
13567 (CLASS_MAX_NREGS): Update comment.
13568 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
13569 (ia64_hard_regno_nregs): New function.
13570 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
13571 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
13572 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
13573 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
13574 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
13576 (m32c_hard_regno_nregs): Likewise. Make static.
13577 (TARGET_HARD_REGNO_NREGS): Redefine.
13578 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
13579 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
13580 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
13581 (m68k_hard_regno_nregs): New function.
13582 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
13583 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
13584 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
13585 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
13586 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
13587 Take and return an unsigned int.
13588 (TARGET_HARD_REGNO_NREGS): Redefine.
13589 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
13590 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
13591 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
13592 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
13593 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
13594 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
13595 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
13596 (msp430_hard_regno_nregs): Make static. Take and return an
13598 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
13599 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
13600 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
13601 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
13602 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
13603 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
13604 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
13605 (TARGET_HARD_REGNO_NREGS): Redefine.
13606 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
13607 (PA_HARD_REGNO_NREGS): ...this.
13608 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
13609 (PA_HARD_REGNO_NREGS): ...this.
13610 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13611 (pa_hard_regno_nregs): New function.
13612 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
13613 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
13614 (pdp11_hard_regno_nregs): New function.
13615 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
13616 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
13617 (rs6000_hard_regno_nregs_hook): New function.
13618 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
13619 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
13620 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
13621 Take and return an unsigned int. Move earlier in file.
13622 (TARGET_HARD_REGNO_NREGS): Redefine.
13623 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
13624 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
13625 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
13626 (rl78_hard_regno_nregs): Make static. Take and return an
13628 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
13629 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
13630 (rs6000_hard_regno_nregs_hook): New function.
13631 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
13632 * config/rx/rx.c (rx_hard_regno_nregs): New function.
13633 (TARGET_HARD_REGNO_NREGS): Redefine.
13634 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
13635 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
13636 instead of HARD_REGNO_NREGS.
13637 (s390_hard_regno_nregs): New function.
13638 (s390_hard_regno_mode_ok): Add comment from s390.h.
13639 (TARGET_HARD_REGNO_NREGS): Redefine.
13640 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
13641 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
13642 (sh_hard_regno_nregs): New function.
13643 (sh_pass_in_reg_p): Use it.
13644 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
13645 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
13646 (sparc_hard_regno_nregs): New function.
13647 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
13648 * config/spu/spu.c (spu_hard_regno_nregs): New function.
13649 (spu_function_arg_advance): Use it, supplying a valid register number.
13650 (TARGET_HARD_REGNO_NREGS): Redefine.
13651 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
13652 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
13653 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
13654 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
13655 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
13656 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
13657 (CLASS_MAX_NREGS): Remove copy of old documentation.
13658 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
13659 (visium_hard_regno_nregs): New function.
13660 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
13661 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
13662 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
13663 xtensa_hard_regno_nregs): New function.
13664 * system.h (HARD_REGNO_NREGS): Poison.
13666 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13668 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
13669 hard_regno_nregs instead of HARD_REGNO_NREGS.
13670 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
13671 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
13672 (c6x_expand_epilogue): Likewise.
13673 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
13674 (frv_read_iacc_argument): Likewise.
13675 * config/sh/sh.c: Include regs.h.
13676 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13677 (regs_used): Likewise.
13678 (output_stack_adjust): Likewise.
13679 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
13680 * expmed.c: Include regs.h.
13681 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
13682 * ree.c: Include regs.h.
13683 (combine_reaching_defs): Use hard_regno_nregs instead of
13685 (add_removable_extension): Likewise.
13687 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13689 * regs.h (hard_regno_nregs): Turn into a function.
13690 (end_hard_regno): Update accordingly.
13691 * caller-save.c (setup_save_areas): Likewise.
13692 (save_call_clobbered_regs): Likewise.
13693 (replace_reg_with_saved_mem): Likewise.
13694 (insert_restore): Likewise.
13695 (insert_save): Likewise.
13696 * combine.c (can_change_dest_mode): Likewise.
13697 (move_deaths): Likewise.
13698 (distribute_notes): Likewise.
13699 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
13700 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
13701 (rs6000_split_multireg_move): Likewise.
13702 (rs6000_register_move_cost): Likewise.
13703 (rs6000_memory_move_cost): Likewise.
13704 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
13705 (rs6000_split_multireg_move): Likewise.
13706 (rs6000_register_move_cost): Likewise.
13707 (rs6000_memory_move_cost): Likewise.
13708 * cselib.c (cselib_reset_table): Likewise.
13709 (cselib_lookup_1): Likewise.
13710 * emit-rtl.c (set_mode_and_regno): Likewise.
13711 * function.c (aggregate_value_p): Likewise.
13712 * ira-color.c (setup_profitable_hard_regs): Likewise.
13713 (check_hard_reg_p): Likewise.
13714 (calculate_saved_nregs): Likewise.
13715 (assign_hard_reg): Likewise.
13716 (improve_allocation): Likewise.
13717 (calculate_spill_cost): Likewise.
13718 * ira-emit.c (modify_move_list): Likewise.
13719 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
13720 (ira_hard_reg_in_set_p): Likewise.
13721 * ira.c (setup_reg_mode_hard_regset): Likewise.
13722 (clarify_prohibited_class_mode_regs): Likewise.
13723 (check_allocation): Likewise.
13724 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13725 (lra_setup_reg_renumber): Likewise.
13726 (setup_try_hard_regno_pseudos): Likewise.
13727 (spill_for): Likewise.
13728 (assign_hard_regno): Likewise.
13729 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
13730 * lra-constraints.c (in_class_p): Likewise.
13731 (lra_constraint_offset): Likewise.
13732 (simplify_operand_subreg): Likewise.
13733 (lra_constraints): Likewise.
13734 (split_reg): Likewise.
13735 (split_if_necessary): Likewise.
13736 (invariant_p): Likewise.
13737 (inherit_in_ebb): Likewise.
13738 * lra-lives.c (process_bb_lives): Likewise.
13739 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
13740 (get_hard_regs): Likewise.
13741 (do_remat): Likewise.
13742 * lra-spills.c (assign_spill_hard_regs): Likewise.
13743 * mode-switching.c (create_pre_exit): Likewise.
13744 * postreload.c (reload_combine_recognize_pattern): Likewise.
13745 * recog.c (peep2_find_free_register): Likewise.
13746 * regcprop.c (kill_value_regno): Likewise.
13747 (set_value_regno): Likewise.
13748 (copy_value): Likewise.
13749 (maybe_mode_change): Likewise.
13750 (find_oldest_value_reg): Likewise.
13751 (copyprop_hardreg_forward_1): Likewise.
13752 * regrename.c (check_new_reg_p): Likewise.
13753 (regrename_do_replace): Likewise.
13754 * reload.c (push_reload): Likewise.
13755 (combine_reloads): Likewise.
13756 (find_dummy_reload): Likewise.
13757 (operands_match_p): Likewise.
13758 (find_reloads): Likewise.
13759 (find_equiv_reg): Likewise.
13760 (reload_adjust_reg_for_mode): Likewise.
13761 * reload1.c (count_pseudo): Likewise.
13762 (count_spilled_pseudo): Likewise.
13763 (find_reg): Likewise.
13764 (clear_reload_reg_in_use): Likewise.
13765 (free_for_value_p): Likewise.
13766 (allocate_reload_reg): Likewise.
13767 (choose_reload_regs): Likewise.
13768 (reload_adjust_reg_for_temp): Likewise.
13769 (emit_reload_insns): Likewise.
13770 (delete_output_reload): Likewise.
13771 * rtlanal.c (subreg_get_info): Likewise.
13772 * sched-deps.c (sched_analyze_reg): Likewise.
13773 * sel-sched.c (init_regs_for_mode): Likewise.
13774 (mark_unavailable_hard_regs): Likewise.
13775 (choose_best_reg_1): Likewise.
13776 (verify_target_availability): Likewise.
13777 * valtrack.c (dead_debug_insert_temp): Likewise.
13778 * var-tracking.c (track_loc_p): Likewise.
13779 (emit_note_insn_var_location): Likewise.
13780 * varasm.c (make_decl_rtl): Likewise.
13781 * reginfo.c (choose_hard_reg_mode): Likewise.
13782 (init_reg_modes_target): Refer directly to
13783 this_target_regs->x_hard_regno_nregs.
13785 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13787 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
13788 instead of hard_regno_nregs.
13790 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13792 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
13793 end_hard_regno instead of hard_regno_nregs.
13794 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
13795 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
13796 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
13797 * ira-color.c (improve_allocation): Likewise.
13798 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13799 * lra-lives.c (mark_regno_live): Likewise.
13800 (mark_regno_dead): Likewise.
13801 * lra-remat.c (operand_to_remat): Likewise.
13802 * lra.c (collect_non_operand_hard_regs): Likewise.
13803 * postreload.c (reload_combine_note_store): Likewise.
13804 (move2add_valid_value_p): Likewise.
13805 * reload.c (regno_clobbered_p): Likewise.
13807 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13809 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
13811 * config/v850/v850.c (v850_reorg): Likewise.
13812 * reload.c (refers_to_regno_for_reload_p): Likewise.
13813 (find_equiv_reg): Likewise.
13814 * reload1.c (reload_reg_reaches_end_p): Likewise.
13816 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13818 * caller-save.c (add_used_regs): Use REG_NREGS instead of
13820 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
13821 * config/arm/arm.c (output_move_neon): Likewise.
13822 (arm_attr_length_move_neon): Likewise.
13823 (neon_split_vcombine): Likewise.
13824 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
13825 (c6x_mark_reg_written): Likewise.
13826 (c6x_dwarf_register_span): Likewise.
13827 * config/i386/i386.c (ix86_save_reg): Likewise.
13828 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
13829 (rws_access_reg): Likewise.
13830 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
13831 * mode-switching.c (create_pre_exit): Likewise.
13832 * ree.c (combine_reaching_defs): Likewise.
13833 (add_removable_extension): Likewise.
13834 * regcprop.c (find_oldest_value_reg): Likewise.
13835 (copyprop_hardreg_forward_1): Likewise.
13836 * reload.c (reload_inner_reg_of_subreg): Likewise.
13837 (push_reload): Likewise.
13838 (combine_reloads): Likewise.
13839 (find_dummy_reload): Likewise.
13840 (reload_adjust_reg_for_mode): Likewise.
13841 * reload1.c (find_reload_regs): Likewise.
13842 (forget_old_reloads_1): Likewise.
13843 (reload_reg_free_for_value_p): Likewise.
13844 (reload_adjust_reg_for_temp): Likewise.
13845 (emit_reload_insns): Likewise.
13846 (delete_output_reload): Likewise.
13847 * sel-sched.c (choose_best_reg_1): Likewise.
13848 (choose_best_pseudo_reg): Likewise.
13850 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13851 Alan Hayward <alan.hayward@arm.com>
13852 David Sherwood <david.sherwood@arm.com>
13854 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
13855 * target.def (slow_unaligned_access): New hook.
13856 * targhooks.h (default_slow_unaligned_access): Declare.
13857 * targhooks.c (default_slow_unaligned_access): New function.
13858 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
13859 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
13860 * doc/tm.texi: Regenerate.
13861 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
13862 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
13863 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
13865 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
13866 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
13868 (rs6000_slow_unaligned_access): New function.
13869 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13870 (expand_block_compare): Likewise.
13871 (expand_strn_compare): Likewise.
13872 (rs6000_rtx_costs): Likewise.
13873 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
13874 (riscv_slow_unaligned_access): Likewise.
13875 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
13876 (riscv_slow_unaligned_access_p): ...this and make static.
13877 (riscv_option_override): Update accordingly.
13878 (riscv_slow_unaligned_access): New function.
13879 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13880 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
13881 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
13882 (rs6000_slow_unaligned_access): New function.
13883 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
13884 (rs6000_rtx_costs): Likewise.
13885 * config/rs6000/rs6000-string.c (expand_block_compare)
13886 (expand_strn_compare): Use targetm.slow_unaligned_access instead
13887 of SLOW_UNALIGNED_ACCESS.
13888 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
13889 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
13890 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
13891 of SLOW_UNALIGNED_ACCESS.
13892 * expmed.c (simple_mem_bitfield_p): Likewise.
13893 * expr.c (alignment_for_piecewise_move): Likewise.
13894 (emit_group_load_1): Likewise.
13895 (emit_group_store): Likewise.
13896 (copy_blkmode_from_reg): Likewise.
13897 (emit_push_insn): Likewise.
13898 (expand_assignment): Likewise.
13899 (store_field): Likewise.
13900 (expand_expr_real_1): Likewise.
13901 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
13902 * lra-constraints.c (simplify_operand_subreg): Likewise.
13903 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
13904 * gimple-ssa-store-merging.c: Likewise in block comment at start
13906 * tree-ssa-strlen.c: Include target.h.
13907 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
13908 of SLOW_UNALIGNED_ACCESS.
13909 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
13911 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
13913 PR rtl-optimization/82185
13914 * expmed.c (emit_store_flag_int): Only test tem if it has been
13917 2017-09-12 Richard Biener <rguenther@suse.de>
13919 PR middle-end/82149
13920 * match.pd ((FTYPE) N CMP CST): Fix typo.
13922 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
13924 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
13926 (mips_near_type_p): Add 'short_call' attribute as a synonym
13928 * doc/extend.texi (short_call): Document new function attribute.
13930 2017-09-12 Jakub Jelinek <jakub@redhat.com>
13933 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
13934 assertion check that in the condition.
13935 (get_atomic_generic_size): Likewise. Before testing if parameter
13936 has pointer type, if it has array type, call for C++
13937 default_conversion to perform array-to-pointer conversion.
13939 2017-09-12 Richard Biener <rguenther@suse.de>
13941 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
13942 for operations we cannot scalarize.
13944 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
13946 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
13947 vectors heap vectors. Clean up comments.
13948 Make visited_bbs a reference.
13949 (profitable_jump_thread_path): Make GC
13950 vectors heap vectors. Clean up comments.
13952 (convert_and_register_jump_thread_path): Make GC vectors heap
13954 (check_subpath_and_update_thread_path): Same. Clean up comments.
13955 Make visited_bbs a reference.
13956 (handle_phi): Abstract common code to to
13957 register_jump_thread_path_if_profitable.
13958 Rename VAR_BB to DEF_BB.
13960 Make GC vectors heap vectors.
13961 Make visited_bbs a reference.
13962 (handle_assignment): Same.
13963 (register_jump_thread_path_if_profitable): New.
13964 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
13966 Make GC vectors heap vectors. Clean up comments.
13967 Make visited_bbs a reference.
13968 (find_jump_threads_backwards): Make visited_bbs live in the stack.
13969 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
13972 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
13975 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
13976 words of E_DImode object are reachable by xtensa_uimm8x4 access.
13978 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
13980 Revert r251800 and r251799.
13982 2017-09-11 Martin Jambor <mjambor@suse.cz>
13985 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
13986 arguments in advance.
13987 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
13988 use it to find predecessor edges.
13989 (naive_outof_ssa): Collect vector of predecessors.
13991 2017-09-08 Jason Merrill <jason@redhat.com>
13993 PR c++/70029 - ICE with ref-qualifier and -flto
13994 * langhooks.h (struct lang_hooks_for_types): Add
13995 copy_lang_qualifiers.
13996 * attribs.c (build_type_attribute_qual_variant): Use it.
13997 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
13999 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
14000 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
14002 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
14005 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
14006 (*mulsi3_sp64): New instruction.
14007 (mulsi3): New expander.
14009 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
14011 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
14013 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14015 * sancov.c: Include memmodel.h.
14017 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
14020 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
14023 2017-09-07 Carl Love <cel@us.ibm.com>
14025 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
14026 the sldi instruction.
14028 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
14030 * sancov.c: Include tm_p.h.
14032 2017-09-07 Jakub Jelinek <jakub@redhat.com>
14035 * output.h (switch_to_other_text_partition): New declaration.
14036 * varasm.c (switch_to_other_text_partition): New function.
14037 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
14038 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
14039 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
14040 to the other text partition before emitting LCL label and switch back
14041 after emitting the word after it.
14043 2017-09-07 Richard Biener <rguenther@suse.de>
14045 * passes.def (pass_split_crit_edges): Remove instance before PRE.
14046 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
14047 critical edges here, after loop init.
14048 (pass_data_pre): Remove PROP_no_crit_edges flags.
14049 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
14050 for valueization of call args to avoid leaking VN_TOP.
14051 (visit_use): Assert we do not visit default defs.
14052 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
14053 Use error_mark_node to more easily detect leaking VN_TOP.
14054 All default-defs are varying, not VN_TOP. Mark them visited.
14055 (run_scc_vn): Make code match comment.
14057 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14059 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
14060 OPTION_MASK_FLOAT128_KEYWORD.
14061 (POWERPC_MASKS): Likewise.
14062 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
14063 support for the -mfloat128-type option, and make -mfloat128
14064 default on PowerPC Linux systems. Define or undefine
14065 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
14066 Define __float128 to be __ieee128 if IEEE 128-bit support is
14067 enabled, or undefine it.
14068 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
14069 Delete defining __FLOAT128_TYPE__.
14070 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
14071 -mfloat128-type option and make -mfloat128 default on PowerPC
14073 (TARGET_FLOAT128_TYPE): Likewise.
14074 (-mfloat128-type): Likewise.
14075 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14076 Delete the -mfloat128-type option and make -mfloat128 default on
14077 PowerPC Linux systems. Always use __ieee128 to be the keyword for
14078 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
14079 128-bit floating point is enabled. Change tests from using
14080 -mfloat128-type to -mfloat128.
14081 (rs6000_mangle_type): Use the correct mangling for the __float128
14082 type even if normal long double is restricted to 64-bits.
14083 (floatn_mode): Enable the _Float128 type by default on VSX Linux
14085 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
14086 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
14087 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
14089 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
14090 documentation for -mfloat128.
14092 2017-09-06 Olivier Hainque <hainque@adacore.com>
14094 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
14096 2017-09-06 Wish Wu <wishwu007@gmail.com>
14097 Jakub Jelinek <jakub@redhat.com>
14099 * asan.c (initialize_sanitizer_builtins): Add
14100 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
14101 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
14102 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
14103 BT_FN_VOID_UINT64_PTR variables.
14104 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
14105 (BT_FN_VOID_UINT16_UINT16): Likewise.
14106 (BT_FN_VOID_UINT32_UINT32): Likewise.
14107 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
14108 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
14109 (BT_FN_VOID_UINT64_PTR): Likewise.
14110 * common.opt (flag_sanitize_coverage): New variable.
14111 (fsanitize-coverage=trace-pc): Remove.
14112 (fsanitize-coverage=): Add.
14113 * flag-types.h (enum sanitize_coverage_code): New enum.
14114 * fold-const.c (fold_range_test): Disable non-short-circuit
14115 optimization if flag_sanitize_coverage.
14116 (fold_truth_andor): Likewise.
14117 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14118 * opts.c (COVERAGE_SANITIZER_OPT): Define.
14119 (coverage_sanitizer_opts): New array.
14120 (get_closest_sanitizer_option): Add OPTS argument, handle also
14121 OPT_fsanitize_coverage_.
14122 (parse_sanitizer_options): Adjusted to also handle
14123 OPT_fsanitize_coverage_.
14124 (common_handle_option): Add OPT_fsanitize_coverage_.
14125 * sancov.c (instrument_comparison, instrument_switch): New function.
14126 (sancov_pass): Add trace-cmp support.
14127 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
14128 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
14129 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
14130 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
14131 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
14132 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
14133 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
14134 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
14135 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
14136 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
14138 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14140 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14141 error. Only quit immediately if parsing is complete.
14142 (BEGIN): Initialize fatal_err and parse_done.
14143 (begin fpu, end fpu): Check number of arguments.
14144 (begin arch, end arch): Likewise.
14145 (begin cpu, end cpu): Likewise.
14146 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14147 (optalias): Likewise.
14149 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
14151 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14152 * config/arm/arm-isa.h: Delete. Move definitions to ...
14153 * arm-cpus.in: ... here. Use new feature and fgroup values.
14154 * config/arm/arm.c (arm_option_override): Use lower case for feature
14156 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14157 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14158 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14160 (print_isa_bits_for): New function.
14161 (gen_isa): New function.
14162 (gen_comm_data): Use print_isa_bits_for.
14163 (define feature): New keyword.
14164 (define fgroup): New keyword.
14165 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
14166 (arm-isa.h): Add rule to generate file.
14167 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14168 case for feature bit names.
14170 2017-09-06 Richard Biener <rguenther@suse.de>
14172 * tree-ssa-pre.c (NECESSARY): Remove.
14173 (create_expression_by_pieces): Do not touch pass-local flags.
14174 (insert_into_preds_of_block): Likewise.
14175 (do_pre_regular_insertion): Likewise.
14176 (eliminate_insert): Likewise.
14177 (eliminate_dom_walker::before_dom_children): Likewise.
14178 (fini_eliminate): Do not look at inserted_exprs.
14179 (mark_operand_necessary): Remove.
14180 (remove_dead_inserted_code): Replace with simple work-list
14181 algorithm based on inserted_exprs and SSA uses.
14182 (pass_pre::execute): Re-order fini_eliminate and
14183 remove_dead_inserted_code.
14185 2017-09-06 Olivier Hainque <hainque@adacore.com>
14187 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
14188 for VxWorks 7. Adjust surrounding comments.
14190 2017-09-06 Richard Biener <rguenther@suse.de>
14192 * gimple-ssa-strength-reduction.c
14193 (find_candidates_dom_walker::before_dom_children): Also allow
14196 2017-09-06 Richard Biener <rguenther@suse.de>
14198 PR tree-optimization/82108
14199 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
14200 for gap in the non-permutation SLP case.
14202 2017-09-06 Martin Jambor <mjambor@suse.cz>
14204 PR tree-optimization/82078
14205 * tree-sra.c (sort_and_splice_var_accesses): Move call to
14206 add_access_to_work_queue...
14207 (build_accesses_from_assign): ...here.
14208 (propagate_all_subaccesses): Make sure racc is the group
14209 representative, if there is one.
14211 2017-09-06 Jakub Jelinek <jakub@redhat.com>
14213 PR middle-end/82095
14214 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
14217 2017-09-06 Richard Biener <rguenther@suse.de>
14219 * gimple-ssa-strength-reduction.c
14220 (find_candidates_dom_walker::before_doom_children): Use a
14221 type and not a mode check.
14223 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
14226 * config/arm/predicates.md (arm_general_adddi_operand): Create new
14228 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
14230 2017-09-05 Jeff Law <law@redhat.com>
14232 PR tree-optimization/64910
14233 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
14234 cases where we have 3 or more operands.
14236 2017-09-05 Jakub Jelinek <jakub@redhat.com>
14238 PR middle-end/81768
14239 * omp-low.c (lower_omp_for): Recompute tree invariant if
14240 gimple_omp_for_initial/final is ADDR_EXPR.
14242 PR middle-end/81768
14243 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
14244 into gimple val before gimplification fo the COND_EXPR.
14246 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
14248 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
14249 REGION_COPY argument.
14250 (thread_through_all_blocks): Remove unused argument to
14251 duplicate_thread_path.
14253 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14254 Alan Hayward <alan.hayward@arm.com>
14255 David Sherwood <david.sherwood@arm.com>
14257 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
14258 Take a scalar_mode rather than a machine_mode.
14259 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14260 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
14261 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
14262 (aarch64_gen_adjusted_ldpstp): Likewise.
14263 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
14265 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14266 Alan Hayward <alan.hayward@arm.com>
14267 David Sherwood <david.sherwood@arm.com>
14269 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
14270 Take a scalar_int_mode instead of a machine_mode.
14271 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14272 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14273 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14274 (aarch64_simd_attr_length_rglist): Delete.
14275 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
14276 a scalar_int_mode instead of a machine_mode.
14277 (aarch64_add_offset): Likewise.
14278 (aarch64_internal_mov_immediate): Likewise
14279 (aarch64_add_constant_internal): Likewise.
14280 (aarch64_add_constant): Likewise.
14281 (aarch64_movw_imm): Likewise.
14282 (aarch64_rtx_arith_op_extract_p): Likewise.
14283 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
14284 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
14285 Remove assert that the mode isn't a vector.
14286 (aarch64_output_scalar_simd_mov_immediate): Likewise.
14287 (aarch64_expand_mov_immediate): Update calls after above changes.
14288 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
14289 (aarch64_and_bitmask_imm): Check for scalar integer modes.
14290 (aarch64_move_imm): Likewise.
14291 (aarch64_can_const_movi_rtx_p): Likewise.
14292 (aarch64_strip_extend): Likewise.
14293 (aarch64_extr_rtx_p): Likewise.
14294 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
14295 a CONST_INT when the mode parameter is VOIDmode.
14296 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
14298 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14300 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
14301 * stor-layout.c (bitwise_mode_for_mode): Likewise.
14302 (bitwise_type_for_mode): Update accordingly.
14304 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14306 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
14307 * stor-layout.c (mode_for_size_tree): Likewise.
14308 (mode_for_array): Update accordingly.
14309 (layout_decl): Likewise.
14310 (compute_record_mode): Likewise. Only set the mode once.
14312 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14314 * target.def (get_mask_mode): Change return type to opt_mode.
14316 * doc/tm.texi: Regenerate.
14317 * targhooks.h (default_get_mask_mode): Return an opt_mode.
14318 * targhooks.c (default_get_mask_mode): Likewise.
14319 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
14320 * optabs-query.c (can_vec_mask_load_store_p): Update use of
14321 targetm.get_mask_mode.
14322 * tree.c (build_truth_vector_type): Likewise.
14324 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14326 * machmode.h (mode_for_vector): Return an opt_mode.
14327 * stor-layout.c (mode_for_vector): Likewise.
14328 (mode_for_int_vector): Update accordingly.
14329 (layout_type): Likewise.
14330 * config/i386/i386.c (emit_memmov): Likewise.
14331 (ix86_expand_set_or_movmem): Likewise.
14332 (ix86_expand_vector_init): Likewise.
14333 (ix86_get_mask_mode): Likewise.
14334 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
14336 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
14337 * expmed.c (extract_bit_field_1): Likewise.
14338 * expr.c (expand_expr_real_2): Likewise.
14339 * optabs-query.c (can_vec_perm_p): Likewise.
14340 (can_vec_mask_load_store_p): Likewise.
14341 * optabs.c (expand_vec_perm): Likewise.
14342 * targhooks.c (default_get_mask_mode): Likewise.
14343 * tree-vect-stmts.c (vectorizable_store): Likewise.
14344 (vectorizable_load): Likewise.
14345 (get_vectype_for_scalar_type_and_size): Likewise.
14347 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14349 * machmode.h (mode_for_int_vector): New function.
14350 * stor-layout.c (mode_for_int_vector): Likewise.
14351 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
14352 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
14353 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
14354 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
14355 (s390_expand_vcond): Likewise.
14357 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14359 * machmode.h (opt_machine_mode): New type.
14360 (opt_mode<T>): Allow construction from anything that can be
14362 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
14363 (mode_for_size): Return an opt_machine_mode.
14364 * stor-layout.c (mode_for_size): Likewise.
14365 (mode_for_size_tree): Update call accordingly.
14366 (bitwise_mode_for_mode): Likewise.
14367 (make_fract_type): Likewise.
14368 (make_accum_type): Likewise.
14369 * caller-save.c (replace_reg_with_saved_mem): Update call
14371 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14372 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14373 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14374 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
14375 * expmed.c (extract_bit_field_1): Likewise.
14376 * reload.c (get_secondary_mem): Likewise.
14377 * varasm.c (assemble_integer): Likewise.
14378 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
14381 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14383 * machmode.h (decimal_float_mode_for_size): New function.
14384 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
14385 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
14386 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
14387 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
14388 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
14389 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
14391 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14393 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
14394 (get_builtin_sync_mode): Likewise.
14395 (expand_ifn_atomic_compare_exchange): Likewise.
14396 (expand_builtin_atomic_clear): Likewise.
14397 (expand_builtin_atomic_test_and_set): Likewise.
14398 (fold_builtin_atomic_always_lock_free): Likewise.
14399 * calls.c (compute_argument_addresses): Likewise.
14400 (emit_library_call_value_1): Likewise.
14401 (store_one_arg): Likewise.
14402 * combine.c (combine_instructions): Likewise.
14403 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
14404 * config/arm/arm.c (arm_function_value): Likewise.
14405 (aapcs_allocate_return_reg): Likewise.
14406 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
14407 * config/i386/i386.c (construct_container): Likewise.
14408 (ix86_gimplify_va_arg): Likewise.
14409 (ix86_expand_sse_cmp): Likewise.
14410 (emit_memmov): Likewise.
14411 (emit_memset): Likewise.
14412 (expand_small_movmem_or_setmem): Likewise.
14413 (ix86_expand_pextr): Likewise.
14414 (ix86_expand_pinsr): Likewise.
14415 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
14416 * config/microblaze/microblaze.c (microblaze_block_move_straight):
14418 * config/mips/mips.c (mips_function_value_1) Likewise.
14419 (mips_block_move_straight): Likewise.
14420 (mips_expand_ins_as_unaligned_store): Likewise.
14421 * config/powerpcspe/powerpcspe.c
14422 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14423 (rs6000_darwin64_record_arg_flush): Likewise.
14424 * config/rs6000/rs6000.c
14425 (rs6000_darwin64_record_arg_advance_flush): Likewise.
14426 (rs6000_darwin64_record_arg_flush): Likewise.
14427 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
14428 (sparc_function_value_1): Likewise.
14429 * config/spu/spu.c (adjust_operand): Likewise.
14430 (spu_emit_branch_or_set): Likewise.
14431 (arith_immediate_p): Likewise.
14432 * emit-rtl.c (gen_lowpart_common): Likewise.
14433 * expr.c (expand_expr_real_1): Likewise.
14434 * function.c (assign_parm_setup_block): Likewise.
14435 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
14436 * reload1.c (alter_reg): Likewise.
14437 * stor-layout.c (mode_for_vector): Likewise.
14438 (layout_type): Likewise.
14440 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
14442 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
14443 (spu_convert_move): Likewise.
14444 * lower-subreg.c (resolve_simple_move): Likewise.
14446 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14449 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
14450 define_insn to a define_expand.
14451 (altivec_vsum2sws_direct): New define_insn.
14452 (altivec_vsumsws): Convert from a define_insn to a define_expand.
14454 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14456 * config/arm/arm.c (arm_option_params_internal): Improve setting of
14459 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
14464 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
14465 realign stack if stack alignment needed is less than incoming
14468 2017-09-05 Marek Polacek <polacek@redhat.com>
14471 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
14474 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
14476 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
14478 2017-09-05 Richard Biener <rguenther@suse.de>
14480 PR tree-optimization/82084
14481 * fold-const.c (can_native_encode_string_p): Handle wide characters.
14483 2017-09-05 Richard Biener <rguenther@suse.de>
14485 PR tree-optimization/82102
14486 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
14488 2017-09-05 Martin Liska <mliska@suse.cz>
14490 PR tree-optimization/82032
14491 * tree-cfg.c (generate_range_test): New function.
14492 * tree-cfg.h (generate_range_test): Declared here.
14493 * tree-cfgcleanup.c (convert_single_case_switch): New function.
14494 (cleanup_control_expr_graph): Use it.
14495 * tree-switch-conversion.c (try_switch_expansion): Remove
14497 (emit_case_nodes): Use generate_range_test.
14499 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14502 * config/i386/i386.md (*<btsc><mode>_mask): Add
14503 TARGET_USE_BT to insn constraint.
14504 (*btr<mode>_mask): Ditto.
14506 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
14508 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
14509 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
14511 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
14514 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
14515 TARGET_NEON and TARGET_IWMMXT.
14516 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
14517 TARGET_NEON and TARGET_IWMMXT.
14518 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
14520 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
14522 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
14523 (ix86_rewrite_tls_address): Ditto.
14524 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
14525 (ix86_rewrite_tls_address_1): Ditto.
14526 (ix86_rewrite_tls_address): Ditto.
14527 * config/i386/predicates.md (tls_address_pattern): New predicate.
14528 * config/i386/i386.md (TLS address splitter): New splitter.
14530 2017-09-04 Richard Biener <rguenther@suse.de>
14532 PR tree-optimization/82084
14533 * fold-const.h (can_native_encode_string_p): Declare.
14534 * fold-const.c (can_native_encode_string_p): Factor out from ...
14535 (native_encode_string): ... here.
14536 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
14537 vectorizing stores from constants we later cannot handle.
14539 2017-09-04 Marek Polacek <polacek@redhat.com>
14542 * doc/invoke.texi: Update -Wtautological-compare documentation.
14544 2017-09-04 Jeff Law <law@redhat.com>
14546 PR tree-optimization/64910
14547 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
14548 swap the first and last operand if the last is a constant.
14550 2017-09-04 Marek Polacek <polacek@redhat.com>
14553 * convert.c (do_narrow): When sanitizing signed integer overflows,
14554 bail out for signed types.
14555 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
14557 2017-09-04 Richard Biener <rguenther@suse.de>
14559 PR tree-optimization/82060
14560 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14561 Move devirtualization after stmt folding and before EH/AB/noreturn
14562 cleanup to get the stmt refs canonicalized. Use a bool instead
14563 of gimple_modified_p since that doesn't work for NOPs. Schedule
14564 NOPs generated by folding for removal.
14566 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14567 Alan Hayward <alan.hayward@arm.com>
14568 David Sherwood <david.sherwood@arm.com>
14570 * coretypes.h (pad_direction): New enum.
14571 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
14572 (FUNCTION_ARG_PADDING): Likewise.
14573 * target.def (function_arg_padding): New hook.
14574 * targhooks.h (default_function_arg_padding): Declare.
14575 * targhooks.c (default_function_arg_padding): New function.
14576 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
14577 (TARGET_FUNCTION_ARG_PADDING): ...this.
14578 * doc/tm.texi: Regenerate.
14579 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
14580 instead of direction.
14581 (compute_argument_addresses): Likewise.
14582 (load_register_parameters): Likewise.
14583 (emit_library_call_value_1): Likewise.
14584 (store_one_arg): Use targetm.calls.function_arg_padding instead
14585 of FUNCTION_ARG_PADDING.
14586 (must_pass_in_stack_var_size_or_pad): Likewise.
14587 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
14588 (emit_group_store): Likewise.
14589 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
14590 instead of FUNCTION_ARG_PADDING.
14591 (emit_push_insn): Likewise, and propagate enum change throughout
14593 * function.h (direction): Delete.
14594 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
14596 * function.c (assign_parm_find_stack_rtl): Likewise.
14597 (assign_parm_setup_block_p): Likewise.
14598 (assign_parm_setup_block): Likewise.
14599 (gimplify_parameters): Likewise.
14600 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
14601 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
14603 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
14604 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14605 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
14606 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
14607 (aarch64_function_arg_padding): ...this new function.
14608 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
14609 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14610 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
14611 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14612 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
14613 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14614 (arm_pad_arg_upward): Replace with...
14615 (arm_function_arg_padding): ...this new function.
14616 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
14618 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
14619 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
14620 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14621 (ia64_hpux_function_arg_padding): Replace with...
14622 (ia64_function_arg_padding): ...this new function. Use pad_direction
14623 instead of direction. Check for TARGET_HPUX.
14624 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
14625 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14626 (iq2000_function_arg_padding): New function.
14627 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
14628 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
14629 (mips_function_arg_padding): ...this new function.
14630 (mips_pad_reg_upward): Update accordingly.
14631 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14632 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
14633 targetm.calls.function_arg_padding.
14634 (FUNCTION_ARG_PADDING): Delete.
14635 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
14636 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
14637 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14638 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
14639 (nios2_block_reg_padding): Return pad_direction instead of direction.
14640 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
14641 instead of direction.
14642 (nios2_function_arg_padding): Likewise. Make static.
14643 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14644 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
14645 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
14646 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
14647 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14648 (pa_function_arg_padding): Make static. Return pad_direction instead
14650 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
14651 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14652 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
14653 instead of direction. Use targetm.calls.function_arg_padding.
14654 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
14655 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14656 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
14657 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
14658 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
14660 (function_arg_padding): Rename to...
14661 (rs6000_function_arg_padding): ...this. Make static. Return
14662 pad_direction instead of direction.
14663 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14664 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
14665 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
14666 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
14667 instead of direction. Use targetm.calls.function_arg_padding.
14668 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
14669 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
14670 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
14671 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
14672 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14673 (function_arg_padding): Rename to...
14674 (rs6000_function_arg_padding): ...this. Make static. Return
14675 pad_direction instead of direction.
14676 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
14677 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
14678 * config/s390/s390.c (s390_function_arg_padding): New function.
14679 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14680 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
14681 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
14682 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
14683 (function_arg_padding): Rename to...
14684 (sparc_function_arg_padding): ...this. Make static. Return
14685 pad_direction instead of direction.
14686 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
14687 * config/spu/spu.c (spu_function_arg_padding): New function.
14688 (TARGET_FUNCTION_ARG_PADDING): Redefine.
14689 * system.h (FUNCTION_ARG_PADDING): Poison.
14691 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14692 Alan Hayward <alan.hayward@arm.com>
14693 David Sherwood <david.sherwood@arm.com>
14695 * target.def (modes_tieable_p): New hook.
14696 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
14697 (TARGET_MODES_TIEABLE_P): ...this.
14698 * doc/tm.texi.in: Regenerate.
14699 * hooks.h (hook_bool_mode_mode_true): Declare.
14700 * hooks.c (hook_bool_mode_mode_true): New function.
14701 * combine.c (subst): Use targetm.modes_tieable_p instead of
14703 * dse.c (find_shift_sequence): Likewise.
14704 * expmed.c (extract_low_bits): Likewise.
14705 * lower-subreg.c: Include target.h.
14706 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
14708 * rtlanal.c (rtx_cost): Likewise.
14709 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
14710 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
14711 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
14712 (TARGET_MODES_TIEABLE_P): Redefine.
14713 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
14714 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
14715 (TARGET_MODES_TIEABLE_P): Redefine.
14716 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
14717 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
14718 (arc_modes_tieable_p): New function.
14719 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
14720 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
14721 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
14722 (arm_modes_tieable_p): Make static.
14723 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
14724 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
14725 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
14726 (TARGET_MODES_TIEABLE_P): Redefine.
14727 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
14728 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
14729 (TARGET_MODES_TIEABLE_P): Redefine.
14730 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
14731 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
14732 (cr16_modes_tieable_p): New function.
14733 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
14734 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
14735 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
14736 (TRULY_NOOP_TRUNCATION): Update comment.
14737 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
14738 (TRULY_NOOP_TRUNCATION): Update comment.
14739 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
14740 (frv_modes_tieable_p): New function.
14741 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
14742 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
14743 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
14744 (TARGET_MODES_TIEABLE_P): Redefine.
14745 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
14746 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
14747 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
14748 (TARGET_MODES_TIEABLE_P): Redefine.
14749 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
14750 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
14751 (ia64_modes_tieable_p): New function.
14752 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
14753 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
14754 (iq2000_modes_tieable_p): New function.
14755 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
14756 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
14757 (lm32_modes_tieable_p): New function.
14758 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
14759 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
14760 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
14761 (TARGET_MODES_TIEABLE_P): Redefine.
14762 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
14763 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
14764 (m32r_modes_tieable_p): New function.
14765 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
14766 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
14767 (m68k_modes_tieable_p): New function.
14768 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
14769 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
14770 (mcore_modes_tieable_p): New function.
14771 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
14772 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
14774 (TARGET_MODES_TIEABLE_P): Redefine.
14775 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
14776 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
14777 * config/mips/mips.c (mips_modes_tieable_p): Make static.
14778 (TARGET_MODES_TIEABLE_P): Redefine.
14779 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
14780 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
14781 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
14782 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
14783 (mn10300_modes_tieable_p): ...this and make static.
14784 (TARGET_MODES_TIEABLE_P): Redefine.
14785 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
14786 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
14787 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
14788 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
14789 (msp430_modes_tieable_p): Make static.
14790 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
14791 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
14792 (TARGET_MODES_TIEABLE_P): Redefine.
14793 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
14794 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
14795 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
14796 (TARGET_MODES_TIEABLE_P): Redefine.
14797 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
14798 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
14799 * config/pa/pa.c (pa_modes_tieable_p): Make static.
14800 (TARGET_MODES_TIEABLE_P): Redefine.
14801 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
14802 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
14803 (pdp11_modes_tieable_p): New function.
14804 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
14805 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
14806 (rs6000_modes_tieable_p): New function.
14807 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14808 * config/powerpcspe/powerpcspe.md: Update comment.
14809 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
14810 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
14811 (TARGET_MODES_TIEABLE_P): Redefine.
14812 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
14813 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
14814 (rl78_modes_tieable_p): New function.
14815 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
14816 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
14817 (rs6000_modes_tieable_p): New function.
14818 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
14819 * config/rs6000/rs6000.md: Update comment.
14820 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
14821 * config/rx/rx.c (rx_modes_tieable_p): New function.
14822 (TARGET_MODES_TIEABLE_P): Redefine.
14823 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
14824 * config/s390/s390.c (s390_modes_tieable_p): New function.
14825 (TARGET_MODES_TIEABLE_P): Redefine.
14826 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
14827 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
14828 (sh_modes_tieable_p): New function.
14829 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
14830 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
14831 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
14832 (sparc_modes_tieable_p): Make static.
14833 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
14834 * config/spu/spu.c (spu_modes_tieable_p): New function.
14835 (TARGET_MODES_TIEABLE_P): Redefine.
14836 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
14837 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
14838 (TARGET_MODES_TIEABLE_P): Redefine.
14839 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
14840 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
14841 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
14842 * config/v850/v850.c (v850_modes_tieable_p): New function.
14843 (TARGET_MODES_TIEABLE_P): Redefine.
14844 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
14845 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
14846 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
14847 (visium_modes_tieable_p): New function.
14848 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
14849 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
14850 (xtensa_modes_tieable_p): New function.
14851 * system.h (MODES_TIEABLE_P): Poison.
14853 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
14854 Alan Hayward <alan.hayward@arm.com>
14855 David Sherwood <david.sherwood@arm.com>
14857 * target.def (hard_regno_mode_ok): New hook.
14858 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
14859 (TARGET_HARD_REGNO_MODE_OK): ...this.
14860 * doc/tm.texi.in: Regenerate.
14861 * hooks.h (hook_bool_uint_mode_true): Declare.
14862 * hooks.c (hook_bool_uint_mode_true): New function.
14863 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
14864 HARD_REGNO_MODE_OK.
14865 * genpreds.c (write_insn_preds_c): Add an include of target.h.
14866 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
14867 instead of HARD_REGNO_MODE_OK.
14868 * caller-save.c: Include target.h.
14869 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
14870 HARD_REGNO_MODE_OK.
14871 * combine.c (can_combine_p): Likewise.
14872 (combinable_i3pat): Likewise.
14873 (can_change_dest_mode): Likewise.
14874 * expr.c (init_expr_target): Likewise.
14875 (convert_move): Likewise.
14876 (convert_modes): Likewise.
14877 * ira.c (setup_prohibited_class_mode_regs): Likewise.
14878 (setup_prohibited_mode_move_regs): Likewise.
14879 * ira.h (target_ira): Likewise.
14880 * lra-assigns.c (find_hard_regno_for_1): Likewise.
14881 * lra-constraints.c (process_alt_operands): Likewise.
14882 (split_reg): Likewise.
14883 * recog.c (peep2_find_free_register): Likewise.
14884 * ree.c (combine_reaching_defs): Likewise.
14885 * regcprop.c (maybe_mode_change): Likewise.
14886 * reginfo.c (init_reg_sets_1): Likewise.
14887 (choose_hard_reg_mode): Likewise.
14888 (simplifiable_subregs): Likewise.
14889 * regrename.c (check_new_reg_p): Likewise.
14890 * reload.c (find_valid_class): Likewise.
14891 (find_valid_class_1): Likewise.
14892 (reload_inner_reg_of_subreg): Likewise.
14893 (push_reload): Likewise.
14894 (combine_reloads): Likewise.
14895 (find_dummy_reload): Likewise.
14896 (find_reloads): Likewise.
14897 * reload1.c (find_reg): Likewise.
14898 (set_reload_reg): Likewise.
14899 (allocate_reload_reg): Likewise.
14900 (choose_reload_regs): Likewise.
14901 (reload_adjust_reg_for_temp): Likewise.
14902 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
14903 (simplify_subreg_regno): Likewise.
14904 * sel-sched.c (init_regs_for_mode): Likewise.
14905 * varasm.c (make_decl_rtl): Likewise.
14906 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
14907 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
14908 HARD_REGNO_MODE_OK.
14909 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
14910 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
14911 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14912 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
14913 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
14914 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14915 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
14916 (arc_mode_class): Delete.
14917 (HARD_REGNO_MODE_OK): Delete.
14918 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14919 (arc_hard_regno_mode_ok): Rename old array to...
14920 (arc_hard_regno_mode_ok_modes): ...this.
14921 (arc_conditional_register_usage): Update accordingly.
14922 (arc_mode_class): Make static.
14923 (arc_hard_regno_mode_ok): New function.
14924 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
14925 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
14926 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14927 (arm_hard_regno_mode_ok): Make static.
14928 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
14929 HARD_REGNO_MODE_OK.
14930 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
14931 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
14932 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
14934 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14935 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
14936 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
14937 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
14938 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
14939 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14940 * config/bfin/predicates.md (valid_reg_operand): Use
14941 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
14942 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
14943 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
14944 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14945 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
14946 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
14947 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14948 (cr16_hard_regno_mode_ok): Make static and return a bool.
14949 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
14950 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14951 (cris_hard_regno_mode_ok): New function.
14952 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
14953 (epiphany_mode_class): Delete.
14954 (HARD_REGNO_MODE_OK): Delete.
14955 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
14956 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14957 (hard_regno_mode_ok): Rename to...
14958 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
14959 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
14960 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
14961 HARD_REGNO_MODE_OK.
14962 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
14963 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
14964 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14965 (frv_hard_regno_mode_ok): Make static and return a bool.
14966 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
14967 HARD_REGNO_MODE_OK.
14968 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
14969 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
14970 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
14971 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
14973 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14974 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
14975 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
14976 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
14978 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14979 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
14980 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14981 (ia64_hard_regno_mode_ok): New function.
14982 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
14983 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14984 (iq2000_hard_regno_mode_ok): New function.
14985 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
14986 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
14987 (lm32_hard_regno_mode_ok): New function.
14988 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
14989 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
14990 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
14991 instead of HARD_REGNO_MODE_OK.
14992 (m32c_hard_regno_ok): Rename to...
14993 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
14994 (m32c_cannot_change_mode_class): Update accordingly.
14995 (TARGET_HARD_REGNO_MODE_OK): Redefine.
14996 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
14997 (m32r_mode_class): Delete.
14998 (HARD_REGNO_MODE_OK): Delete.
14999 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15000 (m32r_hard_regno_mode_ok): Rename to...
15001 (m32r_hard_regno_modes): ...this.
15002 (m32r_mode_class): Make static.
15003 (m32r_hard_regno_mode_ok): New function.
15004 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
15005 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
15006 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15007 (m68k_hard_regno_mode_ok): Make static.
15008 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
15009 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15010 (mcore_hard_regno_mode_ok): New function.
15011 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
15012 (HARD_REGNO_MODE_OK): Delete.
15013 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
15015 (microblaze_hard_regno_mode_ok_p): ...this and make static.
15016 (microblaze_hard_regno_mode_ok): New function.
15017 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15018 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
15019 (mips_hard_regno_mode_ok): Delete.
15020 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
15021 (mips_hard_regno_mode_ok_p): ...this and make static.
15022 (mips_hard_regno_mode_ok_p): Rename to...
15023 (mips_hard_regno_mode_ok_uncached): ...this.
15024 (mips_hard_regno_mode_ok): New function.
15025 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
15026 of HARD_REGNO_MODE_OK.
15027 (mips_option_override): Update after above name changes.
15028 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15029 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
15030 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
15031 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
15032 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
15033 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15034 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
15035 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
15036 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
15037 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15038 (msp430_hard_regno_mode_ok): Make static and return a bool.
15039 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
15040 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
15041 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
15043 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15044 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
15045 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
15046 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
15047 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
15048 (PA_HARD_REGNO_MODE_OK): ...this
15049 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
15050 (PA_HARD_REGNO_MODE_OK): ...this.
15051 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15052 (pa_hard_regno_mode_ok): New function.
15053 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
15054 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15055 (pdp11_hard_regno_mode_ok): New function.
15056 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
15057 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
15059 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
15061 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15062 (rs6000_hard_regno_mode_ok): Rename to...
15063 (rs6000_hard_regno_mode_ok_uncached): ...this.
15064 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15065 (rs6000_hard_regno_mode_ok): New function.
15066 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
15067 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
15068 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
15069 (riscv_hard_regno_mode_ok): ...this and make static.
15070 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15071 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
15072 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
15073 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15074 (rl78_hard_regno_mode_ok): Make static and return bool.
15075 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
15076 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
15078 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
15079 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15080 (rs6000_hard_regno_mode_ok): Rename to...
15081 (rs6000_hard_regno_mode_ok_uncached): ...this.
15082 (rs6000_init_hard_regno_mode_ok): Update accordingly.
15083 (rs6000_hard_regno_mode_ok): New function.
15084 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
15085 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
15086 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15087 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
15088 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
15089 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
15090 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15091 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
15092 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
15093 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15094 (sh_hard_regno_mode_ok): Make static.
15095 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
15096 instead of HARD_REGNO_MODE_OK.
15097 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
15098 (sparc_mode_class): Delete.
15099 (HARD_REGNO_MODE_OK): Delete.
15100 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15101 (hard_regno_mode_classes): Make static.
15102 (sparc_mode_class): Likewise.
15103 (sparc_hard_regno_mode_ok): New function.
15104 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
15105 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
15106 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
15108 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15109 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
15110 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
15111 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
15112 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
15113 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15114 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
15115 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
15116 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
15117 (visium_hard_regno_mode_ok): New function.
15118 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
15119 instead of HARD_REGNO_MODE_OK.
15120 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
15121 (HARD_REGNO_MODE_OK): Delete.
15122 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
15123 (xtensa_hard_regno_mode_ok_p): ...this and make static.
15124 (xtensa_option_override): Update accordingly.
15125 (TARGET_HARD_REGNO_MODE_OK): Redefine.
15126 (xtensa_hard_regno_mode_ok): New function.
15127 * system.h (HARD_REGNO_MODE_OK): Poison.
15129 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15130 Alan Hayward <alan.hayward@arm.com>
15131 David Sherwood <david.sherwood@arm.com>
15133 * target.def (hard_regno_call_part_clobbered): New hook.
15134 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
15135 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
15136 * doc/tm.texi: Regenerate.
15137 * hooks.h (hook_bool_uint_mode_false): Declare.
15138 * hooks.c (hook_bool_uint_mode_false): New function.
15139 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15140 * cselib.c (cselib_process_insn): Use
15141 targetm.hard_regno_call_part_clobbered instead of
15142 HARD_REGNO_CALL_PART_CLOBBERED.
15143 * ira-conflicts.c (ira_build_conflicts): Likewise.
15144 * ira-costs.c (ira_tune_allocno_costs): Likewise.
15145 * lra-constraints.c (need_for_call_save_p): Likewise.
15146 * lra-lives.c: Include target.h.
15147 (check_pseudos_live_through_calls): Use
15148 targetm.hard_regno_call_part_clobbered instead of
15149 HARD_REGNO_CALL_PART_CLOBBERED.
15150 * regcprop.c: Include target.h.
15151 (copyprop_hardreg_forward_1): Use
15152 targetm.hard_regno_call_part_clobbered instead of
15153 HARD_REGNO_CALL_PART_CLOBBERED.
15154 * reginfo.c (choose_hard_reg_mode): Likewise.
15155 * regrename.c (check_new_reg_p): Likewise.
15156 * reload.c (find_equiv_reg): Likewise.
15157 * reload1.c (emit_reload_insns): Likewise.
15158 * sched-deps.c (deps_analyze_insn): Likewise.
15159 * sel-sched.c (init_regs_for_mode): Likewise.
15160 (mark_unavailable_hard_regs): Likewise.
15161 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
15162 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15163 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
15165 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15166 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15167 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
15169 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
15171 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15172 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15173 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
15175 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15176 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15177 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
15179 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15180 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
15182 * config/powerpcspe/powerpcspe.c
15183 (rs6000_hard_regno_call_part_clobbered): New function.
15184 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15185 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15186 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
15188 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15189 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15190 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
15192 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
15193 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
15194 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
15196 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15197 Alan Hayward <alan.hayward@arm.com>
15198 David Sherwood <david.sherwood@arm.com>
15200 * rtl.h (subreg_memory_offset): Declare.
15201 * emit-rtl.c (subreg_memory_offset): New function.
15202 * expmed.c (store_bit_field_1): Use it.
15203 * expr.c (undefined_operand_subword_p): Likewise.
15204 * simplify-rtx.c (simplify_subreg): Likewise.
15206 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
15208 PR rtl-optimization/57448
15211 * optabs.c (expand_atomic_load): Place compiler memory barriers if
15212 using atomic_load pattern.
15213 (expand_atomic_store): Likewise.
15215 2017-09-04 Jakub Jelinek <jakub@redhat.com>
15218 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
15219 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
15220 handling. Use replace_call_with_value with NULL instead of
15221 gsi_replace, unlink_stmt_vdef and release_defs.
15223 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
15226 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
15228 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
15231 * rtl.h (emit_library_call_value_1): Declare.
15232 (emit_library_call): Replace declaration with a series of overloads.
15233 Remove the parameter count argument.
15234 (emit_library_call_value): Likewise.
15235 * calls.c (emit_library_call_value_1): Make global. Replace varargs
15236 with an "rtx_mode_t *".
15237 (emit_library_call_value): Delete.
15238 (emit_library_call): Likewise.
15239 * asan.c (asan_emit_stack_protection): Update calls accordingly.
15240 (asan_emit_allocas_unpoison): Likewise.
15241 * builtins.c (expand_builtin_powi): Likewise.
15242 (expand_asan_emit_allocas_unpoison): Likewise.
15243 * cfgexpand.c (expand_main_function): Likewise.
15244 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
15245 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
15246 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
15247 * config/arm/arm.c (arm_trampoline_init): Likewise.
15248 (arm_call_tls_get_addr): Likewise.
15249 (arm_expand_divmod_libfunc): Likewise.
15250 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
15251 (smulsi3_highpart): Likewise.
15252 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
15253 (c6x_expand_compare): Likewise.
15254 (c6x_expand_movmem): Likewise.
15255 * config/frv/frv.c (frv_trampoline_init): Likewise.
15256 * config/i386/i386.c (ix86_trampoline_init): Likewise.
15257 (ix86_expand_divmod_libfunc): Likewise.
15258 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
15259 (ia64_expand_compare): Likewise.
15260 (ia64_profile_hook): Likewise.
15261 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
15262 (nonlocal_goto): Likewise.
15263 (restore_stack_nonlocal): Likewise.
15264 * config/m32r/m32r.c (block_move_call): Likewise.
15265 (m32r_trampoline_init): Likewise.
15266 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
15267 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
15268 (m68k_call_m68k_read_tp): Likewise.
15269 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
15270 (microblaze_expand_divide): Likewise.
15271 * config/mips/mips.h (mips_args): Likewise.
15272 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
15273 (MIPS_ICACHE_SYNC): Likewise.
15274 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
15275 (nios2_trampoline_init): Likewise.
15276 * config/pa/pa.c (hppa_tls_call): Likewise.
15277 (pa_trampoline_init): Likewise.
15278 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
15279 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
15280 (expand_strn_compare): Likewise.
15281 (rs6000_generate_compare): Likewise.
15282 (rs6000_expand_float128_convert): Likewise.
15283 (output_profile_hook): Likewise.
15284 (rs6000_trampoline_init): Likewise.
15285 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
15286 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
15287 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
15288 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
15289 (rs6000_generate_compare): Likewise.
15290 (rs6000_expand_float128_convert): Likewise.
15291 (output_profile_hook): Likewise.
15292 (rs6000_trampoline_init): Likewise.
15293 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
15294 * config/sh/sh.c (sh_trampoline_init): Likewise.
15295 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
15296 (sparc_emit_float_lib_cmp): Likewise.
15297 (sparc32_initialize_trampoline): Likewise.
15298 (sparc64_initialize_trampoline): Likewise.
15299 (sparc_profile_hook): Likewise.
15300 * config/spu/spu.c (ea_load_store): Likewise.
15301 * config/spu/spu.md (floatunssidf2): Likewise.
15302 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
15303 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
15304 * config/visium/visium.c (expand_block_move_4): Likewise.
15305 (expand_block_move_2): Likewise.
15306 (expand_block_move_1): Likewise.
15307 (expand_block_set_4): Likewise.
15308 (expand_block_set_2): Likewise.
15309 (expand_block_set_1): Likewise.
15310 (visium_trampoline_init): Likewise.
15311 (visium_profile_hook): Likewise.
15312 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
15313 (xtensa_setup_frame_addresses): Likewise.
15314 (xtensa_trampoline_init): Likewise.
15315 * except.c (sjlj_emit_function_enter): Likewise.
15316 (sjlj_emit_function_exit): Likewise.
15317 * explow.c (allocate_dynamic_stack_space): Likewise.
15318 (probe_stack_range): Likewise.
15319 * expr.c (convert_mode_scalar): Likewise.
15320 * optabs.c (expand_binop): Likewise.
15321 (expand_twoval_binop_libfunc): Likewise.
15322 (expand_unop): Likewise.
15323 (prepare_cmp_insn): Likewise.
15324 (prepare_float_lib_cmp): Likewise.
15325 (expand_float): Likewise.
15326 (expand_fix): Likewise.
15327 (expand_fixed_convert): Likewise.
15328 (maybe_emit_sync_lock_test_and_set): Likewise.
15329 (expand_atomic_compare_and_swap): Likewise.
15330 (expand_mem_thread_fence): Likewise.
15331 (expand_atomic_fetch_op): Likewise.
15333 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
15335 * doc/generic.texi (OpenACC): Adjust URL.
15336 * doc/invoke.texi (C Dialect Options): Ditto.
15338 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
15340 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
15341 predicate for operand 1. Add (m,<S>) constraint.
15342 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
15343 Prevent memory operand 1 with register operand 2.
15345 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
15347 PR rtl-optimization/82024
15348 * combine.c (try_combine): If the combination result is a PARALLEL,
15349 and we only need to retain the SET in there that would be placed
15350 at I2, check that we can place that at I3 instead, before doing so.
15352 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15355 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
15357 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
15358 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
15361 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
15362 Jeff Law <law@redhat.com>
15364 * varasm.c (bss_initializer_p): Do not put constants into .bss
15365 (categorize_decl_for_section): Handle bss_initializer_p returning
15366 false when DECL_INITIAL is NULL.
15368 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15371 * config/s390/s390.c (s390_can_inline_p): New function.
15373 2017-09-01 Jeff Law <law@redhat.com>
15375 PR tree-optimization/82052
15376 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
15377 Always initialize the returned slot after a hash table miss
15378 when INSERT is true.
15380 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
15382 * config/s390/s390.md (mem_signal_fence): Remove.
15383 * doc/md.texi (mem_signal_fence): Remove.
15384 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
15386 * target-insns.def (mem_signal_fence): Remove.
15388 2017-09-01 Jakub Jelinek <jakub@redhat.com>
15391 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
15394 * asan.c (create_odr_indicator): Strip name encoding from assembler
15395 name before appending it after __odr_asan_.
15397 2017-09-01 Martin Liska <mliska@suse.cz>
15399 PR tree-optimization/82059
15400 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
15401 frequency only when an edge is redirected.
15403 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15405 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
15406 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
15407 (arc_conditional_register_usage): Remove ARC600 lp_count
15409 (arc_file_start): Emit Tag_ARC_CPU_variation.
15410 (arc_can_use_doloop_p): New conditions to use ZOLs.
15411 (hwloop_fail): New function.
15412 (hwloop_optimize): Likewise.
15413 (hwloop_pattern_reg): Likewise.
15414 (arc_doloop_hooks): New struct, to be used with reorg_loops.
15415 (arc_reorg_loops): New function, calls reorg_loops.
15416 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
15417 (arc600_corereg_hazard): Remove ZOL checking, case handled by
15419 (arc_loop_hazard): Remove function, functionality moved into
15421 (arc_hazard): Remove arc_loop_hazard call.
15422 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
15423 into hwloop_optimize.
15424 (arc_label_align): Remove ZOL handling.
15425 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
15426 * config/arc/arc.md (doloop_begin): Remove pattern.
15427 (doloop_begin_i): Likewise.
15428 (doloop_end_i): Likewise.
15429 (doloop_fallback): Likewise.
15430 (doloop_fallback_m): Likewise.
15431 (doloop_end): Reimplement expand.
15432 (arc_lp): New pattern for LP instruction.
15433 (loop_end): New pattern.
15434 (loop_fail): Likewise.
15435 (decrement_and_branch_until_zero): Likewise.
15436 * config/arc/arc.opt (mlpc-width): New option.
15437 * doc/invoke.texi (mlpc-width): Document option.
15439 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15441 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
15442 (arc_ccfsm_advance): Fix checking for delay slots.
15443 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
15445 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15447 * config/arc/arc.md (movqi_insn): Add stores to save constant long
15449 (movhi_insn): Update store instruction constraint which are saving
15450 6-bit short immediates.
15451 (movsi_insn): Consider also short scaled load operations.
15452 (zero_extendhisi2_i): Use Usd constraint instead of T.
15453 (extendhisi2_i): Add q constraint.
15454 (arc_clzsi2): Add type and length attributes.
15455 (arc_ctzsi2): Likewise.
15456 * config/arc/constraints.md (Usc): Update constraint, the
15457 assembler can parse two relocations for a single instruction.
15459 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
15461 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
15462 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
15464 2017-08-31 Olivier Hainque <hainque@adacore.com>
15466 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
15467 match as powerpc-wrs-vxworks*.
15469 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
15471 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
15472 register constraint for by-element operand.
15473 (aarch64_mls_elt_merge<mode>): Likewise.
15475 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15477 * config/arc/arc.c (arc_can_follow_jump): Check for short
15480 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15482 * config.gcc: Use g.opt for arc.
15483 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
15484 functionality moved to ...
15485 (legitimate_scaled_address_p): New function, ...here.
15486 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
15487 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
15488 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
15490 (arc_override_options): Handle G option.
15491 (arc_output_pic_addr_const): Correct function definition.
15492 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
15493 (arc_decl_anon_ns_mem_p): Delete.
15494 (arc_in_small_data_p): Overhaul this function to take into
15495 consideration the value given via G option.
15496 (arc_rewrite_small_data_1): Renamed and corrected old
15497 arc_rewrite_small_data function.
15498 (arc_rewrite_small_data): New function.
15499 (small_data_pattern): Don't use pic_offset_table_rtx.
15500 * config/arc/arc.h (CC1_SPEC): Recognize G option.
15501 * config/arc/simdext.md (movmisalignv2hi): Use
15502 prepare_move_operands function.
15504 (movmisalign*): Likewise.
15505 * doc/invoke.texi (ARC options): Document -G option.
15507 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
15509 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
15511 * config/arc/arc.c (arc_print_operand): Output scalled address for
15512 sdata whenever is possible.
15513 (arc_in_small_data_p): Allow sdata for 64bit datum when double
15514 load/stores are available.
15515 (compact_sda_memory_operand): Check for the alignment required by
15516 code density instructions.
15517 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
15519 * config/arc/constraints.md (Usd): Update constraint.
15520 (Us0): New constraint.
15521 (Usc): Update constraint.
15523 2017-08-31 Richard Biener <rguenther@suse.de>
15525 PR middle-end/82054
15526 * dwarf2out.c (dwarf2out_early_global_decl): Process each
15527 function only once.
15529 2017-08-31 Tamar Christina <tamar.christina@arm.com>
15531 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
15532 Resize type_signature.
15534 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15535 Alan Hayward <alan.hayward@arm.com>
15536 David Sherwood <david.sherwood@arm.com>
15538 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
15539 subregs whose inner modes can be stored in GPRs.
15540 (aarch64_classify_index): Likewise.
15542 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15543 Alan Hayward <alan.hayward@arm.com>
15544 David Sherwood <david.sherwood@arm.com>
15546 * config/aarch64/iterators.md (V_cmp_result): Rename to...
15547 (V_INT_EQUIV): ...this.
15548 (v_cmp_result): Rename to...
15549 (v_int_equiv): ...this.
15550 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
15551 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
15552 (copysign<mode>3): Likewise.
15553 (aarch64_simd_bsl<mode>_internal): Likewise.
15554 (aarch64_simd_bsl<mode>): Likewise.
15555 (vec_cmp<mode><mode>): Likewise.
15556 (vcond<mode><mode>): Likewise.
15557 (vcond<v_cmp_mixed><mode>): Likewise.
15558 (vcondu<mode><v_cmp_mixed>): Likewise.
15559 (aarch64_cm<optab><mode>): Likewise.
15560 (aarch64_cmtst<mode>): Likewise.
15561 (aarch64_fac<optab><mode>): Likewise.
15562 (vec_perm_const<mode>): Likewise.
15563 (vcond_mask_<mode><v_cmp_result>): Rename to...
15564 (vcond_mask_<mode><v_int_equiv>): ...this.
15565 (vec_cmp<mode><v_cmp_result>): Rename to...
15566 (vec_cmp<mode><v_int_equiv>): ...this.
15568 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
15569 Alan Hayward <alan.hayward@arm.com>
15570 David Sherwood <david.sherwood@arm.com>
15572 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
15574 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
15575 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
15576 (UNSPEC_LD4_DREG): New unspecs.
15577 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
15578 (aarch64_ld2<mode>_dreg_be): Replace with...
15579 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
15581 (aarch64_ld3<mode>_dreg_le)
15582 (aarch64_ld3<mode>_dreg_be): Replace with...
15583 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
15585 (aarch64_ld4<mode>_dreg_le)
15586 (aarch64_ld4<mode>_dreg_be): Replace with...
15587 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
15590 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15592 PR tree-optimization/81987
15593 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
15594 insert an initializer in a location not dominated by the stride
15597 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15599 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
15600 on the entire header of the finally block in the fallthru case.
15602 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
15604 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
15606 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
15608 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
15609 rs6000_emit_move_from_cr and call renamed function.
15610 (rs6000_emit_prologue): Call renamed functions.
15611 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
15612 movesi_from_cr, remove volatile CRs.
15614 2017-08-30 Jon Beniston <jon@beniston.com>
15615 Richard Biener <rguenther@suse.de>
15617 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
15618 of VECTOR_MODE_P check.
15619 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
15620 element vector types.
15622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15624 * df.h (df_read_modify_subreg_p): Remove in favor of...
15625 * rtl.h (read_modify_subreg_p): ...this new function. Take a
15626 const_rtx instead of an rtx.
15627 * cprop.c (local_cprop_find_used_regs): Update accordingly.
15628 * df-problems.c (df_word_lr_mark_ref): Likewise.
15629 * ira-lives.c (mark_pseudo_reg_live): Likewise.
15630 (mark_pseudo_reg_dead): Likewise.
15631 (mark_ref_dead): Likewise.
15632 * reginfo.c (init_subregs_of_mode): Likewise.
15633 * sched-deps.c (sched_analyze_1): Likewise.
15634 * df-scan.c (df_def_record_1): Likewise.
15635 (df_uses_record): Likewise.
15636 (df_read_modify_subreg_p): Remove in favor of...
15637 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
15638 const_rtx instead of an rtx.
15640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15641 Alan Hayward <alan.hayward@arm.com>
15642 David Sherwood <david.sherwood@arm.com>
15644 * rtl.h (partial_subreg_p): New function.
15645 * caller-save.c (save_call_clobbered_regs): Use it.
15646 * calls.c (expand_call): Likewise.
15647 * combine.c (combinable_i3pat): Likewise.
15648 (simplify_set): Likewise.
15649 (make_extraction): Likewise.
15650 (make_compound_operation_int): Likewise.
15651 (gen_lowpart_or_truncate): Likewise.
15652 (force_to_mode): Likewise.
15653 (make_field_assignment): Likewise.
15654 (reg_truncated_to_mode): Likewise.
15655 (record_truncated_value): Likewise.
15656 (move_deaths): Likewise.
15657 * cse.c (record_jump_cond): Likewise.
15658 (cse_insn): Likewise.
15659 * cselib.c (cselib_lookup_1): Likewise.
15660 * expmed.c (extract_bit_field_using_extv): Likewise.
15661 * function.c (assign_parm_setup_reg): Likewise.
15662 * ifcvt.c (noce_convert_multiple_sets): Likewise.
15663 * ira-build.c (create_insn_allocnos): Likewise.
15664 * lra-coalesce.c (merge_pseudos): Likewise.
15665 * lra-constraints.c (match_reload): Likewise.
15666 (simplify_operand_subreg): Likewise.
15667 (curr_insn_transform): Likewise.
15668 * lra-lives.c (process_bb_lives): Likewise.
15669 * lra.c (new_insn_reg): Likewise.
15670 (lra_substitute_pseudo): Likewise.
15671 * regcprop.c (mode_change_ok): Likewise.
15672 (maybe_mode_change): Likewise.
15673 (copyprop_hardreg_forward_1): Likewise.
15674 * reload.c (push_reload): Likewise.
15675 (find_reloads): Likewise.
15676 (find_reloads_subreg_address): Likewise.
15677 * reload1.c (alter_reg): Likewise.
15678 (eliminate_regs_1): Likewise.
15679 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
15681 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
15683 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
15684 back to if statements, including unpack.
15686 2017-08-30 Martin Liska <mliska@suse.cz>
15688 PR inline-asm/82001
15689 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
15691 (func_checker::compare_asm_inputs_outputs): ... this function.
15692 (func_checker::compare_gimple_asm): Use the function to compare
15693 also ASM constrains.
15694 * ipa-icf-gimple.h: Rename the function.
15696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15697 Alan Hayward <alan.hayward@arm.com>
15698 David Sherwood <david.sherwood@arm.com>
15700 * coretypes.h (complex_mode): New type.
15701 * gdbhooks.py (build_pretty_printer): Handle it.
15702 * machmode.h (complex_mode): New class.
15703 (complex_mode::includes_p): New function.
15704 (is_complex_int_mode): Likewise.
15705 (is_complex_float_mode): Likewise.
15706 * genmodes.c (get_mode_class): Handle complex mode classes.
15707 * function.c (expand_function_end): Use is_complex_int_mode.
15709 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15710 Alan Hayward <alan.hayward@arm.com>
15711 David Sherwood <david.sherwood@arm.com>
15713 * coretypes.h (scalar_mode_pod): New typedef.
15714 * gdbhooks.py (build_pretty_printer): Handle it.
15715 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
15716 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
15717 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
15718 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
15719 as_a <scalar_mode>.
15721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15722 Alan Hayward <alan.hayward@arm.com>
15723 David Sherwood <david.sherwood@arm.com>
15725 * machmode.h (mode_for_vector): Take a scalar_mode instead
15727 * stor-layout.c (mode_for_vector): Likewise.
15728 * explow.c (promote_mode): Use as_a <scalar_mode>.
15729 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
15731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15732 Alan Hayward <alan.hayward@arm.com>
15733 David Sherwood <david.sherwood@arm.com>
15735 * target.def (preferred_simd_mode): Take a scalar_mode
15736 instead of a machine_mode.
15737 * targhooks.h (default_preferred_simd_mode): Likewise.
15738 * targhooks.c (default_preferred_simd_mode): Likewise.
15739 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
15740 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
15741 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
15742 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
15743 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
15744 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
15745 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
15746 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
15748 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
15749 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
15750 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
15751 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
15752 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
15753 * doc/tm.texi: Regenerate.
15754 * optabs-query.c (can_vec_mask_load_store_p): Return false for
15757 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15758 Alan Hayward <alan.hayward@arm.com>
15759 David Sherwood <david.sherwood@arm.com>
15761 * target.def (scalar_mode_supported_p): Take a scalar_mode
15762 instead of a machine_mode.
15763 * targhooks.h (default_scalar_mode_supported_p): Likewise.
15764 * targhooks.c (default_scalar_mode_supported_p): Likewise.
15765 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
15766 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
15767 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
15768 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
15769 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
15770 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
15771 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
15772 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
15773 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
15774 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
15775 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
15776 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
15778 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
15779 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
15780 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
15781 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
15782 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
15784 * doc/tm.texi: Regenerate.
15786 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15787 Alan Hayward <alan.hayward@arm.com>
15788 David Sherwood <david.sherwood@arm.com>
15790 * coretypes.h (opt_scalar_mode): New typedef.
15791 * gdbhooks.py (build_pretty_printers): Handle it.
15792 * machmode.h (mode_iterator::get_2xwider): Add overload for
15794 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
15796 * expr.c (convert_mode_scalar): Likewise.
15797 * omp-low.c (omp_clause_aligned_alignment): Likewise.
15798 * optabs.c (expand_float): Likewise.
15799 (expand_fix): Likewise.
15800 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15803 Alan Hayward <alan.hayward@arm.com>
15804 David Sherwood <david.sherwood@arm.com>
15806 * optabs.c (expand_float): Explicitly check for scalars before
15807 using a branching expansion.
15808 (expand_fix): Likewise.
15810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15811 Alan Hayward <alan.hayward@arm.com>
15812 David Sherwood <david.sherwood@arm.com>
15814 * expr.c (convert_mode): Split scalar handling out into...
15815 (convert_mode_scalar): ...this new function. Treat the modes
15818 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15819 Alan Hayward <alan.hayward@arm.com>
15820 David Sherwood <david.sherwood@arm.com>
15822 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
15824 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
15826 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15827 Alan Hayward <alan.hayward@arm.com>
15828 David Sherwood <david.sherwood@arm.com>
15830 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
15831 rather than a machine_mode.
15832 (fixed_from_string): Likewise.
15833 (fixed_convert): Likewise.
15834 (fixed_convert_from_int): Likewise.
15835 (fixed_convert_from_real): Likewise.
15836 (real_convert_from_fixed): Likewise.
15837 * fixed-value.c (fixed_from_double_int): Likewise.
15838 (fixed_from_string): Likewise.
15839 (fixed_convert): Likewise.
15840 (fixed_convert_from_int): Likewise.
15841 (fixed_convert_from_real): Likewise.
15842 (real_convert_from_fixed): Likewise.
15843 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
15845 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15846 Alan Hayward <alan.hayward@arm.com>
15847 David Sherwood <david.sherwood@arm.com>
15849 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
15850 of separate mode class checks. Do not allow vector modes here.
15851 (immed_wide_int_const): Use as_a <scalar_mode>.
15852 * explow.c (trunc_int_for_mode): Likewise.
15853 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
15854 (wi::shwi): Likewise.
15855 (wi::min_value): Likewise.
15856 (wi::max_value): Likewise.
15857 * dwarf2out.c (loc_descriptor): Likewise.
15858 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
15859 for CONST_WIDE_INT.
15861 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15862 Alan Hayward <alan.hayward@arm.com>
15863 David Sherwood <david.sherwood@arm.com>
15865 * tree.h (SCALAR_TYPE_MODE): New macro.
15866 * expr.c (expand_expr_addr_expr_1): Use it.
15867 (expand_expr_real_2): Likewise.
15868 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
15869 (fold_convert_const_fixed_from_int): Likewise.
15870 (fold_convert_const_fixed_from_real): Likewise.
15871 (native_encode_fixed): Likewise
15872 (native_encode_complex): Likewise
15873 (native_encode_vector): Likewise.
15874 (native_interpret_fixed): Likewise.
15875 (native_interpret_real): Likewise.
15876 (native_interpret_complex): Likewise.
15877 (native_interpret_vector): Likewise.
15878 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
15879 (simd_clone_adjust_argument_types): Likewise.
15880 (simd_clone_init_simd_arrays): Likewise.
15881 (simd_clone_adjust): Likewise.
15882 * stor-layout.c (layout_type): Likewise.
15883 * tree.c (build_minus_one_cst): Likewise.
15884 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
15885 * tree-inline.c (estimate_move_cost): Likewise.
15886 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
15887 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
15888 (vectorizable_reduction): Likewise.
15889 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
15890 (vect_recog_mixed_size_cond_pattern): Likewise.
15891 (check_bool_pattern): Likewise.
15892 (adjust_bool_pattern): Likewise.
15893 (search_type_for_mask_1): Likewise.
15894 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
15895 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
15896 (vectorizable_load): Likewise.
15897 (vectorizable_store): Likewise.
15898 * ubsan.c (ubsan_encode_value): Likewise.
15899 * varasm.c (output_constant): Likewise.
15901 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15902 Alan Hayward <alan.hayward@arm.com>
15903 David Sherwood <david.sherwood@arm.com>
15905 * coretypes.h (scalar_mode): New class.
15906 * machmode.h (scalar_mode): Likewise.
15907 (scalar_mode::includes_p): New function.
15908 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
15909 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
15910 * genmodes.c (get_mode_class): Handle remaining scalar modes.
15911 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
15912 * expmed.c (store_bit_field_1): Likewise.
15913 (extract_bit_field_1): Likewise.
15914 * expr.c (write_complex_part): Likewise.
15915 (read_complex_part): Likewise.
15916 (emit_move_complex_push): Likewise.
15917 (expand_expr_real_2): Likewise.
15918 * function.c (assign_parm_setup_reg): Likewise.
15919 (assign_parms_unsplit_complex): Likewise.
15920 * optabs.c (expand_binop): Likewise.
15921 * rtlanal.c (subreg_get_info): Likewise.
15922 * simplify-rtx.c (simplify_immed_subreg): Likewise.
15923 * varasm.c (output_constant_pool_2): Likewise.
15925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15926 Alan Hayward <alan.hayward@arm.com>
15927 David Sherwood <david.sherwood@arm.com>
15929 * expmed.c (extract_high_half): Use scalar_int_mode and remove
15931 (expmed_mult_highpart_optab): Likewise.
15932 (expmed_mult_highpart): Likewise.
15934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
15935 Alan Hayward <alan.hayward@arm.com>
15936 David Sherwood <david.sherwood@arm.com>
15938 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15939 instead of a machine_mode.
15940 (builtin_memset_read_str): Likewise.
15941 * builtins.c (c_readstr): Likewise.
15942 (builtin_memcpy_read_str): Likewise.
15943 (builtin_strncpy_read_str): Likewise.
15944 (builtin_memset_read_str): Likewise.
15945 (builtin_memset_gen_str): Likewise.
15946 (expand_builtin_signbit): Use scalar_int_mode for local variables.
15947 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
15948 instead of a machine_mode.
15949 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
15951 (make_extraction): Likewise.
15952 (try_widen_shift_mode): Take and return scalar_int_modes instead
15954 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
15955 a scalar_int_mode instead of a machine_mode.
15956 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
15957 (avr_addr_space_pointer_mode): Likewise.
15958 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
15959 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
15960 (msp430_unwind_word_mode): Likewise.
15961 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
15962 (spu_addr_space_pointer_mode): Likewise.
15963 (spu_addr_space_address_mode): Likewise.
15964 (spu_libgcc_cmp_return_mode): Likewise.
15965 (spu_libgcc_shift_count_mode): Likewise.
15966 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
15967 (rl78_addr_space_pointer_mode): Likewise.
15968 (fl78_unwind_word_mode): Likewise.
15969 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
15971 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
15972 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
15973 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
15974 (mips_valid_pointer_mode): Likewise.
15975 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
15976 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
15977 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
15979 (ft32_addr_space_address_mode): Likewise.
15980 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
15981 scalar_int_mode instead of a machine_mode.
15982 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
15984 (m32c_addr_space_address_mode): Likewise.
15985 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
15986 (rs6000_eh_return_filter_mode): Likewise.
15987 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
15988 (rs6000_eh_return_filter_mode): Likewise.
15989 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
15990 (s390_libgcc_shift_count_mode): Likewise.
15991 (s390_unwind_word_mode): Likewise.
15992 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
15994 * target.def (mode_rep_extended): Likewise.
15995 (valid_pointer_mode): Likewise.
15996 (addr_space.valid_pointer_mode): Likewise.
15997 (eh_return_filter_mode): Return a scalar_int_mode rather than
15999 (libgcc_cmp_return_mode): Likewise.
16000 (libgcc_shift_count_mode): Likewise.
16001 (unwind_word_mode): Likewise.
16002 (addr_space.pointer_mode): Likewise.
16003 (addr_space.address_mode): Likewise.
16004 * doc/tm.texi: Regenerate.
16005 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
16007 (do_jump): Use scalar_int_mode for local variables.
16008 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
16009 rather than a machine_mode.
16010 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
16011 (scompare_loc_descriptor_wide): Likewise.
16012 (scompare_loc_descriptor_narrow): Likewise.
16013 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
16015 * except.c (sjlj_emit_dispatch_table): Likewise.
16016 (expand_builtin_eh_copy_values): Likewise.
16017 * explow.c (convert_memory_address_addr_space_1): Likewise.
16018 Take a scalar_int_mode rather than a machine_mode.
16019 (convert_memory_address_addr_space): Take a scalar_int_mode rather
16020 than a machine_mode.
16021 (memory_address_addr_space): Use scalar_int_mode for local variables.
16022 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
16023 rather than a machine_mode.
16024 * expmed.c (mask_rtx): Likewise.
16025 (init_expmed_one_conv): Likewise.
16026 (expand_mult_highpart_adjust): Likewise.
16027 (extract_high_half): Likewise.
16028 (expmed_mult_highpart_optab): Likewise.
16029 (expmed_mult_highpart): Likewise.
16030 (expand_smod_pow2): Likewise.
16031 (expand_sdiv_pow2): Likewise.
16032 (emit_store_flag_int): Likewise.
16033 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
16035 (extract_low_bits): Likewise.
16036 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
16038 * expr.c (pieces_addr::adjust): Likewise.
16039 (can_store_by_pieces): Likewise.
16040 (store_by_pieces): Likewise.
16041 (clear_by_pieces_1): Likewise.
16042 (expand_expr_addr_expr_1): Likewise.
16043 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
16044 (expand_expr_real_1): Likewise.
16045 (try_casesi): Likewise.
16046 * final.c (shorten_branches): Likewise.
16047 * fold-const.c (fold_convert_const_int_from_fixed): Change the
16048 type of "mode" to machine_mode.
16049 * internal-fn.c (expand_arith_overflow_result_store): Take a
16050 scalar_int_mode rather than a machine_mode.
16051 (expand_mul_overflow): Use scalar_int_mode for local variables.
16052 * loop-doloop.c (doloop_modify): Likewise.
16053 (doloop_optimize): Likewise.
16054 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
16055 than a machine_mode.
16056 (expand_doubleword_shift_condmove): Likewise.
16057 (expand_doubleword_shift): Likewise.
16058 (expand_doubleword_clz): Likewise.
16059 (expand_doubleword_popcount): Likewise.
16060 (expand_doubleword_parity): Likewise.
16061 (expand_absneg_bit): Use scalar_int_mode for local variables.
16062 (prepare_float_lib_cmp): Likewise.
16063 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
16064 rather than a machine_mode.
16065 (convert_memory_address_addr_space): Likewise.
16066 (get_mode_bounds): Likewise.
16067 (get_address_mode): Return a scalar_int_mode rather than a
16069 * rtlanal.c (get_address_mode): Likewise.
16070 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
16071 than a machine_mode.
16072 * targhooks.c (default_mode_rep_extended): Likewise.
16073 (default_valid_pointer_mode): Likewise.
16074 (default_addr_space_valid_pointer_mode): Likewise.
16075 (default_eh_return_filter_mode): Return a scalar_int_mode rather
16076 than a machine_mode.
16077 (default_libgcc_cmp_return_mode): Likewise.
16078 (default_libgcc_shift_count_mode): Likewise.
16079 (default_unwind_word_mode): Likewise.
16080 (default_addr_space_pointer_mode): Likewise.
16081 (default_addr_space_address_mode): Likewise.
16082 * targhooks.h (default_eh_return_filter_mode): Likewise.
16083 (default_libgcc_cmp_return_mode): Likewise.
16084 (default_libgcc_shift_count_mode): Likewise.
16085 (default_unwind_word_mode): Likewise.
16086 (default_addr_space_pointer_mode): Likewise.
16087 (default_addr_space_address_mode): Likewise.
16088 (default_mode_rep_extended): Take a scalar_int_mode rather than
16090 (default_valid_pointer_mode): Likewise.
16091 (default_addr_space_valid_pointer_mode): Likewise.
16092 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
16094 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
16095 rather than a machine_mode.
16096 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
16097 for local variables.
16098 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16099 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
16100 than a machine_mode.
16102 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16103 Alan Hayward <alan.hayward@arm.com>
16104 David Sherwood <david.sherwood@arm.com>
16106 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
16107 the mode argument to scalar_int_mode.
16108 (do_jump_by_parts_zero_rtx): Likewise.
16109 (do_jump_by_parts_equality_rtx): Likewise.
16110 (do_jump_by_parts_greater): Take a mode argument.
16111 (do_jump_by_parts_equality): Likewise.
16112 (do_jump_1): Update calls accordingly.
16114 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16115 Alan Hayward <alan.hayward@arm.com>
16116 David Sherwood <david.sherwood@arm.com>
16118 * is-a.h (safe_dyn_cast): New function.
16119 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
16120 (jump_table_for_label): Likewise.
16121 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
16122 instead of an rtx_insn *.
16123 (shorten_branches): Use dyn_cast instead of LABEL_P and
16124 JUMP_TABLE_DATA_P. Use jump_table_for_label and
16125 rtx_jump_table_data::get_data_mode.
16126 (final_scan_insn): Likewise.
16128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16129 Alan Hayward <alan.hayward@arm.com>
16130 David Sherwood <david.sherwood@arm.com>
16132 * combine.c (try_combine): Use is_a <scalar_int_mode> when
16133 trying to combine a full-register integer set with a subreg
16136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16137 Alan Hayward <alan.hayward@arm.com>
16138 David Sherwood <david.sherwood@arm.com>
16140 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
16141 that is always either address_mode or pointer_mode.
16143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16144 Alan Hayward <alan.hayward@arm.com>
16145 David Sherwood <david.sherwood@arm.com>
16147 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
16148 when the two are known to be equal.
16150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16151 Alan Hayward <alan.hayward@arm.com>
16152 David Sherwood <david.sherwood@arm.com>
16154 * simplify-rtx.c (simplify_const_unary_operation): Use
16155 is_a <scalar_int_mode> instead of checking for a nonzero
16156 precision. Forcibly convert op_mode to a scalar_int_mode
16157 in that case. More clearly differentiate the operand and
16158 result modes and use the former when deciding what the value
16159 of a count-bits operation should be. Use is_int_mode instead
16160 of checking for a MODE_INT. Remove redundant check for whether
16161 this mode has a zero precision.
16163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16164 Alan Hayward <alan.hayward@arm.com>
16165 David Sherwood <david.sherwood@arm.com>
16167 * optabs.c (widen_leading): Change the type of the mode argument
16168 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
16169 (widen_bswap): Likewise.
16170 (expand_parity): Likewise.
16171 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
16172 (expand_ffs): Likewise.
16173 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
16182 (expand_expr_real_1): Update call accordingly.
16184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16185 Alan Hayward <alan.hayward@arm.com>
16186 David Sherwood <david.sherwood@arm.com>
16188 * expmed.c (store_bit_field_using_insv): Add op0_mode and
16189 value_mode arguments. Use scalar_int_mode internally.
16190 (store_bit_field_1): Rename the new integer mode from imode
16191 to op0_mode and use it instead of GET_MODE (op0). Update calls
16192 to store_split_bit_field, store_bit_field_using_insv and
16193 store_fixed_bit_field.
16194 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
16195 Use scalar_int_mode internally. Use a bit count rather than a mode
16196 when calculating the largest bit size for get_best_mode.
16197 Update calls to store_split_bit_field and store_fixed_bit_field_1.
16198 (store_fixed_bit_field_1): Add mode and value_mode arguments.
16199 Remove assertion that OP0 has a scalar integer mode.
16200 (store_split_bit_field): Add op0_mode and value_mode arguments.
16201 Update calls to extract_fixed_bit_field.
16202 (extract_bit_field_using_extv): Add an op0_mode argument.
16203 Use scalar_int_mode internally.
16204 (extract_bit_field_1): Rename the new integer mode from imode to
16205 op0_mode and use it instead of GET_MODE (op0). Update calls to
16206 extract_split_bit_field, extract_bit_field_using_extv and
16207 extract_fixed_bit_field.
16208 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
16209 to extract_split_bit_field and extract_fixed_bit_field_1.
16210 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
16211 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
16212 on the target mode.
16213 (extract_split_bit_field): Add an op0_mode argument. Update call
16214 to extract_fixed_bit_field.
16216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16217 Alan Hayward <alan.hayward@arm.com>
16218 David Sherwood <david.sherwood@arm.com>
16220 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
16221 * explow.c (hard_function_value): Likewise.
16222 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
16223 convert_to_mode call outside the loop.
16224 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
16225 for the mode iterator. Require the mode specified by max_pieces
16227 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
16229 (copy_blkmode_to_reg): Likewise.
16230 (set_storage_via_setmem): Likewise.
16231 * optabs.c (prepare_cmp_insn): Likewise.
16232 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
16233 * stor-layout.c (finish_bitfield_representative): Likewise.
16235 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16236 Alan Hayward <alan.hayward@arm.com>
16237 David Sherwood <david.sherwood@arm.com>
16239 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
16240 * expr.c (convert_move): Use them.
16241 (convert_modes): Likewise.
16242 (store_expr_with_bounds): Likewise.
16244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16245 Alan Hayward <alan.hayward@arm.com>
16246 David Sherwood <david.sherwood@arm.com>
16248 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
16249 parameter for the mode of "x". Remove the "known_x", "known_mode"
16250 and "known_ret" arguments. Change the type of the mode argument
16251 to scalar_int_mode.
16252 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
16253 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
16254 (reg_num_sign_bit_copies_for_combine): Likewise.
16255 * rtlanal.c (nonzero_bits1): Likewise.
16256 (num_sign_bit_copies1): Likewise.
16257 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
16258 (reg_num_sign_bit_copies_general): Likewise.
16259 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
16260 (reg_nonzero_bits_general): Likewise.
16262 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16263 Alan Hayward <alan.hayward@arm.com>
16264 David Sherwood <david.sherwood@arm.com>
16266 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
16267 than in subroutines. Return 1 for non-integer modes.
16268 (cached_num_sign_bit_copies): Change the type of the mode parameter
16269 to scalar_int_mode.
16270 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
16271 classes. Handle CONST_INT_P first and then check whether X also
16272 has a scalar integer mode. Check the same thing for inner registers
16273 of a SUBREG and for values that are being extended or truncated.
16275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16276 Alan Hayward <alan.hayward@arm.com>
16277 David Sherwood <david.sherwood@arm.com>
16279 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
16280 in subroutines. Return the mode mask for non-integer modes.
16281 (cached_nonzero_bits): Change the type of the mode parameter
16282 to scalar_int_mode.
16283 (nonzero_bits1): Likewise. Remove early exit for other mode
16284 classes. Handle CONST_INT_P first and then check whether X
16285 also has a scalar integer mode.
16287 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16288 Alan Hayward <alan.hayward@arm.com>
16289 David Sherwood <david.sherwood@arm.com>
16291 * expr.c (widest_int_mode_for_size): Make the comment match the code.
16292 Return a scalar_int_mode and assert that the size is greater than
16294 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
16295 (op_by_pieces_d::op_by_pieces_d): Likewise.
16296 (op_by_pieces_d::run): Likewise.
16297 (can_store_by_pieces): Likewise.
16299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16300 Alan Hayward <alan.hayward@arm.com>
16301 David Sherwood <david.sherwood@arm.com>
16303 * combine.c (extract_left_shift): Add a mode argument and update
16305 (make_compound_operation_int): Change the type of the mode parameter
16306 to scalar_int_mode and update the call to extract_left_shift.
16308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16309 Alan Hayward <alan.hayward@arm.com>
16310 David Sherwood <david.sherwood@arm.com>
16312 * combine.c (simplify_and_const_int): Change the type of the mode
16313 parameter to scalar_int_mode.
16314 (simplify_and_const_int_1): Likewise. Update recursive call.
16316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16317 Alan Hayward <alan.hayward@arm.com>
16318 David Sherwood <david.sherwood@arm.com>
16320 * combine.c (simplify_compare_const): Check that the mode is a
16321 scalar_int_mode (rather than VOIDmode) before testing its
16323 (simplify_comparison): Move COMPARISON_P handling out of the
16324 loop and restrict the latter part of the loop to scalar_int_modes.
16325 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
16326 and when considering SUBREG_REGs. Use is_int_mode instead of
16327 checking GET_MODE_CLASS against MODE_INT.
16329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16330 Alan Hayward <alan.hayward@arm.com>
16331 David Sherwood <david.sherwood@arm.com>
16333 * combine.c (try_widen_shift_mode): Move check for equal modes to...
16334 (simplify_shift_const_1): ...here. Use scalar_int_mode for
16335 shift_unit_mode and for modes involved in scalar shifts.
16337 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16338 Alan Hayward <alan.hayward@arm.com>
16339 David Sherwood <david.sherwood@arm.com>
16341 * combine.c (force_int_to_mode): New function, split out from...
16342 (force_to_mode): ...here. Keep xmode up-to-date and use it
16343 instead of GET_MODE (x).
16345 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16346 Alan Hayward <alan.hayward@arm.com>
16347 David Sherwood <david.sherwood@arm.com>
16349 * optabs-query.h (extraction_insn::struct_mode): Change type to
16350 opt_scalar_int_mode and update comment.
16351 (extraction_insn::field_mode): Change type to scalar_int_mode.
16352 (extraction_insn::pos_mode): Likewise.
16353 * combine.c (make_extraction): Update accordingly.
16354 * optabs-query.c (get_traditional_extraction_insn): Likewise.
16355 (get_optab_extraction_insn): Likewise.
16356 * recog.c (simplify_while_replacing): Likewise.
16357 * expmed.c (narrow_bit_field_mem): Change the type of the mode
16358 parameter to opt_scalar_int_mode.
16360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16361 Alan Hayward <alan.hayward@arm.com>
16362 David Sherwood <david.sherwood@arm.com>
16364 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
16365 to a scalar_int_mode instead of a machine_mode.
16366 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
16367 (get_best_mode): Return a boolean and use a pointer argument to store
16368 the selected mode. Replace the limit mode parameter with a bit limit.
16369 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
16370 for the values returned by bit_field_mode_iterator::next_mode.
16371 (store_bit_field): Update call to get_best_mode.
16372 (store_fixed_bit_field): Likewise.
16373 (extract_fixed_bit_field): Likewise.
16374 * expr.c (optimize_bitfield_assignment_op): Likewise.
16375 * fold-const.c (optimize_bit_field_compare): Likewise.
16376 (fold_truth_andor_1): Likewise.
16377 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
16378 Update for new type of m_mode.
16379 (get_best_mode): As above.
16381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16382 Alan Hayward <alan.hayward@arm.com>
16383 David Sherwood <david.sherwood@arm.com>
16385 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
16386 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
16387 (store_bit_field): Check is_a <scalar_int_mode> before calling
16388 strict_volatile_bitfield_p.
16389 (extract_bit_field): Likewise.
16391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16392 Alan Hayward <alan.hayward@arm.com>
16393 David Sherwood <david.sherwood@arm.com>
16395 * target.def (cstore_mode): Return a scalar_int_mode.
16396 * doc/tm.texi: Regenerate.
16397 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
16398 * targhooks.h (default_cstore_mode): Likewise.
16399 * targhooks.c (default_cstore_mode): Likewise, using a forced
16401 * expmed.c (emit_cstore): Expect the target of the cstore to be
16404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16405 Alan Hayward <alan.hayward@arm.com>
16406 David Sherwood <david.sherwood@arm.com>
16408 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
16410 (niter_desc): Likewise mode.
16411 (iv_analyze): Add a mode parameter.
16413 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
16414 and change its type to scalar_int_mode.
16415 * loop-iv.c: Update commentary at head of file.
16416 (iv_constant): Pass the mode paraeter before the rtx it describes
16417 and change its type to scalar_int_mode. Remove VOIDmode handling.
16418 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
16419 (iv_extend): Likewise.
16420 (shorten_into_mode): Likewise.
16421 (iv_add): Use scalar_int_mode.
16422 (iv_mult): Likewise.
16423 (iv_shift): Likewise.
16424 (canonicalize_iv_subregs): Likewise.
16425 (get_biv_step_1): Pass the outer_mode parameter before the rtx
16426 it describes and change its mode to scalar_int_mode. Also change
16427 the type of the returned inner_mode to scalar_int_mode.
16428 (get_biv_step): Likewise, turning outer_mode from a pointer
16429 into a direct parameter. Update call to get_biv_step_1.
16430 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
16431 iv_constant and get_biv_step.
16432 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
16433 and change its type to scalar_int_mode. Don't initialise iv->mode
16434 to VOIDmode and remove later checks for its still being VOIDmode.
16435 Update calls to iv_analyze_op and iv_analyze_expr. Check
16436 is_a <scalar_int_mode> when changing the mode under consideration.
16437 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
16438 Update call to iv_analyze_expr.
16439 (iv_analyze_op): Add a mode parameter. Reject subregs whose
16440 inner register is not also a scalar_int_mode. Update call to
16442 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
16443 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
16444 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
16445 separate mode class checks. Update calls to iv_analyze. Remove
16446 fix-up of VOIDmodes after iv_analyze_biv.
16447 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
16448 don't have a scalar_int_mode. Update call to biv_p.
16450 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16451 Alan Hayward <alan.hayward@arm.com>
16452 David Sherwood <david.sherwood@arm.com>
16454 * cfgexpand.c (convert_debug_memory_address): Use
16455 as_a <scalar_int_mode>.
16456 * combine.c (expand_compound_operation): Likewise.
16457 (make_extraction): Likewise.
16458 (change_zero_ext): Likewise.
16459 (simplify_comparison): Likewise.
16460 * cse.c (cse_insn): Likewise.
16461 * dwarf2out.c (minmax_loc_descriptor): Likewise.
16462 (mem_loc_descriptor): Likewise.
16463 (loc_descriptor): Likewise.
16464 * expmed.c (init_expmed_one_mode): Likewise.
16465 (synth_mult): Likewise.
16466 (emit_store_flag_1): Likewise.
16467 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
16468 of a comparison with size.
16469 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
16470 (reduce_to_bit_field_precision): Likewise.
16471 * function.c (expand_function_end): Likewise.
16472 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
16473 * loop-doloop.c (doloop_modify): Likewise.
16474 * optabs.c (expand_binop): Likewise.
16475 (expand_unop): Likewise.
16476 (expand_copysign_absneg): Likewise.
16477 (prepare_cmp_insn): Likewise.
16478 (maybe_legitimize_operand): Likewise.
16479 * recog.c (const_scalar_int_operand): Likewise.
16480 * rtlanal.c (get_address_mode): Likewise.
16481 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16482 (simplify_cond_clz_ctz): Likewise.
16483 * tree-nested.c (get_nl_goto_field): Likewise.
16484 * tree.c (build_vector_type_for_mode): Likewise.
16485 * var-tracking.c (use_narrower_mode): Likewise.
16487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16488 Alan Hayward <alan.hayward@arm.com>
16489 David Sherwood <david.sherwood@arm.com>
16491 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
16492 * builtins.c (expand_builtin_signbit): Use it.
16493 * cfgexpand.c (expand_debug_expr): Likewise.
16494 * dojump.c (do_jump): Likewise.
16495 (do_compare_and_jump): Likewise.
16496 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
16497 * expmed.c (make_tree): Likewise.
16498 * expr.c (expand_expr_real_2): Likewise.
16499 (expand_expr_real_1): Likewise.
16500 (try_casesi): Likewise.
16501 * fold-const-call.c (fold_const_call_ss): Likewise.
16502 * fold-const.c (unextend): Likewise.
16503 (extract_muldiv_1): Likewise.
16504 (fold_single_bit_test): Likewise.
16505 (native_encode_int): Likewise.
16506 (native_encode_string): Likewise.
16507 (native_interpret_int): Likewise.
16508 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
16509 * internal-fn.c (expand_addsub_overflow): Likewise.
16510 (expand_neg_overflow): Likewise.
16511 (expand_mul_overflow): Likewise.
16512 (expand_arith_overflow): Likewise.
16513 * match.pd: Likewise.
16514 * stor-layout.c (layout_type): Likewise.
16515 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16516 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
16517 * tree-ssanames.c (get_range_info): Likewise.
16518 * tree-switch-conversion.c (array_value_type) Likewise.
16519 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
16520 (vect_recog_divmod_pattern): Likewise.
16521 (vect_recog_mixed_size_cond_pattern): Likewise.
16522 * tree-vrp.c (extract_range_basic): Likewise.
16523 (simplify_float_conversion_using_ranges): Likewise.
16524 * tree.c (int_fits_type_p): Likewise.
16525 * ubsan.c (instrument_bool_enum_load): Likewise.
16526 * varasm.c (mergeable_string_section): Likewise.
16527 (narrowing_initializer_constant_valid_p): Likewise.
16528 (output_constant): Likewise.
16530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16531 Alan Hayward <alan.hayward@arm.com>
16532 David Sherwood <david.sherwood@arm.com>
16534 * machmode.h (NARROWEST_INT_MODE): New macro.
16535 * expr.c (alignment_for_piecewise_move): Use it instead of
16536 GET_CLASS_NARROWEST_MODE (MODE_INT).
16537 (push_block): Likewise.
16538 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
16540 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16543 Alan Hayward <alan.hayward@arm.com>
16544 David Sherwood <david.sherwood@arm.com>
16546 * postreload.c (move2add_valid_value_p): Change the type of the
16547 mode parameter to scalar_int_mode.
16548 (move2add_use_add2_insn): Add a mode parameter and use it instead
16550 (move2add_use_add3_insn): Likewise.
16551 (reload_cse_move2add): Update accordingly.
16553 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16554 Alan Hayward <alan.hayward@arm.com>
16555 David Sherwood <david.sherwood@arm.com>
16557 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
16559 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
16560 * optabs.c (expand_unop): Likewise.
16562 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16563 Alan Hayward <alan.hayward@arm.com>
16564 David Sherwood <david.sherwood@arm.com>
16566 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
16567 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
16568 (popcount_loc_descriptor): Likewise.
16569 (bswap_loc_descriptor): Likewise.
16570 (rotate_loc_descriptor): Likewise.
16571 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
16572 calling the functions above.
16574 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16575 Alan Hayward <alan.hayward@arm.com>
16576 David Sherwood <david.sherwood@arm.com>
16578 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
16580 (try_combine): Likewise.
16581 (simplify_if_then_else): Likewise.
16582 * cse.c (cse_insn): Likewise.
16583 * dwarf2out.c (mem_loc_descriptor): Likewise.
16584 * emit-rtl.c (gen_lowpart_common): Likewise.
16585 * simplify-rtx.c (simplify_truncation): Likewise.
16586 (simplify_binary_operation_1): Likewise.
16587 (simplify_const_relational_operation): Likewise.
16588 (simplify_ternary_operation): Likewise.
16589 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
16591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16592 Alan Hayward <alan.hayward@arm.com>
16593 David Sherwood <david.sherwood@arm.com>
16595 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
16596 * reload.c (push_reload): Likewise.
16597 (find_reloads): Likewise.
16599 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16600 Alan Hayward <alan.hayward@arm.com>
16601 David Sherwood <david.sherwood@arm.com>
16603 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
16604 (make_compound_operation_int): Likewise.
16605 (change_zero_ext): Likewise.
16606 * expr.c (convert_move): Likewise.
16607 (convert_modes): Likewise.
16608 * fwprop.c (forward_propagate_subreg): Likewise.
16609 * loop-iv.c (get_biv_step_1): Likewise.
16610 * optabs.c (widen_operand): Likewise.
16611 * postreload.c (move2add_valid_value_p): Likewise.
16612 * recog.c (simplify_while_replacing): Likewise.
16613 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16614 (simplify_binary_operation_1): Likewise. Remove redundant
16615 mode equality check.
16617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16618 Alan Hayward <alan.hayward@arm.com>
16619 David Sherwood <david.sherwood@arm.com>
16621 * combine.c (combine_simplify_rtx): Add checks for
16622 is_a <scalar_int_mode>.
16623 (simplify_if_then_else): Likewise.
16624 (make_field_assignment): Likewise.
16625 (simplify_comparison): Likewise.
16626 * ifcvt.c (noce_try_bitop): Likewise.
16627 * loop-invariant.c (canonicalize_address_mult): Likewise.
16628 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16631 Alan Hayward <alan.hayward@arm.com>
16632 David Sherwood <david.sherwood@arm.com>
16634 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
16635 is_a <scalar_int_mode> instead of != BLKmode.
16637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16638 Alan Hayward <alan.hayward@arm.com>
16639 David Sherwood <david.sherwood@arm.com>
16641 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
16642 instead of != VOIDmode.
16643 * combine.c (if_then_else_cond): Likewise.
16644 (change_zero_ext): Likewise.
16645 * dwarf2out.c (mem_loc_descriptor): Likewise.
16646 (loc_descriptor): Likewise.
16647 * rtlanal.c (canonicalize_condition): Likewise.
16648 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
16650 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16651 Alan Hayward <alan.hayward@arm.com>
16652 David Sherwood <david.sherwood@arm.com>
16654 * simplify-rtx.c (simplify_binary_operation_1): Use
16655 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
16657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16658 Alan Hayward <alan.hayward@arm.com>
16659 David Sherwood <david.sherwood@arm.com>
16661 * wide-int.h (int_traits<unsigned char>) New class.
16662 (int_traits<unsigned short>) Likewise.
16663 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
16664 Use GET_MODE_UNIT_PRECISION and remove redundant test for
16666 * combine.c (set_nonzero_bits_and_sign_copies): Use
16667 is_a <scalar_int_mode>.
16668 (find_split_point): Likewise.
16669 (combine_simplify_rtx): Likewise.
16670 (simplify_logical): Likewise.
16671 (expand_compound_operation): Likewise.
16672 (expand_field_assignment): Likewise.
16673 (make_compound_operation): Likewise.
16674 (extended_count): Likewise.
16675 (change_zero_ext): Likewise.
16676 (simplify_comparison): Likewise.
16677 * dwarf2out.c (scompare_loc_descriptor): Likewise.
16678 (ucompare_loc_descriptor): Likewise.
16679 (minmax_loc_descriptor): Likewise.
16680 (mem_loc_descriptor): Likewise.
16681 (loc_descriptor): Likewise.
16682 * expmed.c (init_expmed_one_mode): Likewise.
16683 * lra-constraints.c (lra_constraint_offset): Likewise.
16684 * optabs.c (prepare_libcall_arg): Likewise.
16685 * postreload.c (move2add_note_store): Likewise.
16686 * reload.c (operands_match_p): Likewise.
16687 * rtl.h (load_extend_op): Likewise.
16688 * rtlhooks.c (gen_lowpart_general): Likewise.
16689 * simplify-rtx.c (simplify_truncation): Likewise.
16690 (simplify_unary_operation_1): Likewise.
16691 (simplify_binary_operation_1): Likewise.
16692 (simplify_const_binary_operation): Likewise.
16693 (simplify_const_relational_operation): Likewise.
16694 (simplify_subreg): Likewise.
16695 * stor-layout.c (bitwise_mode_for_mode): Likewise.
16696 * var-tracking.c (adjust_mems): Likewise.
16697 (prepare_call_arguments): Likewise.
16699 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16700 Alan Hayward <alan.hayward@arm.com>
16701 David Sherwood <david.sherwood@arm.com>
16703 * machmode.h (is_int_mode): New fuction.
16704 * combine.c (find_split_point): Use it.
16705 (combine_simplify_rtx): Likewise.
16706 (simplify_if_then_else): Likewise.
16707 (simplify_set): Likewise.
16708 (simplify_shift_const_1): Likewise.
16709 (simplify_comparison): Likewise.
16710 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
16711 * cse.c (notreg_cost): Likewise.
16712 (cse_insn): Likewise.
16713 * cselib.c (cselib_lookup_1): Likewise.
16714 * dojump.c (do_jump_1): Likewise.
16715 (do_compare_rtx_and_jump): Likewise.
16716 * dse.c (get_call_args): Likewise.
16717 * dwarf2out.c (rtl_for_decl_init): Likewise.
16718 (native_encode_initializer): Likewise.
16719 * expmed.c (emit_store_flag_1): Likewise.
16720 (emit_store_flag): Likewise.
16721 * expr.c (convert_modes): Likewise.
16722 (store_field): Likewise.
16723 (expand_expr_real_1): Likewise.
16724 * fold-const.c (fold_read_from_constant_string): Likewise.
16725 * gimple-ssa-sprintf.c (get_format_string): Likewise.
16726 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
16727 * optabs.c (expand_binop): Likewise.
16728 (expand_unop): Likewise.
16729 (expand_abs_nojump): Likewise.
16730 (expand_one_cmpl_abs_nojump): Likewise.
16731 * simplify-rtx.c (mode_signbit_p): Likewise.
16732 (val_signbit_p): Likewise.
16733 (val_signbit_known_set_p): Likewise.
16734 (val_signbit_known_clear_p): Likewise.
16735 (simplify_relational_operation_1): Likewise.
16736 * tree.c (vector_type_mode): Likewise.
16738 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16739 Alan Hayward <alan.hayward@arm.com>
16740 David Sherwood <david.sherwood@arm.com>
16742 * machmode.h (smallest_mode_for_size): Fix formatting.
16743 (smallest_int_mode_for_size): New function.
16744 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
16745 instead of smallest_mode_for_size.
16746 * combine.c (make_extraction): Likewise.
16747 * config/arc/arc.c (arc_expand_movmem): Likewise.
16748 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
16749 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
16750 * config/s390/s390.c (s390_expand_insv): Likewise.
16751 * config/sparc/sparc.c (assign_int_registers): Likewise.
16752 * config/spu/spu.c (spu_function_value): Likewise.
16753 (spu_function_arg): Likewise.
16754 * coverage.c (get_gcov_type): Likewise.
16755 (get_gcov_unsigned_t): Likewise.
16756 * dse.c (find_shift_sequence): Likewise.
16757 * expmed.c (store_bit_field_1): Likewise.
16758 * expr.c (convert_move): Likewise.
16759 (store_field): Likewise.
16760 * internal-fn.c (expand_arith_overflow): Likewise.
16761 * optabs-query.c (get_best_extraction_insn): Likewise.
16762 * optabs.c (expand_twoval_binop_libfunc): Likewise.
16763 * stor-layout.c (layout_type): Likewise.
16764 (initialize_sizetypes): Likewise.
16765 * targhooks.c (default_get_mask_mode): Likewise.
16766 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
16768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16769 Alan Hayward <alan.hayward@arm.com>
16770 David Sherwood <david.sherwood@arm.com>
16772 * machmode.h (opt_mode::else_blk): New function.
16773 (int_mode_for_mode): Declare.
16774 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
16775 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
16777 * cfgexpand.c (expand_debug_expr): Likewise.
16778 * combine.c (gen_lowpart_or_truncate): Likewise.
16779 (gen_lowpart_for_combine): Likewise.
16780 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
16781 * config/avr/avr.c (avr_to_int_mode): Likewise.
16782 (avr_out_plus_1): Likewise.
16783 (avr_out_plus): Likewise.
16784 (avr_out_round): Likewise.
16785 * config/i386/i386.c (ix86_split_to_parts): Likewise.
16786 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
16787 (s390_expand_vcond): Likewise.
16788 * config/spu/spu.c (spu_split_immediate): Likewise.
16789 (spu_expand_mov): Likewise.
16790 * dse.c (get_stored_val): Likewise.
16791 * expmed.c (store_bit_field_1): Likewise.
16792 (convert_extracted_bit_field): Use int_mode_for_mode instead of
16794 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
16795 (extract_low_bits): Likewise.
16796 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
16797 handling rather than repeating the check.
16798 (emit_group_store): Likewise.
16799 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
16800 * optabs.c (expand_absneg_bit): Likewise.
16801 (expand_copysign_absneg): Likewise.
16802 (expand_copysign_bit): Likewise.
16803 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
16804 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
16805 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
16806 * var-tracking.c (prepare_call_arguments): Likewise.
16807 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
16808 int_mode_for_mode instead of mode_for_size.
16809 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
16811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16812 Alan Hayward <alan.hayward@arm.com>
16813 David Sherwood <david.sherwood@arm.com>
16815 * machmode.h (int_mode_for_size): New function.
16816 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
16817 instead of mode_for_size.
16818 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
16820 * combine.c (expand_field_assignment): Use int_mode_for_size
16821 instead of mode_for_size.
16822 (make_extraction): Likewise.
16823 (simplify_shift_const_1): Likewise.
16824 (simplify_comparison): Likewise.
16825 * dojump.c (do_jump): Likewise.
16826 * dwarf2out.c (mem_loc_descriptor): Likewise.
16827 * emit-rtl.c (init_derived_machine_modes): Likewise.
16828 * expmed.c (flip_storage_order): Likewise.
16829 (convert_extracted_bit_field): Likewise.
16830 * expr.c (copy_blkmode_from_reg): Likewise.
16831 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
16832 * internal-fn.c (expand_mul_overflow): Likewise.
16833 * lower-subreg.c (simple_move): Likewise.
16834 * optabs-libfuncs.c (init_optabs): Likewise.
16835 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
16836 * tree.c (vector_type_mode): Likewise.
16837 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
16838 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
16839 * tree-vect-generic.c (expand_vector_parallel): Likewise.
16840 * tree-vect-stmts.c (vectorizable_load): Likewise.
16841 (vectorizable_store): Likewise.
16843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16844 Alan Hayward <alan.hayward@arm.com>
16845 David Sherwood <david.sherwood@arm.com>
16847 * coretypes.h (pod_mode): New type.
16848 (scalar_int_mode_pod): New typedef.
16849 * machmode.h (pod_mode): New class.
16850 (int_n_data_t::m): Change type to scalar_int_mode_pod.
16851 * genmodes.c (emit_mode_int_n): Update accordingly.
16852 * lower-subreg.h (target_lower_subreg): Change type to
16853 scalar_int_mode_pod.
16854 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
16855 scalar_int_mode_pod.
16857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16858 Alan Hayward <alan.hayward@arm.com>
16859 David Sherwood <david.sherwood@arm.com>
16861 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
16862 machine_mode to scalar_int_mode.
16863 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
16864 (rs6000_option_override_internal): Remove cast to int.
16865 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
16866 machine_mode to scalar_int_mode.
16867 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
16868 (rs6000_option_override_internal): Remove cast to int.
16869 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
16870 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
16872 * config/s390/s390.c (s390_expand_builtin): Likewise.
16873 * coretypes.h (scalar_int_mode): New type.
16874 (opt_scalar_int_mode): New typedef.
16875 * machmode.h (scalar_int_mode): New class.
16876 (scalar_int_mode::includes_p): New function.
16877 (byte_mode): Change type to scalar_int_mode.
16878 (word_mode): Likewise.
16879 (ptr_mode): Likewise.
16880 * emit-rtl.c (byte_mode): Likewise.
16881 (word_mode): Likewise.
16882 (ptr_mode): Likewise.
16883 (init_derived_machine_modes): Update accordingly.
16884 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
16885 and MODE_PARTIAL_INT.
16886 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
16887 opt_scalar_int_mode.
16889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16890 Alan Hayward <alan.hayward@arm.com>
16891 David Sherwood <david.sherwood@arm.com>
16893 * target.def (libgcc_floating_mode_supported_p): Take a
16895 * doc/tm.texi: Regenerate.
16896 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
16898 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
16899 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
16902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16903 Alan Hayward <alan.hayward@arm.com>
16904 David Sherwood <david.sherwood@arm.com>
16906 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
16907 * doc/tm.texi: Regenerate.
16908 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
16909 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
16910 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
16911 * targhooks.h (default_floatn_mode): Likewise.
16912 * targhooks.c (default_floatn_mode): Likewise.
16913 * tree.c (build_common_tree_nodes): Update accordingly.
16915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16916 Alan Hayward <alan.hayward@arm.com>
16917 David Sherwood <david.sherwood@arm.com>
16919 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
16920 (mode_iterator::iterate_p): Likewise.
16921 (mode_iterator::get_wider): Likewise.
16922 * expr.c (init_expr_target): Use opt_scalar_float_mode.
16924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16925 Alan Hayward <alan.hayward@arm.com>
16926 David Sherwood <david.sherwood@arm.com>
16928 * coretypes.h (opt_scalar_float_mode): New typedef.
16929 * machmode.h (float_mode_for_size): New function.
16930 * emit-rtl.c (double_mode): Delete.
16931 (init_emit_once): Use float_mode_for_size.
16932 * stor-layout.c (layout_type): Likewise.
16933 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
16935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16936 Alan Hayward <alan.hayward@arm.com>
16937 David Sherwood <david.sherwood@arm.com>
16939 * output.h (assemble_real): Take a scalar_float_mode.
16940 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
16941 * config/arm/arm.md (consttable_4): Likewise.
16942 (consttable_8): Likewise.
16943 (consttable_16): Likewise.
16944 * config/mips/mips.md (consttable_float): Likewise.
16945 * config/s390/s390.c (s390_output_pool_entry): Likewise.
16946 * varasm.c (assemble_real): Take a scalar_float_mode.
16947 (output_constant_pool_2): Update accordingly.
16948 (output_constant): Likewise.
16950 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16951 Alan Hayward <alan.hayward@arm.com>
16952 David Sherwood <david.sherwood@arm.com>
16954 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
16955 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
16956 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
16957 (native_encode_real): Likewise.
16958 (native_interpret_real): Likewise.
16959 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
16960 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
16962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16963 Alan Hayward <alan.hayward@arm.com>
16964 David Sherwood <david.sherwood@arm.com>
16966 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
16967 <scalar_float_mode>. Simplify.
16968 (gen_extend_conv_libfunc): Likewise.
16970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
16971 Alan Hayward <alan.hayward@arm.com>
16972 David Sherwood <david.sherwood@arm.com>
16974 * coretypes.h (scalar_float_mode): New type.
16975 * machmode.h (mode_traits::from_int): Use machine_mode if
16976 USE_ENUM_MODES is defined.
16977 (is_a): New function.
16979 (dyn_cast): Likewise.
16980 (scalar_float_mode): New class.
16981 (scalar_float_mode::includes_p): New function.
16982 (is_float_mode): Likewise.
16983 * gdbhooks.py (MachineModePrinter): New class.
16984 (build_pretty_printer): Use it for scalar_float_mode.
16985 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
16986 (format_helper::format_helper): Turn into a template.
16987 * genmodes.c (get_mode_class): New function.
16988 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
16989 or machine_mode if none.
16990 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
16991 as_a <scalar_float_mode>.
16992 * dwarf2out.c (mem_loc_descriptor): Likewise.
16993 (insert_float): Likewise.
16994 (add_const_value_attribute): Likewise.
16995 * simplify-rtx.c (simplify_immed_subreg): Likewise.
16996 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
16997 (expand_unop): Update accordingly.
16998 (expand_abs_nojump): Likewise.
16999 (expand_copysign_absneg): Take a scalar_float_mode.
17000 (expand_copysign_bit): Likewise.
17001 (expand_copysign): Update accordingly.
17003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17004 Alan Hayward <alan.hayward@arm.com>
17005 David Sherwood <david.sherwood@arm.com>
17007 * coretypes.h (opt_mode): New class.
17008 * machmode.h (opt_mode): Likewise.
17009 (opt_mode::else_void): New function.
17010 (opt_mode::require): Likewise.
17011 (opt_mode::exists): Likewise.
17012 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
17013 (GET_MODE_2XWIDER_MODE): Likewise.
17014 (mode_iterator::get_wider): Update accordingly.
17015 (mode_iterator::get_2xwider): Likewise.
17016 (mode_iterator::get_known_wider): Likewise, turning into a template.
17017 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
17018 forcing a wider mode to exist.
17019 * config/cr16/cr16.h (LONG_REG_P): Likewise.
17020 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17021 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
17022 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
17023 * lower-subreg.c (init_lower_subreg): Likewise.
17024 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
17025 on the final iteration.
17026 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
17027 a wider mode exists before asking for a move pattern.
17028 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
17029 forcing a wider mode to exist.
17030 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
17031 returning false if no such mode exists.
17032 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
17033 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
17034 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
17035 Avoid checking for a MODE_INT if we already know the mode is not a
17037 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
17038 forcing a wider mode to exist.
17039 (expmed_mult_highpart_optab): Likewise.
17040 (expmed_mult_highpart): Likewise.
17041 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
17043 * lto-streamer-in.c (lto_input_mode_table): Likewise.
17044 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
17045 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
17046 * internal-fn.c (expand_mul_overflow): Update use of
17047 GET_MODE_2XWIDER_MODE.
17048 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17049 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
17050 GET_MODE_WIDER_MODE.
17051 (convert_plusminus_to_widen): Likewise.
17052 * tree-switch-conversion.c (array_value_type): Likewise.
17053 * var-tracking.c (emit_note_insn_var_location): Likewise.
17054 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
17055 Return false inside rather than outside the loop if no wider mode
17057 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
17058 and GET_MODE_2XWIDER_MODE
17059 (can_compare_p): Use else_void.
17060 * gdbhooks.py (OptMachineModePrinter): New class.
17061 (build_pretty_printer): Use it for opt_mode.
17063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17064 Alan Hayward <alan.hayward@arm.com>
17065 David Sherwood <david.sherwood@arm.com>
17067 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
17068 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
17070 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17071 Alan Hayward <alan.hayward@arm.com>
17072 David Sherwood <david.sherwood@arm.com>
17074 * machmode.h (mode_traits): New structure.
17075 (get_narrowest_mode): New function.
17076 (mode_iterator::start): Likewise.
17077 (mode_iterator::iterate_p): Likewise.
17078 (mode_iterator::get_wider): Likewise.
17079 (mode_iterator::get_known_wider): Likewise.
17080 (mode_iterator::get_2xwider): Likewise.
17081 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
17082 (FOR_EACH_MODE): Likewise.
17083 (FOR_EACH_MODE_FROM): Likewise.
17084 (FOR_EACH_MODE_UNTIL): Likewise.
17085 (FOR_EACH_WIDER_MODE): Likewise.
17086 (FOR_EACH_2XWIDER_MODE): Likewise.
17087 * builtins.c (expand_builtin_strlen): Use new mode iterators.
17088 * combine.c (simplify_comparison): Likewise
17089 * config/i386/i386.c (type_natural_mode): Likewise.
17090 * cse.c (cse_insn): Likewise.
17091 * dse.c (find_shift_sequence): Likewise.
17092 * emit-rtl.c (init_derived_machine_modes): Likewise.
17093 (init_emit_once): Likewise.
17094 * explow.c (hard_function_value): Likewise.
17095 * expmed.c (extract_fixed_bit_field_1): Likewise.
17096 (extract_bit_field_1): Likewise.
17097 (expand_divmod): Likewise.
17098 (emit_store_flag_1): Likewise.
17099 * expr.c (init_expr_target): Likewise.
17100 (convert_move): Likewise.
17101 (alignment_for_piecewise_move): Likewise.
17102 (widest_int_mode_for_size): Likewise.
17103 (emit_block_move_via_movmem): Likewise.
17104 (copy_blkmode_to_reg): Likewise.
17105 (set_storage_via_setmem): Likewise.
17106 (compress_float_constant): Likewise.
17107 * omp-low.c (omp_clause_aligned_alignment): Likewise.
17108 * optabs-query.c (get_best_extraction_insn): Likewise.
17109 * optabs.c (expand_binop): Likewise.
17110 (expand_twoval_unop): Likewise.
17111 (expand_twoval_binop): Likewise.
17112 (widen_leading): Likewise.
17113 (widen_bswap): Likewise.
17114 (expand_parity): Likewise.
17115 (expand_unop): Likewise.
17116 (prepare_cmp_insn): Likewise.
17117 (prepare_float_lib_cmp): Likewise.
17118 (expand_float): Likewise.
17119 (expand_fix): Likewise.
17120 (expand_sfix_optab): Likewise.
17121 * postreload.c (move2add_use_add2_insn): Likewise.
17122 * reg-stack.c (reg_to_stack): Likewise.
17123 * reginfo.c (choose_hard_reg_mode): Likewise.
17124 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
17125 * stor-layout.c (mode_for_size): Likewise.
17126 (smallest_mode_for_size): Likewise.
17127 (mode_for_vector): Likewise.
17128 (finish_bitfield_representative): Likewise.
17129 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
17130 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
17131 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
17132 * var-tracking.c (prepare_call_arguments): Likewise.
17134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17135 Alan Hayward <alan.hayward@arm.com>
17136 David Sherwood <david.sherwood@arm.com>
17138 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
17139 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
17140 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
17141 * machmode.h (mode_size): Move earlier in file.
17142 (mode_precision): Likewise.
17143 (mode_inner): Likewise.
17144 (mode_nunits): Likewise.
17145 (mode_unit_size): Likewise.
17146 (unit_unit_precision): Likewise.
17147 (mode_wider): Likewise.
17148 (mode_2xwider): Likewise.
17149 (machine_mode): New class.
17150 (mode_to_bytes): New function.
17151 (mode_to_bits): Likewise.
17152 (mode_to_precision): Likewise.
17153 (mode_to_inner): Likewise.
17154 (mode_to_unit_size): Likewise.
17155 (mode_to_unit_precision): Likewise.
17156 (mode_to_nunits): Likewise.
17157 (GET_MODE_SIZE): Use mode_to_bytes.
17158 (GET_MODE_BITSIZE): Use mode_to_bits.
17159 (GET_MODE_PRECISION): Use mode_to_precision.
17160 (GET_MODE_INNER): Use mode_to_inner.
17161 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
17162 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
17163 (GET_MODE_NUNITS): Use mode_to_nunits.
17164 * system.h (ALWAYS_INLINE): New macro.
17165 * config/powerpcspe/powerpcspe-c.c
17166 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
17167 int for arg1_mode and arg2_mode.
17169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17170 Alan Hayward <alan.hayward@arm.com>
17171 David Sherwood <david.sherwood@arm.com>
17173 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17174 Prefix mode names with E_ in case statements.
17175 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17176 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
17177 (aarch64_split_simd_move): Likewise.
17178 (aarch64_gen_storewb_pair): Likewise.
17179 (aarch64_gen_loadwb_pair): Likewise.
17180 (aarch64_gen_store_pair): Likewise.
17181 (aarch64_gen_load_pair): Likewise.
17182 (aarch64_get_condition_code_1): Likewise.
17183 (aarch64_constant_pool_reload_icode): Likewise.
17184 (get_rsqrte_type): Likewise.
17185 (get_rsqrts_type): Likewise.
17186 (get_recpe_type): Likewise.
17187 (get_recps_type): Likewise.
17188 (aarch64_gimplify_va_arg_expr): Likewise.
17189 (aarch64_simd_container_mode): Likewise.
17190 (aarch64_emit_load_exclusive): Likewise.
17191 (aarch64_emit_store_exclusive): Likewise.
17192 (aarch64_expand_compare_and_swap): Likewise.
17193 (aarch64_gen_atomic_cas): Likewise.
17194 (aarch64_emit_bic): Likewise.
17195 (aarch64_emit_atomic_swap): Likewise.
17196 (aarch64_emit_atomic_load_op): Likewise.
17197 (aarch64_evpc_trn): Likewise.
17198 (aarch64_evpc_uzp): Likewise.
17199 (aarch64_evpc_zip): Likewise.
17200 (aarch64_evpc_ext): Likewise.
17201 (aarch64_evpc_rev): Likewise.
17202 (aarch64_evpc_dup): Likewise.
17203 (aarch64_gen_ccmp_first): Likewise.
17204 (aarch64_gen_ccmp_next): Likewise.
17205 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
17206 (alpha_emit_xfloating_libcall): Likewise.
17207 (emit_insxl): Likewise.
17208 (alpha_arg_type): Likewise.
17209 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
17210 (arc_preferred_simd_mode): Likewise.
17211 (arc_secondary_reload): Likewise.
17212 (get_arc_condition_code): Likewise.
17213 (arc_print_operand): Likewise.
17214 (arc_legitimate_constant_p): Likewise.
17215 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17216 * config/arc/arc.md (casesi_load): Likewise.
17217 (casesi_compact_jump): Likewise.
17218 * config/arc/predicates.md (proper_comparison_operator): Likewise.
17219 (cc_use_register): Likewise.
17220 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17221 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17222 (arm_init_iwmmxt_builtins): Likewise.
17223 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
17224 (neon_expand_vector_init): Likewise.
17225 (arm_attr_length_move_neon): Likewise.
17226 (maybe_get_arm_condition_code): Likewise.
17227 (arm_emit_vector_const): Likewise.
17228 (arm_preferred_simd_mode): Likewise.
17229 (arm_output_iwmmxt_tinsr): Likewise.
17230 (thumb1_output_casesi): Likewise.
17231 (thumb2_output_casesi): Likewise.
17232 (arm_emit_load_exclusive): Likewise.
17233 (arm_emit_store_exclusive): Likewise.
17234 (arm_expand_compare_and_swap): Likewise.
17235 (arm_evpc_neon_vuzp): Likewise.
17236 (arm_evpc_neon_vzip): Likewise.
17237 (arm_evpc_neon_vrev): Likewise.
17238 (arm_evpc_neon_vtrn): Likewise.
17239 (arm_evpc_neon_vext): Likewise.
17240 (arm_validize_comparison): Likewise.
17241 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
17242 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
17243 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
17244 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
17245 (c6x_preferred_simd_mode): Likewise.
17246 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
17247 (epiphany_rtx_costs): Likewise.
17248 * config/epiphany/predicates.md (proper_comparison_operator):
17250 * config/frv/frv.c (condexec_memory_operand): Likewise.
17251 (frv_emit_move): Likewise.
17252 (output_move_single): Likewise.
17253 (output_condmove_single): Likewise.
17254 (frv_hard_regno_mode_ok): Likewise.
17255 (frv_matching_accg_mode): Likewise.
17256 * config/h8300/h8300.c (split_adds_subs): Likewise.
17257 (h8300_rtx_costs): Likewise.
17258 (h8300_print_operand): Likewise.
17259 (compute_mov_length): Likewise.
17260 (output_logical_op): Likewise.
17261 (compute_logical_op_length): Likewise.
17262 (compute_logical_op_cc): Likewise.
17263 (h8300_shift_needs_scratch_p): Likewise.
17264 (output_a_shift): Likewise.
17265 (compute_a_shift_length): Likewise.
17266 (compute_a_shift_cc): Likewise.
17267 (expand_a_rotate): Likewise.
17268 (output_a_rotate): Likewise.
17269 * config/i386/i386.c (classify_argument): Likewise.
17270 (function_arg_advance_32): Likewise.
17271 (function_arg_32): Likewise.
17272 (function_arg_64): Likewise.
17273 (function_value_64): Likewise.
17274 (ix86_gimplify_va_arg): Likewise.
17275 (ix86_legitimate_constant_p): Likewise.
17276 (put_condition_code): Likewise.
17277 (split_double_mode): Likewise.
17278 (ix86_avx256_split_vector_move_misalign): Likewise.
17279 (ix86_expand_vector_logical_operator): Likewise.
17280 (ix86_split_idivmod): Likewise.
17281 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
17282 (ix86_build_const_vector): Likewise.
17283 (ix86_build_signbit_mask): Likewise.
17284 (ix86_match_ccmode): Likewise.
17285 (ix86_cc_modes_compatible): Likewise.
17286 (ix86_expand_branch): Likewise.
17287 (ix86_expand_sse_cmp): Likewise.
17288 (ix86_expand_sse_movcc): Likewise.
17289 (ix86_expand_int_sse_cmp): Likewise.
17290 (ix86_expand_vec_perm_vpermi2): Likewise.
17291 (ix86_expand_vec_perm): Likewise.
17292 (ix86_expand_sse_unpack): Likewise.
17293 (ix86_expand_int_addcc): Likewise.
17294 (ix86_split_to_parts): Likewise.
17295 (ix86_vectorize_builtin_gather): Likewise.
17296 (ix86_vectorize_builtin_scatter): Likewise.
17297 (avx_vpermilp_parallel): Likewise.
17298 (inline_memory_move_cost): Likewise.
17299 (ix86_tieable_integer_mode_p): Likewise.
17300 (x86_maybe_negate_const_int): Likewise.
17301 (ix86_expand_vector_init_duplicate): Likewise.
17302 (ix86_expand_vector_init_one_nonzero): Likewise.
17303 (ix86_expand_vector_init_one_var): Likewise.
17304 (ix86_expand_vector_init_concat): Likewise.
17305 (ix86_expand_vector_init_interleave): Likewise.
17306 (ix86_expand_vector_init_general): Likewise.
17307 (ix86_expand_vector_set): Likewise.
17308 (ix86_expand_vector_extract): Likewise.
17309 (emit_reduc_half): Likewise.
17310 (ix86_emit_i387_round): Likewise.
17311 (ix86_mangle_type): Likewise.
17312 (ix86_expand_round_sse4): Likewise.
17313 (expand_vec_perm_blend): Likewise.
17314 (canonicalize_vector_int_perm): Likewise.
17315 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
17316 (expand_vec_perm_1): Likewise.
17317 (expand_vec_perm_interleave3): Likewise.
17318 (expand_vec_perm_even_odd_pack): Likewise.
17319 (expand_vec_perm_even_odd_1): Likewise.
17320 (expand_vec_perm_broadcast_1): Likewise.
17321 (ix86_vectorize_vec_perm_const_ok): Likewise.
17322 (ix86_expand_vecop_qihi): Likewise.
17323 (ix86_expand_mul_widen_hilo): Likewise.
17324 (ix86_expand_sse2_abs): Likewise.
17325 (ix86_expand_pextr): Likewise.
17326 (ix86_expand_pinsr): Likewise.
17327 (ix86_preferred_simd_mode): Likewise.
17328 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
17329 * config/i386/sse.md (*andnot<mode>3): Likewise.
17330 (<mask_codefor><code><mode>3<mask_name>): Likewise.
17331 (*<code><mode>3): Likewise.
17332 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
17333 (ia64_expand_atomic_op): Likewise.
17334 (ia64_arg_type): Likewise.
17335 (ia64_mode_to_int): Likewise.
17336 (ia64_scalar_mode_supported_p): Likewise.
17337 (ia64_vector_mode_supported_p): Likewise.
17338 (expand_vec_perm_broadcast): Likewise.
17339 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
17340 (iq2000_function_arg_advance): Likewise.
17341 (iq2000_function_arg): Likewise.
17342 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
17343 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
17344 (m68k_libcall_value): Likewise.
17345 (m68k_function_value): Likewise.
17346 (sched_attr_op_type): Likewise.
17347 * config/mcore/mcore.c (mcore_output_move): Likewise.
17348 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
17350 (microblaze_function_arg): Likewise.
17351 * config/mips/mips.c (mips16_build_call_stub): Likewise.
17352 (mips_print_operand): Likewise.
17353 (mips_mode_ok_for_mov_fmt_p): Likewise.
17354 (mips_vector_mode_supported_p): Likewise.
17355 (mips_preferred_simd_mode): Likewise.
17356 (mips_expand_vpc_loongson_even_odd): Likewise.
17357 (mips_expand_vec_unpack): Likewise.
17358 (mips_expand_vi_broadcast): Likewise.
17359 (mips_expand_vector_init): Likewise.
17360 (mips_expand_vec_reduc): Likewise.
17361 (mips_expand_msa_cmp): Likewise.
17362 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
17363 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
17364 (cc_flags_for_mode): Likewise.
17365 * config/msp430/msp430.c (msp430_print_operand): Likewise.
17366 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
17367 (nds32_output_casesi_pc_relative): Likewise.
17368 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17369 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
17370 (nvptx_gen_unpack): Likewise.
17371 (nvptx_gen_pack): Likewise.
17372 (nvptx_gen_shuffle): Likewise.
17373 (nvptx_gen_wcast): Likewise.
17374 (nvptx_preferred_simd_mode): Likewise.
17375 * config/pa/pa.c (pa_secondary_reload): Likewise.
17376 * config/pa/predicates.md (base14_operand): Likewise.
17377 * config/powerpcspe/powerpcspe-c.c
17378 (altivec_resolve_overloaded_builtin): Likewise.
17379 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
17381 (rs6000_preferred_simd_mode): Likewise.
17382 (output_vec_const_move): Likewise.
17383 (rs6000_expand_vector_extract): Likewise.
17384 (rs6000_split_vec_extract_var): Likewise.
17385 (reg_offset_addressing_ok_p): Likewise.
17386 (rs6000_legitimate_offset_address_p): Likewise.
17387 (rs6000_legitimize_address): Likewise.
17388 (rs6000_emit_set_const): Likewise.
17389 (rs6000_const_vec): Likewise.
17390 (rs6000_emit_move): Likewise.
17391 (spe_build_register_parallel): Likewise.
17392 (rs6000_darwin64_record_arg_recurse): Likewise.
17393 (swap_selector_for_mode): Likewise.
17394 (spe_init_builtins): Likewise.
17395 (paired_init_builtins): Likewise.
17396 (altivec_init_builtins): Likewise.
17397 (do_load_for_compare): Likewise.
17398 (rs6000_generate_compare): Likewise.
17399 (rs6000_expand_float128_convert): Likewise.
17400 (emit_load_locked): Likewise.
17401 (emit_store_conditional): Likewise.
17402 (rs6000_output_function_epilogue): Likewise.
17403 (rs6000_handle_altivec_attribute): Likewise.
17404 (rs6000_function_value): Likewise.
17405 (emit_fusion_gpr_load): Likewise.
17406 (emit_fusion_p9_load): Likewise.
17407 (emit_fusion_p9_store): Likewise.
17408 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
17409 (fusion_gpr_mem_load): Likewise.
17410 (fusion_addis_mem_combo_load): Likewise.
17411 (fusion_addis_mem_combo_store): Likewise.
17412 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
17413 (fusion_gpr_mem_load): Likewise.
17414 (fusion_addis_mem_combo_load): Likewise.
17415 (fusion_addis_mem_combo_store): Likewise.
17416 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17418 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
17419 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
17420 (rs6000_preferred_simd_mode): Likewise.
17421 (output_vec_const_move): Likewise.
17422 (rs6000_expand_vector_extract): Likewise.
17423 (rs6000_split_vec_extract_var): Likewise.
17424 (reg_offset_addressing_ok_p): Likewise.
17425 (rs6000_legitimate_offset_address_p): Likewise.
17426 (rs6000_legitimize_address): Likewise.
17427 (rs6000_emit_set_const): Likewise.
17428 (rs6000_const_vec): Likewise.
17429 (rs6000_emit_move): Likewise.
17430 (rs6000_darwin64_record_arg_recurse): Likewise.
17431 (swap_selector_for_mode): Likewise.
17432 (paired_init_builtins): Likewise.
17433 (altivec_init_builtins): Likewise.
17434 (rs6000_expand_float128_convert): Likewise.
17435 (emit_load_locked): Likewise.
17436 (emit_store_conditional): Likewise.
17437 (rs6000_output_function_epilogue): Likewise.
17438 (rs6000_handle_altivec_attribute): Likewise.
17439 (rs6000_function_value): Likewise.
17440 (emit_fusion_gpr_load): Likewise.
17441 (emit_fusion_p9_load): Likewise.
17442 (emit_fusion_p9_store): Likewise.
17443 * config/rx/rx.c (rx_gen_move_template): Likewise.
17444 (flags_from_mode): Likewise.
17445 * config/s390/predicates.md (s390_alc_comparison): Likewise.
17446 (s390_slb_comparison): Likewise.
17447 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
17448 (s390_vector_mode_supported_p): Likewise.
17449 (s390_cc_modes_compatible): Likewise.
17450 (s390_match_ccmode_set): Likewise.
17451 (s390_canonicalize_comparison): Likewise.
17452 (s390_emit_compare_and_swap): Likewise.
17453 (s390_branch_condition_mask): Likewise.
17454 (s390_rtx_costs): Likewise.
17455 (s390_secondary_reload): Likewise.
17456 (__SECONDARY_RELOAD_CASE): Likewise.
17457 (s390_expand_cs): Likewise.
17458 (s390_preferred_simd_mode): Likewise.
17459 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
17460 * config/sh/sh.c (sh_print_operand): Likewise.
17461 (dump_table): Likewise.
17462 (sh_secondary_reload): Likewise.
17463 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17464 * config/sh/sh.md (casesi_worker_1): Likewise.
17465 (casesi_worker_2): Likewise.
17466 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
17467 (fcc_comparison_operator): Likewise.
17468 * config/sparc/sparc.c (sparc_expand_move): Likewise.
17469 (emit_soft_tfmode_cvt): Likewise.
17470 (sparc_preferred_simd_mode): Likewise.
17471 (output_cbranch): Likewise.
17472 (sparc_print_operand): Likewise.
17473 (sparc_expand_vec_perm_bmask): Likewise.
17474 (vector_init_bshuffle): Likewise.
17475 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
17476 (spu_vector_mode_supported_p): Likewise.
17477 (spu_expand_insv): Likewise.
17478 (spu_emit_branch_or_set): Likewise.
17479 (spu_handle_vector_attribute): Likewise.
17480 (spu_builtin_splats): Likewise.
17481 (spu_builtin_extract): Likewise.
17482 (spu_builtin_promote): Likewise.
17483 (spu_expand_sign_extend): Likewise.
17484 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
17485 (tilegx_simd_int): Likewise.
17486 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
17487 (tilepro_simd_int): Likewise.
17488 * config/v850/v850.c (const_double_split): Likewise.
17489 (v850_print_operand): Likewise.
17490 (ep_memory_offset): Likewise.
17491 * config/vax/vax.c (vax_rtx_costs): Likewise.
17492 (vax_output_int_move): Likewise.
17493 (vax_output_int_add): Likewise.
17494 (vax_output_int_subtract): Likewise.
17495 * config/visium/predicates.md (visium_branch_operator): Likewise.
17496 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
17497 (visium_print_operand_address): Likewise.
17498 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17499 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
17500 (xtensa_expand_conditional_branch): Likewise.
17501 (xtensa_copy_incoming_a7): Likewise.
17502 (xtensa_output_literal): Likewise.
17503 * dfp.c (decimal_real_maxval): Likewise.
17504 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
17506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17507 Alan Hayward <alan.hayward@arm.com>
17508 David Sherwood <david.sherwood@arm.com>
17510 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
17511 (mode_nunits_inline): Likewise.
17512 (mode_inner_inline): Likewise.
17513 (mode_unit_size_inline): Likewise.
17514 (mode_unit_precision_inline): Likewise.
17515 (emit_insn_modes_h): Likewise. Also emit a #define of the
17517 (emit_mode_wider): Add an E_ prefix to mode names.
17518 (emit_mode_complex): Likewise.
17519 (emit_mode_inner): Likewise.
17520 (emit_mode_adjustments): Likewise.
17521 (emit_mode_int_n): Likewise.
17522 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
17523 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
17524 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
17525 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
17526 (CRC32_BUILTIN, ENTRY): Likewise.
17527 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
17528 (aarch64_pop_regs): Likewise.
17529 (aarch64_process_components): Likewise.
17530 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
17531 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
17532 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
17533 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
17534 * config/arm/arm.c (arm_init_libfuncs): Likewise.
17535 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
17537 * config/i386/i386-builtin.def (pcmpestr): Likewise.
17538 (pcmpistr): Likewise.
17539 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
17540 * config/mmix/mmix.c (mmix_output_condition): Likewise.
17541 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17543 * config/rl78/rl78.c (mduc_regs): Likewise.
17544 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17545 (htm_expand_builtin): Likewise.
17546 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
17547 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
17548 * config/xtensa/xtensa.c (print_operand): Likewise.
17549 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
17550 (NUM_MODE_VECTOR_INT): Likewise.
17551 * genoutput.c (null_operand): Likewise.
17552 (output_operand_data): Likewise.
17553 * genrecog.c (print_parameter_value): Likewise.
17554 * lra.c (debug_operand_data): Likewise.
17556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17558 * dwarf2out.c (scompare_loc_descriptor_wide)
17559 (scompare_loc_descriptor_narrow): New functions, split out from...
17560 (scompare_loc_descriptor): ...here.
17561 * expmed.c (emit_store_flag_int): New function, split out from...
17562 (emit_store_flag): ...here.
17564 2017-08-30 Richard Biener <rguenther@suse.de>
17566 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
17567 (dwarf2out_early_finish): Move setting of AT_pubnames from
17568 early debug output to early finish.
17570 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17572 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
17573 and -mdata-region to the assembler.
17575 2017-08-30 Richard Biener <rguenther@suse.de>
17577 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
17579 (gen_subprogram_die): Add DW_AT_object_pointer only early.
17580 (dwarf2out_early_global_decl): Only generate a DIE for the
17581 abstract origin if it doesn't already exist or is a declaration DIE.
17582 (resolve_addr): Do not add the linkage name twice when
17583 generating a stub DIE for the DW_TAG_GNU_call_site target.
17585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
17587 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17588 Use machine_mode rather than int for arg1_mode.
17590 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
17593 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
17594 that the second argument of the built-in functions to unpack
17595 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
17596 switch statement instead a lot of if statements.
17597 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
17598 Allow 64-bit values to be in Altivec registers as well as
17599 traditional floating point registers.
17600 (pack<mode>, FMOVE128_VSX iterator): Likewise.
17602 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
17604 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
17605 MAX_REGS_PER_ADDRESS == 1.
17607 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
17609 * config/i386/i386.opt (flag_fentry): Do not init to -1.
17610 * config/i386/i386.c (ix86_option_override_internal): Simplify
17611 setting of opts->x_flag_entry.
17613 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17614 Jakub Jelinek <jakub@redhat.com>
17615 Richard Biener <rguenther@suse.de>
17617 PR tree-optimization/81503
17618 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
17619 folded constant fits in the target type; reorder tests for clarity.
17621 2017-08-29 Martin Liska <mliska@suse.cz>
17623 * passes.def: Include pass_lower_switch.
17624 * stmt.c (dump_case_nodes): Remove and move to
17625 tree-switch-conversion.
17626 (case_values_threshold): Likewise.
17627 (expand_switch_as_decision_tree_p): Likewise.
17628 (emit_case_decision_tree): Likewise.
17629 (expand_case): Likewise.
17630 (balance_case_nodes): Likewise.
17631 (node_has_low_bound): Likewise.
17632 (node_has_high_bound): Likewise.
17633 (node_is_bounded): Likewise.
17634 (emit_case_nodes): Likewise.
17635 (struct simple_case_node): New struct.
17636 (add_case_node): Remove.
17637 (emit_case_dispatch_table): Use vector instead of case_list.
17638 (reset_out_edges_aux): Remove.
17639 (compute_cases_per_edge): Likewise.
17640 (expand_case): Build list of simple_case_node.
17641 (expand_sjlj_dispatch_table): Use it.
17642 * tree-switch-conversion.c (struct case_node): Moved from
17643 stmt.c and adjusted.
17644 (emit_case_nodes): Likewise.
17645 (node_has_low_bound): Likewise.
17646 (node_has_high_bound): Likewise.
17647 (node_is_bounded): Likewise.
17648 (case_values_threshold): Likewise.
17649 (reset_out_edges_aux): Likewise.
17650 (compute_cases_per_edge): Likewise.
17651 (add_case_node): Likewise.
17652 (dump_case_nodes): Likewise.
17653 (balance_case_nodes): Likewise.
17654 (expand_switch_as_decision_tree_p): Likewise.
17655 (emit_jump): Likewise.
17656 (emit_case_decision_tree): Likewise.
17657 (try_switch_expansion): Likewise.
17658 (do_jump_if_equal): Likewise.
17659 (emit_cmp_and_jump_insns): Likewise.
17660 (fix_phi_operands_for_edge): New function.
17661 (record_phi_operand_mapping): Likewise.
17662 (class pass_lower_switch): New pass.
17663 (pass_lower_switch::execute): New function.
17664 (make_pass_lower_switch): Likewise.
17665 (conditional_probability):
17666 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
17667 * tree-pass.h: Add make_pass_lower_switch.
17669 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17672 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
17675 2017-08-29 Richard Biener <rguenther@suse.de>
17677 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
17678 we do not add a DW_AT_inline attribute twice.
17679 (gen_subprogram_die): Remove code setting DW_AT_inline on
17680 DECL_ABSTRACT_P nodes.
17682 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
17684 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
17685 calls to internal functions.
17686 (gimplify_modify_expr): Likewise.
17687 * tree-call-cdce.c (use_internal_fn): Likewise.
17688 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
17689 (convert_to_divmod): Set the nothrow flag.
17690 * tree-if-conv.c (predicate_mem_writes): Likewise.
17691 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17692 (vectorizable_call): Likewise.
17693 (vectorizable_store): Likewise.
17694 (vectorizable_load): Likewise.
17695 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
17696 (vect_recog_mask_conversion_pattern): Likewise.
17698 2017-08-29 Martin Liska <mliska@suse.cz>
17701 * gcc.c (driver_handle_option): Add new argument.
17702 * opts-common.c (handle_option): Pass
17703 target_option_override_hook.
17704 * opts-global.c (lang_handle_option): Add new option.
17705 (set_default_handlers): Add new argument.
17706 (decode_options): Likewise.
17707 * opts.c (target_handle_option): Likewise.
17708 (common_handle_option): Call target_option_override_hook.
17709 * opts.h (struct cl_option_handler_func): Add hook for
17710 target option override.
17711 (struct cl_option_handlers): Likewise.
17712 (set_default_handlers): Add new argument.
17713 (decode_options): Likewise.
17714 (common_handle_option): Likewise.
17715 (target_handle_option): Likewise.
17716 * toplev.c (toplev::main): Pass targetm.target_option.override
17719 2017-08-29 Richard Biener <rguenther@suse.de>
17720 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
17722 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
17723 life to the active subtree.
17725 2017-08-28 Jeff Law <law@redhat.com>
17727 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
17728 derive_equivalences.
17729 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
17730 Code moved into....
17731 (edge_info::derive_equivalences): New private member function
17733 * tree-ssa-dom.c (class edge_info): Changed from a struct
17734 to a class. Add ctor/dtor, methods and data members.
17735 (edge_info::edge_info): Renamed from allocate_edge_info.
17736 Initialize additional members.
17737 (edge_info::~edge_info): New.
17738 (free_dom_edge_info): Delete the edge info.
17739 (record_edge_info): Use new class & associated member functions.
17740 Tighten forms for testing for edge equivalences.
17741 (record_temporary_equivalences): Iterate over the simple
17742 equivalences rather than assuming there's only one per edge.
17743 (cprop_into_successor_phis): Iterate over the simple
17744 equivalences rather than assuming there's only one per edge.
17745 (optimize_stmt): Use operand_equal_p rather than pointer
17746 equality for mini-DSE code.
17748 2017-08-28 Nathan Sidwell <nathan@acm.org>
17750 * gcc.c (execute): Fold SIGPIPE handling into switch
17751 statement. Adjust internal error message.
17753 2017-08-28 Richard Biener <rguenther@suse.de>
17756 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
17757 Do nothing for removed DIEs.
17759 2017-08-28 Richard Biener <rguenther@suse.de>
17761 PR tree-optimization/81977
17762 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
17765 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
17768 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
17769 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
17770 using targetm.gen_mem_thread_fence.
17772 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
17775 * config/i386/i386.md (*<btsc><mode>): Change operand 2
17776 predicate to register_operand. Reorder operands.
17777 (*btr<mode>): Ditto.
17778 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
17779 (*btr<mode>_mask): Ditto.
17781 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
17783 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
17784 * config/rs6000/xmmintrin.h: New file.
17785 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
17787 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17790 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
17791 parameter and_insn and return it.
17792 (recombine_lvx_pattern): Insert a copy to ensure availability of
17793 the base register of the copied masking operation at the point of
17794 the instruction replacement.
17795 (recombine_stvx_pattern): Likewise.
17797 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
17799 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
17800 undocumented switches.
17801 (-mpower9-dform-vector): Likewise.
17802 (-mpower9-dform): Likewise.
17803 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
17804 comments to delete references to -mpower9-dform* switches.
17805 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
17806 Delete reference to -mpower9-dform* switches, test for
17807 -mpower9-vector instead.
17808 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
17809 (OTHER_P9_VECTOR_MASKS): Likewise.
17810 (POWERPC_MASKS): Likewise.
17811 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
17812 tests against -mpower9-dform* to -mpower9-vector. Delete code
17813 that checked for -mpower9-dform* consistancy with other options.
17814 Add test for -mpower9-misc to enable other power9 switches.
17815 (rs6000_init_hard_regno_mode_ok): Likewise.
17816 (rs6000_option_override_internal): Likewise.
17817 (rs6000_emit_prologue): Likewise.
17818 (rs6000_emit_epilogue): Likewise.
17819 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
17820 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
17821 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
17823 (emit_fusion_p9_store): Likewise.
17824 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
17825 resetting these macros if the assembler does not support ISA 3.0
17827 (TARGET_P9_DFORM_VECTOR): Likewise.
17828 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
17829 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
17831 2017-08-25 Alan Modra <amodra@gmail.com>
17833 PR rtl-optimization/81747
17834 * cse.c (cse_extended_basic_block): Don't attempt to record
17835 equivalences for degenerate conditional branches that branch
17836 to their fall-through.
17838 2017-08-24 Martin Sebor <msebor@redhat.com>
17840 PR middle-end/81908
17841 * gimple-fold.c (size_must_be_zero_p): New function.
17842 (gimple_fold_builtin_memory_op): Call it.
17844 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
17846 * config/rs6000/mm_malloc.h: New file.
17848 2017-08-24 Bin Cheng <bin.cheng@arm.com>
17850 PR tree-optimization/81913
17851 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
17852 analysis when either IVs in condition can wrap.
17854 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
17856 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
17857 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
17859 2017-08-24 Richard Biener <rguenther@suse.de>
17862 * targhooks.c (default_target_can_inline_p): Properly
17863 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
17864 is present and always compare.
17865 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
17866 infer -mfpmath=sse from TARGET_SSE_P.
17867 (ix86_can_inline_p): Properly use target_option_default_node when
17868 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
17870 2017-08-24 Richard Biener <rguenther@suse.de>
17873 * dwarf2out.c (output_die): Handle flag_generate_offload like
17875 (output_comp_unit): Likewise.
17876 (gen_array_type_die): Likewise.
17877 (dwarf2out_early_finish): Likewise.
17878 (note_variable_value_in_expr): Likewise.
17879 (dwarf2out_finish): Likewise. Adjust assert.
17880 * cgraphunit.c (symbol_table::compile): Move setting of
17881 flag_generate_offload earlier ...
17882 (symbol_table::finalize_compilation_unit): ... here, before
17883 early debug finalization.
17885 2017-08-24 Richard Biener <rguenther@suse.de>
17887 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
17888 and ipa-fnsummary.h.
17889 (ix86_can_inline_p): When ix86_fpmath flags do not match
17890 check whether the callee uses FP math at all.
17892 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
17894 PR middle-end/81931
17895 * tree-ssanames.c (get_nonzero_bits): Use element_precision
17896 instead of TYPE_PRECISION.
17898 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
17899 Alan Hayward <alan.hayward@arm.com>
17900 David Sherwood <david.sherwood@arm.com>
17902 * combine.c (make_extraction): Use subreg_offset_from_lsb.
17904 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
17906 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
17908 (ix86_frame::stack_realign_allocate): New field.
17909 (struct machine_frame_state): Modify comments.
17910 (machine_frame_state::sp_realigned_fp_end): New field.
17911 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
17912 layout calculation.
17913 (sp_valid_at): Add assertion to assure no attempt to access invalid
17914 offset of a realigned stack.
17915 (fp_valid_at): Likewise.
17916 (choose_baseaddr): Modify comments.
17917 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
17918 ix86_expand_prologue.
17919 (ix86_expand_prologue): Modify stack realignment and allocation.
17920 (ix86_expand_epilogue): Modify comments.
17921 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
17922 avx2_runtime, avx512f, and avx512f_runtime.
17924 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
17926 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
17927 (mstackrealign): Do not init to -1.
17928 * config/i386/i386.c (ix86_option_override_internal):
17929 Check opts_set, not opts when setting default value of
17930 opts->x_ix86_force_align_arg_pointer.
17932 2017-08-23 Richard Biener <rguenther@suse.de>
17934 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
17935 lang_hooks.decl_printable_name.
17936 * print-rtl-function.c (print_rtx_function): Likewise.
17937 * tree-pretty-print.c (dump_function_header): Likewise.
17939 2017-08-23 Richard Biener <rguenther@suse.de>
17942 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
17943 -g0 at compile-time.
17945 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17947 PR middle-end/19706
17948 * doc/sourcebuild.texi (Other hardware attributes):
17951 2017-08-23 Tamar Christina <tamar.christina@arm.com>
17953 PR middle-end/19706
17954 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
17955 Fix single-use check.
17957 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
17959 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
17961 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17963 * doc/install.texi: Modify to add more details on running selected
17966 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
17968 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
17969 is combined with -mabi=ms.
17970 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
17973 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17975 PR tree-optimization/81488
17976 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
17977 and cached_basis fields.
17978 (MAX_SPREAD): New constant.
17979 (alloc_cand_and_find_basis): Initialize new fields.
17980 (clear_visited): New function.
17981 (create_phi_basis_1): Rename from create_phi_basis, set visited
17982 and cached_basis fields.
17983 (create_phi_basis): New wrapper function.
17984 (phi_add_costs_1): Rename from phi_add_costs, add spread
17985 parameter, set visited field, short-circuit when limits reached.
17986 (phi_add_costs): New wrapper function.
17987 (record_phi_increments_1): Rename from record_phi_increments, set
17989 (record_phi_increments): New wrapper function.
17990 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
17991 (phi_incr_cost): New wrapper function.
17992 (all_phi_incrs_profitable_1): Rename from
17993 all_phi_incrs_profitable, set visited field.
17994 (all_phi_incrs_profitable): New wrapper function.
17996 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
17997 Alan Hayward <alan.hayward@arm.com>
17998 David Sherwood <david.sherwood@arm.com>
18000 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
18001 that takes the outer and inner modes.
18002 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
18003 comparison as the canonical test for a paradoxical subreg.
18004 * combine.c (simplify_set): Use paradoxical_subreg_p.
18005 (make_extraction): Likewise.
18006 (force_to_mode): Likewise.
18007 (rtx_equal_for_field_assignment_p): Likewise.
18008 (gen_lowpart_for_combine): Likewise.
18009 (simplify_comparison): Likewise.
18010 * cse.c (equiv_constant): Likewise.
18011 * expmed.c (store_bit_field_1): Likewise.
18012 * final.c (alter_subreg): Likewise.
18013 * fwprop.c (propagate_rtx): Likewise.
18014 (forward_propagate_subreg): Likewise.
18015 * ira-conflicts.c (ira_build_conflicts): Likewise.
18016 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
18017 * lra-constraints.c (curr_insn_transform): Likewise.
18018 (split_reg): Likewise.
18019 * lra-eliminations.c (move_plus_up): Likewise.
18020 (lra_eliminate_regs_1): Likewise.
18021 * recog.c (general_operand): Likewise.
18022 * ree.c (combine_reaching_defs): Likewise.
18023 * reload.c (push_reload): Likewise.
18024 (find_reloads): Likewise.
18025 * reload1.c (elimination_effects): Likewise.
18026 (compute_reload_subreg_offset): Likewise.
18027 (choose_reload_regs): Likewise.
18028 * rtlanal.c (subreg_lsb_1): Likewise.
18029 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18030 (simplify_subreg): Likewise.
18031 * var-tracking.c (track_loc_p): Likewise.
18032 * emit-rtl.c (byte_lowpart_offset): Likewise.
18033 (paradoxical_subreg_p): Delete out-of-line definition.
18035 2017-08-22 Jeff Law <law@redhat.com>
18037 PR tree-optimization/81741
18038 PR tree-optimization/71947
18039 * tree-ssa-dom.c: Include tree-inline.h.
18040 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
18041 equivalences if one is more expensive to compute than the other.
18042 * tree-ssa-scopedtables.h (class const_or_copies): Make
18043 record_const_or_copy_raw method private.
18044 (class avail_exprs_stack): New method simplify_binary_operation.
18045 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
18046 avail_exprs_stack::simplify_binary_operation as needed.
18047 (avail_exprs_stack::simplify_binary_operation): New function.
18049 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
18051 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
18052 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
18053 (DOT_SYMBOLS): Likewise.
18054 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
18055 (RELOCATABLE_NEEDS_FIXUP): Likewise.
18056 (RS6000_ABI_NAME): Likewise.
18057 (TARGET_CMODEL): Likewise.
18058 (TOC_SECTION_ASM_OP): Likewise.
18059 (SET_CMODEL): New macro.
18060 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
18062 2017-08-22 Richard Biener <rguenther@suse.de>
18064 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
18065 to allow for free-lang-data replacements similar to verify_type_variant.
18067 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
18068 Alan Hayward <alan.hayward@arm.com>
18069 David Sherwood <david.sherwood@arm.com>
18071 * config/aarch64/aarch64.md (casesi): Use DImode rather than
18072 VOIDmode for the LABEL_REF.
18074 2017-08-22 Richard Biener <rguenther@suse.de>
18076 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
18078 2017-08-22 Richard Biener <rguenther@suse.de>
18080 * common.opt (feliminate-dwarf2-dups): Ignore.
18081 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
18082 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
18083 same_die_p_wrap, compute_section_prefix,
18084 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
18085 (comdat_symbol_id, comdat_symbol_number): Likewise.
18086 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
18088 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
18089 (output_die): Mark unreachable path unreachable.
18090 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
18091 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
18092 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
18093 (dwarf2out_early_finish): Likewise.
18095 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
18097 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
18099 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
18102 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
18103 not VAR_P. Filter attribute warnings with OPT_Wattributes.
18104 (avr_attribute_table) <io, io_low, address>: Initialize
18105 .decl_required with true.
18107 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
18109 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
18110 undocumented debugging options.
18111 (-mvsx-scalar-double): Likewise.
18112 (-mallow-df-permute): Likewise.
18113 (-mvectorize-builtins): Likewise.
18114 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18115 (rs6000_builtin_vectorized_function): Likewise.
18116 (rs6000_builtin_md_vectorized_function): Likewise.
18117 (rs6000_opt_vars): Likewise.
18119 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
18122 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
18123 (*btrq_imm): Rename from *btrq.
18124 (*btcq_imm): Rename from *btcq.
18125 (btsc): New code attribute.
18126 (*<btsc><mode>): New insn pattern.
18127 (*btr<mode>): Ditto.
18128 (*<btsc><mode>_mask): New insn_and_split pattern.
18129 (*btr<mode>_mask): Ditto.
18131 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18132 Alan Hayward <alan.hayward@arm.com>
18133 David Sherwood <david.sherwood@arm.com>
18135 * function.c (pad_below): Simplify padding calculation.
18137 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18138 Alan Hayward <alan.hayward@arm.com>
18139 David Sherwood <david.sherwood@arm.com>
18141 * target.def (function_prologue): Remove frame size argument.
18142 (function_epilogue): Likewise.
18143 * doc/tm.texi: Regenerate.
18144 * final.c (final_start_function): Update call to function_prologue.
18145 (final_end_function): Update call to function_epilogue.
18146 (default_function_pro_epilogue): Remove frame size argument.
18147 * output.h (default_function_pro_epilogue): Likewise.
18148 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
18149 (arm_output_function_prologue): Likewise.
18150 * config/frv/frv.c (frv_function_prologue): Likewise.
18151 (frv_function_epilogue): Likewise.
18152 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
18153 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
18154 (ia64_output_function_epilogue): Likewise.
18155 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
18156 (m32r_output_function_epilogue): Likewise.
18157 * config/microblaze/microblaze.c (microblaze_function_prologue)
18158 (microblaze_function_epilogue): Likewise.
18159 * config/mips/mips.c (mips_output_function_prologue): Likewise.
18160 (mips_output_function_epilogue): Likewise.
18161 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
18162 (mmix_target_asm_function_epilogue): Likewise.
18163 * config/msp430/msp430.c (msp430_start_function): Likewise.
18164 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
18165 (nds32_asm_function_epilogue): Likewise.
18166 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
18167 * config/pa/pa.c (pa_output_function_prologue): Likewise.
18168 (pa_output_function_epilogue): Likewise.
18169 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
18170 (rs6000_output_function_epilogue): Likewise.
18171 * config/rl78/rl78.c (rl78_start_function): Likewise.
18172 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
18173 (rs6000_output_function_epilogue): Likewise.
18174 * config/rx/rx.c (rx_output_function_prologue): Likewise.
18175 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
18176 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
18177 (sparc_asm_function_epilogue): Likewise.
18179 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18181 * tree.h (type_has_mode_precision_p): New function.
18182 * convert.c (convert_to_integer_1): Use it.
18183 * expr.c (expand_expr_real_2): Likewise.
18184 (expand_expr_real_1): Likewise.
18185 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
18186 * match.pd: Likewise.
18187 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
18188 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
18189 * tree-tailcall.c (process_assignment): Likewise.
18190 * tree-vect-loop.c (vectorizable_reduction): Likewise.
18191 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
18192 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
18193 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18194 (vectorizable_assignment): Likewise.
18195 (vectorizable_shift): Likewise.
18196 (vectorizable_operation): Likewise.
18197 * tree-vrp.c (register_edge_assert_for_2): Likewise.
18199 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
18201 * match.pd: Add pow (C, x) simplification.
18203 2017-08-21 Richard Biener <rguenther@suse.de>
18205 PR tree-optimization/81900
18206 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
18207 for blocks with abnormal predecessors.
18208 (compute_antic): Do not set visited flag prematurely.
18210 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
18213 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
18215 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18217 * stor-layout.h (vector_type_mode): Move to...
18218 * tree.h (vector_type_mode): ...here.
18219 * stor-layout.c (vector_type_mode): Move to...
18220 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
18222 2017-08-21 Richard Biener <rguenther@suse.de>
18224 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
18225 register_external_die hooks.
18226 (debug_false_tree_charstarstar_uhwistar): Declare.
18227 (debug_nothing_tree_charstar_uhwi): Likewise.
18228 * debug.c (do_nothing_debug_hooks): Adjust.
18229 (debug_false_tree_charstarstar_uhwistar): New do nothing.
18230 (debug_nothing_tree_charstar_uhwi): Likewise.
18231 * dbxout.c (dbx_debug_hooks): Adjust.
18232 (xcoff_debug_hooks): Likewise.
18233 * sdbout.c (sdb_debug_hooks): Likewise.
18234 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
18235 * dwarf2out.c (macinfo_label_base): New global.
18236 (dwarf2out_register_external_die): New function for the
18237 register_external_die hook.
18238 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
18239 (dwarf2_debug_hooks): Use them.
18240 (dwarf2_lineno_debug_hooks): Adjust.
18241 (struct die_struct): Add with_offset flag.
18242 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
18243 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
18244 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
18245 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
18246 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
18247 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
18248 defining section names for the early LTO debug variants.
18249 (reset_indirect_string): New helper.
18250 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
18251 (print_dw_val): Add support for offsetted symbol references.
18252 (get_ultimate_context): Split out from is_cxx.
18253 (is_cxx): Use get_ultimate_context.
18254 (is_fortran): Add decl overload.
18255 (compute_comp_unit_symbol): Split out worker from
18256 compute_section_prefix.
18257 (compute_section_prefix): Call compute_comp_unit_symbol and
18258 set comdat_type_p here.
18259 (output_die): Skip DIE symbol output for the LTO added one.
18260 Handle DIE symbol references with offset.
18261 (output_comp_unit): Guard section name mangling properly.
18262 For LTO debug sections emit a symbol at the section beginning
18263 which we use to refer to its DIEs.
18264 (add_abstract_origin_attribute): For DIEs registered via
18265 dwarf2out_register_external_die directly refer to the early
18266 DIE rather than indirectly through the shadow one we created.
18267 Remove obsolete call to dwarf2out_abstract_function for
18268 non-function/block origins.
18269 (gen_array_type_die): When generating early LTO debug do
18270 not emit DW_AT_string_length.
18271 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
18272 late when in LTO. As suggested place a gcc_unreachable for
18273 the DECL_ABSTRACT_P case.
18274 (gen_subprogram_die): Avoid another specification DIE
18275 for early built declarations/definitions for the late LTO case.
18276 (gen_variable_die): Add type references for late duplicated VLA dies
18278 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
18279 we have the abstract instance already.
18280 (process_scope_var): Adjust decl DIE contexts in LTO which
18281 first puts them in limbo.
18282 (gen_decl_die): Do not generate type DIEs late apart from
18283 types for VLAs or for decls we do not yet have a DIE. Do not
18284 call dwarf2out_abstract_function late.
18285 (dwarf2out_early_global_decl): Make sure to create DIEs
18286 for abstract instances of a decl first.
18287 (dwarf2out_late_global_decl): Adjust comment.
18288 (output_macinfo_op): With multiple macro sections use
18289 macinfo_label_base to distinguish labels.
18290 (output_macinfo): Likewise. Update macinfo_label_base.
18291 Pass in the line info label.
18292 (note_variable_value_in_expr): When generating LTO resolve
18293 all variable values here by generating DIEs as needed.
18294 (init_sections_and_labels): Add early LTO debug flag parameter
18295 and generate different sections and names if set. Add generation
18296 counter for the labels so we can have multiple of them.
18297 (reset_dies): Helper to allow DIEs to be output multiple times.
18298 (dwarf2out_finish): When outputting DIEs to the fat part of an
18299 LTO object first reset DIEs.
18300 (dwarf2out_early_finish): Output early DIEs when generating LTO.
18301 (modified_type_die): Check for decl_ultimate_origin being self
18303 (gen_type_die_with_usage): Likewise.
18304 (gen_typedef_die): Allow decl_ultimate_origin being self.
18305 (set_decl_abstract_flags): Remove.
18306 (set_block_abstract_flags): Likewise.
18307 (dwarf2out_abstract_function): Treat the early generated DIEs
18308 as the abstract copy and only add DW_AT_inline and
18309 DW_AT_artificial here and call set_decl_origin_self.
18310 If the DIE has an abstract origin don't do anything.
18311 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
18312 if we have none yet (Go fails to build one, PR78628).
18313 (variably_modified_type_p): Prevent endless recursion for Ada
18314 cyclic pointer types.
18315 * lto-streamer-in.c: Include debug.h.
18316 (dref_queue): New global.
18317 (lto_read_tree_1): Stream in DIE references.
18318 (lto_input_tree): Register DIE references.
18319 (input_function): Stream DECL_DEBUG_ARGS.
18320 * lto-streamer-out.c: Include debug.h.
18321 (lto_write_tree_1): Output DIE references.
18322 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
18323 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
18324 (output_function): Stream DECL_DEBUG_ARGS.
18325 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
18326 Stream DECL_ABSTRACT_ORIGIN.
18327 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
18328 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
18329 DECL_CONTEXT for file-scope decls.
18330 * lto-streamer.h (struct dref_entry): Declare.
18331 (dref_queue): Likewise.
18332 * cfgexpand.c (pass_expand::execute): Do not call the
18333 outlining_inline_function hook here.
18334 * lto-wrapper.c (debug_obj): New global.
18335 (tool_cleanup): Unlink it if required.
18336 (debug_objcopy): New function.
18337 (run_gcc): Handle early debug sections in the IL files by
18338 extracting them to separate files, partially linkin them and
18339 feeding the result back as result to the linker.
18340 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
18341 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
18342 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
18343 sections into a separate segment.
18344 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
18346 (darwin_asm_dwarf_section): Likewise.
18347 (darwin_asm_output_dwarf_offset): Likewise.
18348 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
18350 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18351 Alan Hayward <alan.hayward@arm.com>
18352 David Sherwood <david.sherwood@arm.com>
18354 * read-md.h (md_reader::record_potential_iterator_use): Replace
18355 pointer argument with an rtx and an index.
18356 * read-rtl.c (iterator_group::apply_iterator): Likewise.
18357 (apply_mode_iterator): Likewise.
18358 (apply_code_iterator): Likewise.
18359 (apply_int_iterator): Likewise.
18360 (apply_subst_iterator): Likewise.
18361 (record_iterator_use): Likewise.
18362 (record_attribute_use): Likewise.
18363 (md_reader::record_potential_iterator_use): Likewise. Update calls
18364 to record_iterator_use and apply_iterator.
18365 (iterator_use): Replace ptr with x and index.
18366 (attribute_use): Likewise.
18367 (apply_attribute_uses): Update calls to apply_iterator.
18368 (apply_iterators): Likewise. Update initialization of iterator_use.
18369 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
18370 and record_potential_iterator_use.
18371 (rtx_reader::read_rtx_operand): Likewise.
18373 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
18374 Alan Hayward <alan.hayward@arm.com>
18375 David Sherwood <david.sherwood@arm.com>
18377 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
18380 2017-08-21 Richard Biener <rguenther@suse.de>
18382 PR middle-end/81884
18383 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
18384 at struct end conservatively when comparing common bases.
18386 2017-08-21 Richard Biener <rguenther@suse.de>
18388 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
18389 (mem_ref_in_stmt): Remove.
18390 (determine_max_movement): Use ref index to get at the reference.
18391 (invariantness_dom_walker::before_dom_children): Deal with
18392 lim data already initialized.
18393 (gather_mem_refs_stmt): Initialize lim data and record ref index.
18395 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18397 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
18398 (TARGET_ISA_ROUND): Ditto.
18399 (TARGET_ROUND): Ditto.
18400 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
18401 * config/i386/i386.md: Ditto.
18402 * config/i386/sse.md: Ditto.
18403 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
18404 with OPTION_MASK_ISA_SSE4_1.
18406 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
18409 * doc/extend.texi (x86 Built-in Functions): Correct the name of
18410 __builtin_ia32_lzcnt_u16.
18412 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
18415 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
18416 (rs6000_set_current_function): Rewrite function to use it.
18418 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
18421 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
18422 and TYPE_WARN_IF_NOT_ALIGN.
18423 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
18424 (handle_warn_if_not_align): New.
18425 (place_union_field): Call handle_warn_if_not_align.
18426 (place_field): Call handle_warn_if_not_align.
18427 Copy TYPE_WARN_IF_NOT_ALIGN.
18428 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
18429 (layout_type): Likewise.
18430 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
18432 (tree_decl_common): Add warn_if_not_align.
18433 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
18434 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
18435 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
18436 (DECL_WARN_IF_NOT_ALIGN): Likewise.
18437 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
18438 * doc/extend.texi: Document warn_if_not_aligned attribute.
18439 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
18441 2017-08-17 Martin Liska <mliska@suse.cz>
18444 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
18445 (get_data_dependence): Use it as pointer type.
18446 (distribute_loop): Likewise.
18448 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18450 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
18451 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
18452 (p8_vmrgow_<mode>_direct): New define_insn.
18453 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
18454 handle endianness for vmrgew and vmrgow permute patterns.
18456 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18458 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
18459 * config/rs6000/rs6000-cpus.def: Remove comment.
18460 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
18461 (POWERPC_MASKS): Likewise.
18462 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
18463 use of TARGET_VSX_TIMODE.
18464 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
18465 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
18466 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
18467 (rs6000_option_override_internal): Remove dead code.
18468 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
18469 (rs6000_legitimize_reload_address): Likewise.
18470 (rs6000_legitimate_address_p): Likewise.
18471 (rs6000_opt_masks): Delete "vsx-timode".
18472 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
18473 from function comment.
18474 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
18475 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
18476 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
18478 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
18479 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
18480 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
18481 (VSX_TI): Likewise.
18483 (define_peephole2): Likewise.
18485 2017-08-17 Martin Sebor <msebor@redhat.com>
18488 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
18489 past the end of an array.
18490 (test_pp_format): Add test cases.
18492 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
18494 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
18495 missing ECF_NOTHROW flags.
18497 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
18500 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
18501 operands residing in integer registers.
18502 (*vsx_le_perm_load_<mode>): Likewise.
18503 (*vsx_le_perm_store_<mode>): Likewise.
18504 (define_peephole2): Add peepholes to optimize the above.
18506 2017-08-17 Marek Polacek <polacek@redhat.com>
18508 PR middle-end/81814
18509 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
18510 to mimic what shorten_compare did. Change the return type to bool.
18511 (fold_cond_expr_with_comparison): Update call to
18512 operand_equal_for_comparison_p.
18513 (fold_ternary_loc): Likewise.
18515 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
18517 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
18519 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
18521 2017-08-17 Richard Biener <rguenther@suse.de>
18523 * tree-ssa-structalias.c (solve_graph): When propagating
18524 to successors update the graphs succ edges and avoid duplicate work.
18526 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
18529 * config/i386/i386.c (ix86_option_override_internal): Save target
18530 specific options after ix86_stack_protector_guard_reg was changed.
18532 2017-08-17 Richard Biener <rguenther@suse.de>
18534 PR tree-optimization/81827
18535 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
18536 (new_var_info): Initialize it conservatively.
18537 (get_call_vi): Mark register vars.
18538 (new_scalar_tmp_constraint_exp): Likewise.
18539 (handle_rhs_call): Likewise.
18540 (handle_const_call): Likewise.
18541 (create_function_info_for): Likewise.
18542 (solve_constraints): Sort varinfos to separate register from
18543 non-register vars to pack points-to solution bitmaps during
18546 2017-08-17 Marek Polacek <polacek@redhat.com>
18548 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
18550 2017-08-17 Richard Biener <rguenther@suse.de>
18552 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
18553 to true when overflow is undefined and we saturated the result.
18555 2017-08-17 Alan Modra <amodra@gmail.com>
18558 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
18559 Don't use store multiple if only one reg needs saving.
18560 (interesting_frame_related_regno): New function.
18561 (rs6000_frame_related): Don't emit frame info for regs that
18563 (rs6000_emit_epilogue): Likewise.
18565 2017-08-16 Nathan Sidwell <nathan@acm.org>
18567 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
18568 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
18569 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
18570 * tree.c (free_lang_data_in_type): Use else-if chain. Always
18571 clear TYPE_LANG_1. Remove obsolete member-function stripping.
18572 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
18573 (verify_type): Adjust for TYPE_BINFO move.
18574 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
18575 process TYPE_BINFO directly.
18576 (hash_tree): Likewise.
18577 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
18579 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
18582 2017-08-16 David Malcolm <dmalcolm@redhat.com>
18584 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
18586 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
18589 * config/i386/i386.md (*anddi_1_btr): Change predicates of
18590 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
18591 Add ix86_binary_operator_ok to insn constraint.
18592 (*iordi_1_bts): Ditto.
18593 (*xordi_1_btc): Ditto.
18594 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
18595 Update corresponding peephole2 pattern.
18599 2017-08-16 Bin Cheng <bin.cheng@arm.com>
18601 PR tree-optimization/81832
18602 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
18603 copy loop header which has IFN_LOOP_DIST_ALIAS call.
18605 2017-08-16 Marek Polacek <polacek@redhat.com>
18608 * fold-const.c (fold_indirect_ref_1): Restore original behavior
18609 regarding size_zero_node.
18611 2017-08-16 Martin Liska <mliska@suse.cz>
18614 * config.gcc: Respect previously set extra_objs in case
18617 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
18619 PR tree-optimization/81835
18620 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
18621 the phi SSA_NAME. Check that the condition in a COND_EXPR does
18622 not depend on the phi.
18624 2017-08-16 Alan Modra <amodra@gmail.com>
18626 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
18629 2017-08-16 Alan Modra <amodra@gmail.com>
18631 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
18632 (save_reg_p): ..into this. Update all callers.
18633 (first_reg_to_save): Simplify.
18635 2017-08-16 Alan Modra <amodra@gmail.com>
18637 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
18640 2017-08-15 Joseph Myers <joseph@codesourcery.com>
18644 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
18645 constant count if that count is less than 32.
18647 2017-08-15 Nathan Sidwell <nathan@acm.org>
18649 * gcc.c (execute): Emit friendlier message if inferior is killed
18650 by an external cause.
18652 2017-08-15 Richard Biener <rguenther@suse.de>
18654 PR tree-optimization/81790
18655 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
18656 CONSTRUCTORs from simplifying and VN.
18658 2017-08-14 Martin Sebor <msebor@redhat.com>
18660 * builtin-attrs.def: Add comments.
18662 2017-08-14 Martin Sebor <msebor@redhat.com>
18665 * doc/extend.texi (attribute nonstring): Document new attribute.
18667 2017-08-14 Martin Sebor <msebor@redhat.com>
18670 * tree-diagnostic.c (default_tree_printer): Handle %G.
18671 * gimple-pretty-print.h (percent_G_format): Declare new function.
18672 * gimple-pretty-print.c (percent_G_format): Define.
18673 * tree-pretty-print.c (percent_K_format): Add argument.
18675 2017-08-14 Martin Sebor <msebor@redhat.com>
18677 PR translation/79998
18678 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
18679 Remove a stray space.
18681 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
18684 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
18685 (*iordi_1_bts): Ditto.
18686 (*xordi_1_btc): Ditto.
18688 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18691 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
18692 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18694 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
18695 quoted strings, and make more translator-friendly.
18696 (darwin_rs6000_override_options): Likewise.
18697 (rs6000_option_override_internal): Likewise.
18698 (rs6000_return_in_memory): Fix overlong line.
18699 (init_cmulative_args): Use quoted strings, and make more
18700 translator-friendly.
18701 (rs6000_pass_by_reference): Fix overlong line.
18702 (def_builtin): Use quoted strings.
18703 (altivec_expand_predicate_builtin): Use quoted strings, and make
18704 more translator-friendly.
18705 (htm_expand_builtin): Use quoted strings.
18706 (cpu_expand_builtin): Use quoted strings, and make more
18707 translator-friendly.
18708 (altivec_expand_builtin): Likewise.
18709 (paired_expand_predicate_builtin): Likewise.
18710 (rs6000_invalid_builtin): Likewise.
18711 (builtin_function_type): Use quoted strings.
18712 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
18713 more translator-friendly.
18714 (rs6000_trampoline_init): Likewise.
18715 (rs6000_handle_altivec_attribute): Likewise.
18716 (rs6000_inner_target_options): Use quoted strings.
18717 (rs6000_disable_incompatible_switches): Likewise.
18718 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
18719 strings, and make more translator-friendly.
18720 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
18722 2017-08-14 Bin Cheng <bin.cheng@arm.com>
18724 PR tree-optimization/81799
18725 * tree-loop-distribution.c (version_loop_by_alias_check): Force
18726 cond_expr to simple gimple operand.
18728 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
18730 PR middle-end/46932
18731 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
18733 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
18737 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
18738 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
18739 TARGET_GASISR_PROLOGUES.
18740 * config/avr/avr.c (avr_option_override): Same.
18741 (avr_pass_pre_proep::execute): Same.
18743 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
18746 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
18747 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
18749 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
18751 * config/i386/i386.md (*load_tp_<mode>): Redefine as
18752 define_insn_and_split. Split to a memory load from 0 in
18753 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
18754 using PTR mode iterator.
18755 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
18756 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
18757 (*add_tp_<mode>): Redefine as define_insn_and_split.
18758 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
18759 address space. Merge with *add_tp_x32 using PTR mode iterator.
18760 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
18761 Split to an add with a memory load from 0 in
18762 DEFAULT_TLS_SEG_REG address space.
18764 2017-08-12 Andrew Pinski <apinski@cavium.com>
18766 * config/aarch64/aarch64-option-extensions.def (rdma):
18767 Fix feature string to what Linux prints out in /proc/cpuinfo.
18769 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
18772 * dwarf2out.c (modified_type_die): For C typedef types that have
18773 an ultimate origin, process the ultimate origin instead of the
18775 (gen_typedef_die): Assert that input DECLs have no ultimate
18777 (gen_type_die_with_usage): For typedef variants that have an
18778 ultimate origin, just call gen_decl_die on the original DECL.
18779 (process_scope_var): Avoid creating DIEs for local typedefs and
18780 concrete static variables.
18782 2017-08-12 Alan Modra <amodra@gmail.com>
18786 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
18787 match gnu-user.h startfile.
18788 (ENDFILE_LINUX_SPEC): Similarly.
18790 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
18793 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
18795 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
18797 2017-08-11 Tamar Christina <tamar.christina@arm.com>
18798 * config/aarch64/aarch64.md (mov<mode>): Change.
18799 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
18800 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
18801 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
18803 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
18805 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
18806 for storage order barriers.
18808 2017-08-11 Martin Liska <mliska@suse.cz>
18811 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
18812 variables of void type.
18814 2017-08-11 Martin Liska <mliska@suse.cz>
18816 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
18817 TARGET_SUPPORTS_ALIASES.
18818 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
18819 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
18820 (optimize_weakref): Likewise.
18821 * symtab.c (symtab_node::noninterposable_alias): Likewise.
18822 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
18823 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
18825 2017-08-11 Martin Liska <mliska@suse.cz>
18828 * config/i386/i386.c (make_resolver_func): Do complete
18829 refactoring of the function.
18831 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18834 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
18835 * config/i386/i386.c (ix86_stack_protect_guard): Use
18836 ix86_stack_protect_guard_symbol_str to generate varible declaration.
18837 * doc/invoke.texi (x86 Options): Document
18838 -mstack-protector-guard-symbol= option.
18840 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
18842 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
18843 * config/i386/i386.c (ix86_split_stack_guard): New function.
18844 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
18845 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
18846 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
18847 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
18848 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
18849 (split_stack_space_check): Call ix86_split_stack_guard.
18851 2017-08-10 Martin Sebor <msebor@redhat.com>
18853 * print-tree.c (print_node): Print location using the established
18855 Replace spaces with colons.
18856 (debug_raw, debug): Ditto.
18858 2017-08-10 Martin Sebor <msebor@redhat.com>
18861 * pretty-print.c (pp_format): Correct the handling of %s precision.
18863 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
18866 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
18868 (ix86_finalize_stack_frame_flags): This. Also clear
18869 frame_pointer_needed if -fno-omit-frame-pointer is used without
18871 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
18872 with ix86_finalize_stack_frame_flags.
18873 (ix86_expand_epilogue): Likewise.
18874 (ix86_expand_split_stack_prologue): Likewise.
18875 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
18877 2017-08-10 Martin Liska <mliska@suse.cz>
18880 * c-attribs.c (handle_target_attribute):
18881 Report warning for an empty string argument of target attribute.
18883 2017-08-09 Jakub Jelinek <jakub@redhat.com>
18886 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
18888 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
18889 or DECL_NONLOCAL labels.
18890 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
18891 or DECL_NONLOCAL labels here.
18893 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
18895 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
18896 to indicate when early gimple folding has been disabled.
18897 (rs6000_gimple_fold_builtin): Add debug content.
18898 (rs6000_invalid_builtin): Fix whitespace.
18899 (rs6000_expand_builtin): Fix whitespace.
18900 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
18902 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
18905 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
18906 SAVE_MULTIPLE if not all the registers that saves, should be saved.
18908 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
18910 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
18911 (qdf24xx): Likewise.
18912 * config/aarch64/aarch64.md: Include falkor.md.
18913 * config/aarch64/falkor.md: New.
18915 2017-08-09 Marek Polacek <polacek@redhat.com>
18918 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
18919 * diagnostic.c (emit_diagnostic): Add a comment.
18920 (emit_diagnostic_valist): New function.
18922 2017-08-09 Marek Polacek <polacek@redhat.com>
18925 * input.c (make_location): New overload.
18926 * input.h (make_location): Declare.
18928 2017-08-08 Alan Modra <amodra@gmail.com>
18929 H.J. Lu <hongjiu.lu@intel.com>
18932 * gcc.c (NO_PIE_SPEC): Delete.
18933 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
18935 (LINK_PIE_SPEC): ..to here.
18936 (LINK_COMMAND_SPEC): Support -no-pie.
18937 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
18938 chain of crtbegin*.o selection, update for PIE_SPEC changes and
18940 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
18941 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
18942 (ENDFILE_CRTEND_SPEC): Similarly.
18944 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
18947 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
18948 (mstack-protector-guard-offset=): Ditto.
18949 * config/i386/i386.c (ix86_option_override): Handle
18950 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
18952 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
18953 ix86_stack_protect_guard_offset variables.
18954 (TARGET_STACK_PROTECT_GUARD): Always define.
18955 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
18956 and -mstack-protector-guard-offset= options.
18958 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18960 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
18961 boundary case for the last candidate.
18963 2017-08-08 Bin Cheng <bin.cheng@arm.com>
18965 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
18966 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
18968 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18970 PR middle-end/19706
18971 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
18972 * config/aarch64/aarch64-builtins.c
18973 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
18974 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
18975 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
18977 2017-08-08 Tamar Christina <tamar.christina@arm.com>
18978 Andrew Pinski <pinskia@gmail.com>
18980 PR middle-end/19706
18981 * internal-fn.def (XORSIGN): New.
18982 * optabs.def (xorsign_optab): New.
18983 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
18984 (convert_expand_mult_copysign): New.
18985 (pass_optimize_widening_mul::execute): Call
18986 convert_expand_mult_copysign.
18988 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18990 PR tree-optimization/81354
18991 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
18992 Insert on edges rather than explicitly creating landing pads.
18993 (analyze_candidates_and_replace): Commit edge inserts.
18995 2017-08-08 Richard Biener <rguenther@suse.de>
18997 PR middle-end/81719
18998 * tree-ssa-loop-niter.c: Include tree-dfa.h.
18999 (expand_simple_operations): Also look through ADDR_EXPRs with
19000 MEM_REF bases treating them as POINTER_PLUS_EXPR.
19002 2017-08-08 Richard Biener <rguenther@suse.de>
19004 PR tree-optimization/81723
19005 * tree-vect-slp.c (struct bst_traits): New hash traits.
19006 (bst_fail): New global.
19007 (vect_build_slp_tree_2): New worker, split out from ...
19008 (vect_build_slp_tree): ... this now wrapping it with using
19009 bst_fail set to cache SLP tree build fails. Properly handle
19011 (vect_analyze_slp_instance): Allocate and free bst_fail.
19013 2017-08-08 Martin Liska <mliska@suse.cz>
19016 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
19017 LABEL_DECLs that can be from a different function.
19019 2017-08-08 Bin Cheng <bin.cheng@arm.com>
19021 PR tree-optimization/81744
19022 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
19023 loop's number of iterations.
19025 2017-08-08 Martin Liska <mliska@suse.cz>
19027 * asan.c: Include header files.
19028 * attribs.c (build_decl_attribute_variant): New function moved
19030 (build_type_attribute_qual_variant): Likewise.
19031 (cmp_attrib_identifiers): Likewise.
19032 (simple_cst_list_equal): Likewise.
19033 (omp_declare_simd_clauses_equal): Likewise.
19034 (attribute_value_equal): Likewise.
19035 (comp_type_attributes): Likewise.
19036 (build_type_attribute_variant): Likewise.
19037 (lookup_ident_attribute): Likewise.
19038 (remove_attribute): Likewise.
19039 (merge_attributes): Likewise.
19040 (merge_type_attributes): Likewise.
19041 (merge_decl_attributes): Likewise.
19042 (merge_dllimport_decl_attributes): Likewise.
19043 (handle_dll_attribute): Likewise.
19044 (attribute_list_equal): Likewise.
19045 (attribute_list_contained): Likewise.
19046 * attribs.h (lookup_attribute): New function moved from tree.[ch].
19047 (lookup_attribute_by_prefix): Likewise.
19048 * bb-reorder.c: Include header files.
19049 * builtins.c: Likewise.
19050 * calls.c: Likewise.
19051 * cfgexpand.c: Likewise.
19052 * cgraph.c: Likewise.
19053 * cgraphunit.c: Likewise.
19054 * convert.c: Likewise.
19055 * dwarf2out.c: Likewise.
19056 * final.c: Likewise.
19057 * fold-const.c: Likewise.
19058 * function.c: Likewise.
19059 * gimple-expr.c: Likewise.
19060 * gimple-fold.c: Likewise.
19061 * gimple-pretty-print.c: Likewise.
19062 * gimple.c: Likewise.
19063 * gimplify.c: Likewise.
19064 * hsa-common.c: Likewise.
19065 * hsa-gen.c: Likewise.
19066 * internal-fn.c: Likewise.
19067 * ipa-chkp.c: Likewise.
19068 * ipa-cp.c: Likewise.
19069 * ipa-devirt.c: Likewise.
19070 * ipa-fnsummary.c: Likewise.
19071 * ipa-inline.c: Likewise.
19072 * ipa-visibility.c: Likewise.
19074 * lto-cgraph.c: Likewise.
19075 * omp-expand.c: Likewise.
19076 * omp-general.c: Likewise.
19077 * omp-low.c: Likewise.
19078 * omp-offload.c: Likewise.
19079 * omp-simd-clone.c: Likewise.
19080 * opts-global.c: Likewise.
19081 * passes.c: Likewise.
19082 * predict.c: Likewise.
19083 * sancov.c: Likewise.
19084 * sanopt.c: Likewise.
19085 * symtab.c: Likewise.
19086 * toplev.c: Likewise.
19087 * trans-mem.c: Likewise.
19088 * tree-chkp.c: Likewise.
19089 * tree-eh.c: Likewise.
19090 * tree-into-ssa.c: Likewise.
19091 * tree-object-size.c: Likewise.
19092 * tree-parloops.c: Likewise.
19093 * tree-profile.c: Likewise.
19094 * tree-ssa-ccp.c: Likewise.
19095 * tree-ssa-live.c: Likewise.
19096 * tree-ssa-loop.c: Likewise.
19097 * tree-ssa-sccvn.c: Likewise.
19098 * tree-ssa-structalias.c: Likewise.
19099 * tree-ssa.c: Likewise.
19100 * tree-streamer-in.c: Likewise.
19101 * tree-vectorizer.c: Likewise.
19102 * tree-vrp.c: Likewise.
19103 * tsan.c: Likewise.
19104 * ubsan.c: Likewise.
19105 * varasm.c: Likewise.
19106 * varpool.c: Likewise.
19107 * tree.c: Remove functions moved to attribs.[ch].
19108 * tree.h: Likewise.
19109 * config/aarch64/aarch64.c: Add attrs.h header file.
19110 * config/alpha/alpha.c: Likewise.
19111 * config/arc/arc.c: Likewise.
19112 * config/arm/arm.c: Likewise.
19113 * config/avr/avr.c: Likewise.
19114 * config/bfin/bfin.c: Likewise.
19115 * config/c6x/c6x.c: Likewise.
19116 * config/cr16/cr16.c: Likewise.
19117 * config/cris/cris.c: Likewise.
19118 * config/darwin.c: Likewise.
19119 * config/epiphany/epiphany.c: Likewise.
19120 * config/fr30/fr30.c: Likewise.
19121 * config/frv/frv.c: Likewise.
19122 * config/ft32/ft32.c: Likewise.
19123 * config/h8300/h8300.c: Likewise.
19124 * config/i386/winnt.c: Likewise.
19125 * config/ia64/ia64.c: Likewise.
19126 * config/iq2000/iq2000.c: Likewise.
19127 * config/lm32/lm32.c: Likewise.
19128 * config/m32c/m32c.c: Likewise.
19129 * config/m32r/m32r.c: Likewise.
19130 * config/m68k/m68k.c: Likewise.
19131 * config/mcore/mcore.c: Likewise.
19132 * config/microblaze/microblaze.c: Likewise.
19133 * config/mips/mips.c: Likewise.
19134 * config/mmix/mmix.c: Likewise.
19135 * config/mn10300/mn10300.c: Likewise.
19136 * config/moxie/moxie.c: Likewise.
19137 * config/msp430/msp430.c: Likewise.
19138 * config/nds32/nds32-isr.c: Likewise.
19139 * config/nds32/nds32.c: Likewise.
19140 * config/nios2/nios2.c: Likewise.
19141 * config/nvptx/nvptx.c: Likewise.
19142 * config/pa/pa.c: Likewise.
19143 * config/pdp11/pdp11.c: Likewise.
19144 * config/powerpcspe/powerpcspe.c: Likewise.
19145 * config/riscv/riscv.c: Likewise.
19146 * config/rl78/rl78.c: Likewise.
19147 * config/rx/rx.c: Likewise.
19148 * config/s390/s390.c: Likewise.
19149 * config/sh/sh.c: Likewise.
19150 * config/sol2.c: Likewise.
19151 * config/sparc/sparc.c: Likewise.
19152 * config/spu/spu.c: Likewise.
19153 * config/stormy16/stormy16.c: Likewise.
19154 * config/tilegx/tilegx.c: Likewise.
19155 * config/tilepro/tilepro.c: Likewise.
19156 * config/v850/v850.c: Likewise.
19157 * config/vax/vax.c: Likewise.
19158 * config/visium/visium.c: Likewise.
19159 * config/xtensa/xtensa.c: Likewise.
19161 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19164 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
19165 constraints since the -mupper-regs-* switches have been
19167 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
19168 into a vector from a double word element that was extracted from
19169 another vector, and eliminate extra XXPERMDI instructions.
19170 (vsx_concat_<mode>_2): Likewise.
19171 (vsx_concat_<mode>_3): Likewise.
19172 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
19173 concat to allow optimizing inserts from previous extracts.
19175 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
19177 * config/i386/i386.c (ix86_stack_protect_guard): Generate
19178 memory reference to a SSP offset in TLS address space.
19179 (ix86_print_operand) <case '@'>: Remove.
19180 (ix86_print_operand_punct_valid_p): Remove '@' code.
19181 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
19182 UNSPEC_SP_TLS_TEST.
19183 (stack_tls_protect_set_<mode>): Remove.
19184 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
19185 (stack_tls_protect_test_<mode>): Remove.
19186 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
19188 2017-08-07 Olivier Hainque <hainque@adacore.com>
19191 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
19193 2017-08-07 Douglas Rupp <rupp@adacore.com>
19195 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
19196 variable was referenced as multidir in command.
19198 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19201 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
19204 2017-08-07 Martin Liska <mliska@suse.cz>
19206 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
19207 * config/rl78/rl78.c: Add include of attribs.h.
19208 * config/sh/sh.c: Likewise.
19209 * config/v850/v850.c: Likewise.
19211 2017-08-07 Tom de Vries <tom@codesourcery.com>
19213 PR middle-end/78266
19214 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
19216 2017-08-07 Martin Liska <mliska@suse.cz>
19218 * config/mips/mips.c: Include attribs.h.
19220 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19223 * doc/invoke.texi: Document change in behvaior for -Ofast for
19226 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
19228 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
19230 (aarch64_pop_regs): Likewise.
19231 (aarch64_gen_load_pair): Likewise.
19232 (aarch64_save_callee_saves): Likewise.
19233 (aarch64_restore_callee_saves): Likewise.
19235 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19237 * config/i386/i386.c: Revert the last change.
19239 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19242 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
19244 (ix86_finalize_stack_frame_flags): This. Also clear
19245 frame_pointer_needed if -fno-omit-frame-pointer is used without
19247 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
19248 with ix86_finalize_stack_frame_flags.
19249 (ix86_expand_epilogue): Likewise.
19250 (ix86_expand_split_stack_prologue): Likewise.
19252 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
19255 * config/i386/i386.c (get_builtin_code_for_version): Set priority
19256 to P_AES for Westmere.
19258 2017-08-07 Jonathan Yong <10walls@gmail.com>
19260 * config/i386/mingw.opt (fset-stack-executable): Removed.
19261 * config/i386/cygming.opt (fset-stack-executable): Moved
19263 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
19265 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
19267 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
19269 2017-08-07 Marek Polacek <polacek@redhat.com>
19271 PR middle-end/81737
19272 * fold-const.c (fold_indirect_ref_1): Check type_domain.
19274 2017-08-07 Martin Liska <mliska@suse.cz>
19276 * attribs.h (canonicalize_attr_name): New function.
19277 (cmp_attribs): Move from c-format.c and adjusted.
19278 (is_attribute_p): Moved from tree.h.
19279 * tree-inline.c: Add new includes.
19280 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
19281 (private_is_attribute_p): Remove.
19282 (private_lookup_attribute): Likewise.
19283 (private_lookup_attribute_by_prefix): Simplify.
19284 (remove_attribute): Use is_attribute_p.
19285 * tree.h: Remove removed declarations.
19287 2017-08-07 Jakub Jelinek <jakub@redhat.com>
19289 PR middle-end/81698
19290 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
19291 instead of computing it in the function. Formatting fix.
19292 (expand_case): Don't rely on default_edge being the first edge,
19293 clear it if removing it, pass default_edge to
19294 emit_case_dispatch_table.
19295 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
19298 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
19300 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
19301 insn in the function, emit NOP after the insn.
19303 2017-08-06 Tom de Vries <tom@codesourcery.com>
19305 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
19308 2017-08-06 Tom de Vries <tom@codesourcery.com>
19310 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
19313 2017-08-04 Yury Gribov <tetra2005@gmail.com>
19315 PR tree-optimization/57371
19316 * match.pd: New pattern.
19318 2017-08-04 Marek Polacek <polacek@redhat.com>
19320 PR middle-end/81695
19321 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
19322 perform the computation in offset_int.
19324 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19326 PR tree-optimization/81136
19327 * tree-vectorizer.h: Include tree-hash-traits.h.
19328 (vec_base_alignments): New typedef.
19329 (vec_info): Add a base_alignments field.
19330 (vect_record_base_alignments): Declare.
19331 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
19333 (DR_IS_CONDITIONAL_IN_STMT): New macro.
19334 (create_data_ref): Add an is_conditional_in_stmt argument.
19335 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
19336 the is_conditional_in_stmt field.
19337 (data_ref_loc): Add an is_conditional_in_stmt field.
19338 (get_references_in_stmt): Set the is_conditional_in_stmt field.
19339 (find_data_references_in_stmt): Update call to create_data_ref.
19340 (graphite_find_data_references_in_stmt): Likewise.
19341 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
19342 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
19343 (vect_record_base_alignment): New function.
19344 (vect_record_base_alignments): Likewise.
19345 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
19346 for nested statements even if we fail to compute a misalignment.
19347 Use pooled base alignments for unconditional references.
19348 (vect_find_same_alignment_drs): Compare base addresses instead
19350 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
19351 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
19353 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19355 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
19356 Add an explicit name for the enum. Use auto_vec for slp_instances
19357 and grouped_stores.
19358 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
19360 (_bb_vec_info): Add a constructor and destructor.
19361 (vinfo_for_stmt): Return NULL for uids of -1 as well.
19362 (destroy_loop_vec_info): Delete.
19363 (vect_destroy_datarefs): Likewise.
19364 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
19365 (vec_info::vec_info): New function.
19366 (vec_info::~vec_info): Likewise.
19367 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
19368 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
19369 destroy_loop_vec_info.
19370 * tree-vect-loop.c (new_loop_vec_info): Replace with...
19371 (_loop_vec_info::_loop_vec_info): ...this.
19372 (destroy_loop_vec_info): Replace with...
19373 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
19374 the stmt_vec_infos. Leave handling of vec_info information to its
19375 destructor. Remove explicit vector releases.
19376 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
19377 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
19378 * tree-vect-slp.c (new_bb_vec_info): Replace with...
19379 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
19380 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
19381 (destroy_bb_vec_info): Replace with...
19382 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
19383 information to its destructor.
19384 (vect_slp_analyze_bb_1): Use new and delete instead of
19385 new_bb_vec_info and destroy_bb_vec_info.
19386 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
19389 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
19391 * tree-data-ref.h (subscript): Add access_fn field.
19392 (data_dependence_relation): Add could_be_independent_p.
19393 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
19394 (same_access_functions): Move to tree-data-ref.c.
19395 * tree-data-ref.c (ref_contains_union_access_p): New function.
19396 (access_fn_component_p): Likewise.
19397 (access_fn_components_comparable_p): Likewise.
19398 (dr_analyze_indices): Add a reference to access_fn_component_p.
19399 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
19401 (constant_access_functions): Likewise.
19402 (add_other_self_distances): Likewise.
19403 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
19404 (initialize_data_dependence_relation): Use XCNEW and remove
19405 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
19406 of access functions that have the same type. Allow the
19407 subsequence to end with different bases in some circumstances.
19408 Record the chosen access functions in SUB_ACCESS_FN.
19409 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
19410 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
19411 (subscript_dependence_tester_1): Likewise dra and drb.
19412 (build_classic_dist_vector): Update calls accordingly.
19413 (subscript_dependence_tester): Likewise.
19414 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
19415 DDR_COULD_BE_INDEPENDENT_P.
19416 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
19417 comp_alias_ddrs instead of may_alias_ddrs.
19418 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
19420 (vect_analyze_data_ref_dependence): Use it if
19421 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
19422 distance vectors if that fails.
19423 (dependence_distance_ge_vf): New function.
19424 (vect_prune_runtime_alias_test_list): Use it. Don't clear
19425 LOOP_VINFO_MAY_ALIAS_DDRS.
19427 2017-08-04 Richard Biener <rguenther@suse.de>
19429 PR middle-end/81705
19430 * fold-const.c (fold_binary_loc): Properly restrict
19431 minus_var0 && minus_var1 case when associating undefined overflow
19434 2017-08-04 Tom de Vries <tom@codesourcery.com>
19436 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
19438 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19440 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19441 Don't start diagnostic messages with a capital letter.
19442 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19444 (rs6000_invalid_builtin): Likewise.
19445 (rs6000_trampoline_init): Likewise.
19447 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19450 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
19451 after setting changeable df flags.
19453 2017-08-03 Richard Biener <rguenther@suse.de>
19455 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
19456 up if the use is in USE - X.
19458 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
19460 * toplev.c (dumpfile.h): New include.
19461 (internal_error_reentered): New static function. Use it...
19462 (internal_error_function): ...here to handle reentered internal_error.
19464 2017-08-03 Richard Biener <rguenther@suse.de>
19466 PR middle-end/81148
19467 * fold-const.c (split_tree): Add minus_var and minus_con
19468 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
19469 here but always use minus_*.
19470 (associate_trees): Assert we never associate with MINUS_EXPR
19471 and NULL first operand. Do not recurse for PLUS_EXPR operands
19472 when associating as MINUS_EXPR either.
19473 (fold_binary_loc): Track minus_var and minus_con.
19475 2017-08-03 Tom de Vries <tom@codesourcery.com>
19478 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
19479 ACCEL_COMPILER, apply finish_options on
19480 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
19482 2017-08-03 Tom de Vries <tom@codesourcery.com>
19485 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
19486 function_entry_patch_area_size > 0.
19488 2017-08-03 Jakub Jelinek <jakub@redhat.com>
19491 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
19492 instead of 10??LU, perform unit multiplication in wide_int,
19493 don't change alloc_object_size_limit if the limit is larger
19496 PR tree-optimization/81655
19497 PR tree-optimization/81588
19498 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
19499 the case when ranges[i].low and high are 1 for unsigned type with
19502 PR middle-end/81052
19503 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
19504 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
19506 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19508 * tree-vrp.h: Add include guard.
19510 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
19513 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
19514 (ud2): New insn pattern.
19515 * config/i386/i386.c (ix86_expand_epilogue):
19516 For naked functions, generate ud2 instead of trap insn.
19518 2017-08-02 Marek Polacek <polacek@redhat.com>
19521 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
19523 2017-08-02 Tom de Vries <tom@codesourcery.com>
19524 Cesar Philippidis <cesar@codesourcery.com>
19526 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
19527 Add missing edge probabilities.
19529 2017-08-02 Tamar Christina <tamar.christina@arm.com>
19531 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
19532 Correct endianness.
19534 2017-08-02 Jakub Jelinek <jakub@redhat.com>
19536 PR middle-end/79499
19537 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
19538 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
19539 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
19540 prologue_seq sequences - if any.
19542 2017-08-02 Richard Biener <rguenther@suse.de>
19544 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
19545 via vectors if supported, integer extracts via punning if supported
19546 or otherwise vector extracts.
19548 2017-08-02 Richard Biener <rguenther@suse.de>
19550 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
19552 (bitmap_insert_into_set): ... this.
19554 2017-08-02 Richard Biener <rguenther@suse.de>
19556 PR tree-optimization/81633
19558 2015-08-17 Alan Hayward <alan.hayward@arm.com>
19560 PR tree-optimization/71752
19561 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
19563 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
19565 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
19566 (machine_function::call_ms2sysv_pad_out): Remove field.
19567 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
19568 (ix86_compute_frame_layout): Likewise.
19570 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
19573 * config/i386/i386.c (ix86_set_func_type): Disallow naked
19574 attribute with interrupt attribute.
19576 2017-08-01 Andrew Pinski <apinski@cavium.com>
19578 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
19579 BIT_INSERT_EXPR's operand 1
19580 to see if the types precision matches.
19582 2017-08-01 Martin Liska <mliska@suse.cz>
19584 PR middle-end/70140
19585 * builtins.c (expand_builtin_memcpy_args): Remove.
19586 (expand_builtin_memcpy): Call newly added function
19587 expand_builtin_memory_copy_args.
19588 (expand_builtin_memcpy_with_bounds): Likewise.
19589 (expand_builtin_mempcpy): Remove last argument.
19590 (expand_builtin_mempcpy_with_bounds): Likewise.
19591 (expand_builtin_memory_copy_args): New function created from
19592 expand_builtin_mempcpy_args with small modifications.
19593 (expand_builtin_mempcpy_args): Remove.
19594 (expand_builtin_stpcpy): Remove unused argument.
19595 (expand_builtin): Likewise.
19596 (expand_builtin_with_bounds): Likewise.
19598 2017-08-01 Martin Liska <mliska@suse.cz>
19601 Make mempcpy more optimal (PR middle-end/70140).
19603 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19606 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
19607 __builtin_vec_cmpne verify both arguments are compatible vectors
19608 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
19609 verify arg1_type is a pointer or array type. For __builtin_vec_st,
19610 move computation of aligned to after checking the argument types.
19614 * config/rs6000/vsx.md (vextract_fp_from_shorth,
19615 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
19618 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
19619 Doug Rupp <rupp@adacore.com>
19620 Olivier Hainque <hainque@adacore.com>
19622 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
19623 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
19625 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
19626 for TARGET_OS_CPP_BUILTIN.
19627 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
19628 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
19630 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
19631 passing required abi options to the assembler for EABI configurations.
19632 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
19633 of .text.hot and .text.unlikely sections for kernel modules when
19634 using ARM style exceptions.
19635 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
19636 options. Add EXTRA_CC1_SPEC.
19637 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
19639 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
19641 (ARM_TARGET2_DWARF_FORMAT): Define.
19642 * config/arm/t-vxworks: Adjust multilib control to removal of the
19643 Diab command line options.
19645 2017-08-01 Martin Liska <mliska@suse.cz>
19647 PR gcov-profile/81561
19648 * gcov.c (unblock): Make unblocking safe as we need to preserve
19649 index correspondence of blocks and block_lists.
19651 2017-08-01 Richard Biener <rguenther@suse.de>
19653 PR tree-optimization/81181
19654 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
19655 (compute_antic): ... end of iteration here.
19657 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
19659 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
19660 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
19661 (ftree-slp-vectorize): Likewise.
19662 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
19663 can no longer be set independent of flag_tree_loop_vectorize.
19664 * omp-general.c (emp_max_vf): Likewise.
19665 * opts.c (enable_fdo_optimizations): Remove references to
19666 flag_tree_vectorize, these are now implicit.
19667 (common_handle_option): Remove handling for OPT_ftree_vectorize,
19668 and leave it for the options machinery.
19670 2017-08-01 Martin Liska <mliska@suse.cz>
19672 PR middle-end/70140
19673 * builtins.c (expand_builtin_memcpy_args): Remove.
19674 (expand_builtin_memcpy): Call newly added function
19675 expand_builtin_memory_copy_args.
19676 (expand_builtin_memcpy_with_bounds): Likewise.
19677 (expand_builtin_mempcpy): Remove last argument.
19678 (expand_builtin_mempcpy_with_bounds): Likewise.
19679 (expand_builtin_memory_copy_args): New function created from
19680 expand_builtin_mempcpy_args with small modifications.
19681 (expand_builtin_mempcpy_args): Remove.
19682 (expand_builtin_stpcpy): Remove unused argument.
19683 (expand_builtin): Likewise.
19684 (expand_builtin_with_bounds): Likewise.
19686 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19689 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
19690 print "ds:" only for immediates in generic address space.
19692 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
19695 * config/i386/i386.c (ix86_funciton_naked): New prototype.
19696 (ix86_function_ok_for_sibcall): Return false for naked functions.
19698 2017-08-01 Richard Biener <rguenther@suse.de>
19700 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
19701 (compute_antic): Seed worklist with exit block predecessors.
19702 * cfganal.c (dfs_find_deadend): For a cycle return the source
19703 of the edge closing it.
19705 2017-08-01 Tamar Christina <tamar.christina@arm.com>
19707 * config/aarch64/aarch64.c
19708 (aarch64_can_const_movi_rtx_p): Move 0 check.
19710 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19712 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
19713 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
19715 * match.pd: Ditto in address comparison pattern.
19717 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19719 PR tree-optimization/81627
19720 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
19721 closed ssa form for store-store chain.
19723 2017-08-01 Bin Cheng <bin.cheng@arm.com>
19725 PR tree-optimization/81620
19726 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
19727 for store-store chain.
19729 2017-08-01 Jakub Jelinek <jakub@redhat.com>
19731 PR tree-optimization/81588
19732 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
19733 ranges[i].in_p, invert comparison code ccode. For >/>=,
19734 swap rhs1 and rhs2 and comparison code unconditionally,
19735 for </<= don't do that. Don't swap rhs1/rhs2 again if
19736 ranges[i].in_p, instead invert comparison code ccode if
19737 opcode or oe->rank is BIT_IOR_EXPR.
19740 * optabs.def (vec_extract_optab, vec_init_optab): Change from
19741 a direct optab to conversion optab.
19742 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
19743 with GET_MODE_INNER as last argument instead of optab_handler.
19744 * expmed.c (extract_bit_field_1): Likewise. Use vector from
19745 vector extraction if possible and optab is available.
19746 * expr.c (store_constructor): Use convert_optab_handler instead
19747 of optab_handler. Use vector initialization from smaller
19748 vectors if possible and optab is available.
19749 * tree-vect-stmts.c (vectorizable_load): Likewise.
19750 * doc/md.texi (vec_extract, vec_init): Document that the optabs
19751 now have two modes.
19752 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
19753 of vec_init from half-sized vectors with the same element mode.
19754 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
19755 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
19756 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
19757 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
19758 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
19759 after mode in gen_vec_extract* calls.
19760 (vec_extract<mode>): Renamed to ...
19761 (vec_extract<mode><ssescalarmodelower>): ... this.
19762 (vec_extract<mode><ssehalfvecmodelower>): New expander.
19763 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
19764 element mode after mode in gen_vec_init* calls.
19765 (VEC_INIT_HALF_MODE): New mode iterator.
19766 (vec_init<mode>): Renamed to ...
19767 (vec_init<mode><ssescalarmodelower>): ... this.
19768 (vec_init<mode><ssehalfvecmodelower>): New expander.
19769 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
19770 (vec_extractv2sfsf): ... this.
19771 (vec_initv2sf): Renamed to ...
19772 (vec_initv2sfsf): ... this.
19773 (vec_extractv2si): Renamed to ...
19774 (vec_extractv2sisi): ... this.
19775 (vec_initv2si): Renamed to ...
19776 (vec_initv2sisi): ... this.
19777 (vec_extractv4hi): Renamed to ...
19778 (vec_extractv4hihi): ... this.
19779 (vec_initv4hi): Renamed to ...
19780 (vec_initv4hihi): ... this.
19781 (vec_extractv8qi): Renamed to ...
19782 (vec_extractv8qiqi): ... this.
19783 (vec_initv8qi): Renamed to ...
19784 (vec_initv8qiqi): ... this.
19785 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
19786 (vec_init<mode>): Renamed to ...
19787 (vec_init<mode><VEC_base_l>): ... this.
19788 (vec_extract<mode>): Renamed to ...
19789 (vec_extract<mode><VEC_base_l>): ... this.
19790 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
19791 (vec_initv2sfsf): ... this.
19792 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
19793 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19794 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
19795 element mode after mode in gen_vec_init* calls.
19796 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
19797 (vec_init<mode><Vel>): ... this.
19798 (vec_extract<mode>): Renamed to ...
19799 (vec_extract<mode><Vel>): ... this.
19800 * config/aarch64/iterators.md (Vel): New mode attribute.
19801 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
19802 Add element mode after mode in gen_vec_extract* calls.
19803 * config/s390/vector.md (non_vec_l): New mode attribute.
19804 (vec_extract<mode>): Renamed to ...
19805 (vec_extract<mode><non_vec_l>): ... this.
19806 (vec_init<mode>): Renamed to ...
19807 (vec_init<mode><non_vec_l>): ... this.
19808 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
19809 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
19811 * config/arm/iterators.md (V_elem_l): New mode attribute.
19812 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
19813 (vec_extract<mode><V_elem_l>): ... this.
19814 (vec_extractv2di): Renamed to ...
19815 (vec_extractv2didi): ... this.
19816 (vec_init<mode>): Renamed to ...
19817 (vec_init<mode><V_elem_l>): ... this.
19818 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
19819 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
19820 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
19821 Add element mode after gen_vec_extract* calls.
19822 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
19823 (vec_init<mode><unitmode>): ... this.
19824 (vec_extract<mode>): Renamed to ...
19825 (vec_extract<mode><unitmode>): ... this.
19826 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
19827 (vec_init<mode><unitmode>): ... this.
19828 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
19829 (vec_initv2sfsf): ... this.
19830 (vec_extractv2sf): Renamed to ...
19831 (vec_extractv2sfsf): ... this.
19832 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
19833 Add element mode after gen_vec_extract* calls.
19834 * config/mips/mips.md (unitmode): New mode iterator.
19835 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
19836 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
19837 * config/spu/spu.md (inner_l): New mode attribute.
19838 (vec_init<mode>): Renamed to ...
19839 (vec_init<mode><inner_l>): ... this.
19840 (vec_extract<mode>): Renamed to ...
19841 (vec_extract<mode><inner_l>): ... this.
19842 * config/sparc/sparc.md (veltmode): New mode iterator.
19843 (vec_init<VMALL:mode>): Renamed to ...
19844 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
19845 * config/ia64/vect.md (vec_initv2si): Renamed to ...
19846 (vec_initv2sisi): ... this.
19847 (vec_initv2sf): Renamed to ...
19848 (vec_initv2sfsf): ... this.
19849 (vec_extractv2sf): Renamed to ...
19850 (vec_extractv2sfsf): ... this.
19851 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
19852 (vec_init<mode>): Renamed to ...
19853 (vec_init<mode><VEC_base_l>): ... this.
19854 (vec_extract<mode>): Renamed to ...
19855 (vec_extract<mode><VEC_base_l>): ... this.
19856 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
19857 (vec_initv2sfsf): ... this.
19858 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
19859 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
19860 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
19861 gen_vec_init* calls.
19863 2017-08-01 Richard Biener <rguenther@suse.de>
19865 PR tree-optimization/81297
19866 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
19867 TREE_OVERFLOW from INTEGER_CSTs.
19869 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
19871 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
19873 2017-07-31 Carl Love <cel@us.ibm.com>
19875 * config/rs6000/rs6000-c: Add support for built-in functions
19876 vector signed char vec_xl_be (signed long long, signed char *);
19877 vector unsigned char vec_xl_be (signed long long, unsigned char *);
19878 vector signed int vec_xl_be (signed long long, signed int *);
19879 vector unsigned int vec_xl_be (signed long long, unsigned int *);
19880 vector signed long long vec_xl_be (signed long long, signed long long *);
19881 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
19882 vector signed short vec_xl_be (signed long long, signed short *);
19883 vector unsigned short vec_xl_be (signed long long, unsigned short *);
19884 vector double vec_xl_be (signed long long, double *);
19885 vector float vec_xl_be (signed long long, float *);
19886 * config/rs6000/altivec.h (vec_xl_be): Add #define.
19887 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
19888 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
19890 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
19891 (altivec_expand_builtin): Add switch statement to call
19892 altivec_expand_xl_be for each builtin.
19893 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
19894 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
19895 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
19896 __builtin_vsx_le_be_v16qi.
19897 * doc/extend.texi: Update the built-in documentation file for the
19898 new built-in functions.
19900 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19903 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
19905 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
19907 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19909 * config.gcc: Add z14.
19910 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
19911 CPU model numbers for z13s and z14.
19912 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
19914 * config/s390/s390-opts.h (enum processor_type): Rename
19915 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19916 * config/s390/s390.c (processor_table): Add field for CPU name to
19917 be passed to Binutils.
19918 (s390_asm_output_machine_for_arch): Use the new field in
19919 processor_table for Binutils.
19920 (s390_expand_builtin): Replace arch12 with z14.
19921 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
19922 (s390_get_sched_attrmask): Likewise.
19923 (s390_get_unit_mask): Likewise.
19924 * config/s390/s390.opt: Add z14 to processor_type enum.
19926 2017-07-31 Martin Jambor <mjambor@suse.cz>
19929 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
19930 regardless of optimization level.
19932 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
19933 Martin Liska <mliska@suse.cz>
19935 * predict.def: Remove old comment and adjust probability.
19936 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
19937 PREDICT statements.
19939 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
19942 * config/i386/i386.c (ix86_function_naked): New function.
19943 (ix86_can_use_return_insn_p): Return false for naked functions.
19944 (ix86_expand_prologue): Skip prologue for naked functions.
19945 (ix86_expand_epilogue): Skip epilogue for naked functions
19946 and emit trap instruction.
19947 (ix86_warn_func_return): New function.
19948 (ix86_attribute_table): Add "naked" attribute specification.
19949 (TARGET_WARN_FUNC_RETURN): Define.
19950 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
19952 2017-07-31 Martin Liska <mliska@suse.cz>
19954 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
19955 (dump_gimple_bb_header): Always dump BB info.
19956 (pp_cfg_jump): Do not append info about BB when dumping a jump.
19958 2017-07-31 Martin Liska <mliska@suse.cz>
19961 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
19962 also with current_function_decl non-null equality.
19964 2017-07-31 Jakub Jelinek <jakub@redhat.com>
19967 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
19968 change type to the element type, instead add eltype variable and
19969 use it where we are interested in the element type.
19971 PR tree-optimization/81603
19972 * ipa-polymorphic-call.c
19973 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
19974 offset arithmetic in offset_int, bail out if the resulting bit offset
19975 doesn't fit into shwi.
19977 2017-07-31 Martin Liska <mliska@suse.cz>
19979 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
19980 (gimplify_save_expr): Fix comment.
19982 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
19985 * config/i386/i386.c (ix86_function_arg): Update arguments for
19987 (ix86_compute_frame_layout): Set the initial stack offset to
19988 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
19989 INCOMING_FRAME_SP_OFFSET.
19990 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
19991 stack before exception handler returns.
19992 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
19993 the 'ERROR_CODE' for exception handler.
19995 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
19997 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
19998 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
19999 (ASM_OUTPUT_REG_POP): Ditto.
20000 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
20001 instead of asm_fprintf to output pure string.
20003 2017-07-29 Jakub Jelinek <jakub@redhat.com>
20005 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
20006 to imported_module_or_decl hook.
20007 (debug_nothing_tree_tree_tree_bool): Remove.
20008 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
20009 * debug.c (do_nothing_debug_hooks): Use
20010 debug_nothing_tree_tree_tree_bool_bool instead of
20011 debug_nothing_tree_tree_tree_bool.
20012 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
20013 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
20014 * sdbout.c (sdb_debug_hooks): Likewise.
20015 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
20016 (gen_namespace_die): Add DW_AT_export_symbols attribute if
20018 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
20019 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
20020 attribute, don't add anything.
20022 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20024 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
20025 (fold_build2_stat_loc): Likewise.
20026 (fold_build3_stat_loc): Likewise.
20027 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
20028 (fold_build1_loc): Remove macro.
20029 (fold_build2_loc): Likewise.
20030 (fold_build3_loc): Likewise.
20032 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20034 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
20035 (gimple_build_debug_bind_source_stat): Likewise.
20036 * gimple.h (gimple_build_debug_bind): Remove macro.
20037 (gimple_build_debug_bind_source): Likewise.
20039 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20041 * bitmap.c (bitmap_alloc): Adjust.
20042 (bitmap_gc_alloc): Likewise.
20043 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
20045 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20047 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
20048 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
20049 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
20050 (bitmap_gc_alloc_stat): Likewise.
20051 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
20053 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20055 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
20056 * rtl.h (shallow_copy_rtx): Remove macro.
20058 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20060 * emit-rtl.c (gen_raw_REG): Adjust.
20061 * gengenrtl.c (gendef): Likewise.
20062 * rtl.c (rtx_alloc_stat): Remove _stat from name.
20063 * rtl.h (rtx_alloc): Remove macro.
20065 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20067 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
20068 (build_tree_list_stat): Likewise.
20069 * tree.h (build_tree_list): Remove macro.
20070 (build_tree_list_vec): Likewise.
20072 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20074 * tree.c (make_vector_stat): Remove _stat from name.
20075 (build_vector_stat): Likewise.
20076 * tree.h (make_vector_stat): Remove macro.
20077 (build_vector_stat): Likewise.
20079 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20081 * tree.h (build_var_debug_value): Remove prototype.
20083 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20085 * tree.c (tree_cons_stat): Remove _stat from name.
20086 * tree.h (tree_cons): Remove macro.
20088 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20090 * tree.c (build_vl_exp_stat): Remove _stat from name.
20091 * tree.h (build_vl_exp): Remove macro.
20093 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20095 * tree.c (build_decl_stat): Remove _stat from name.
20096 * tree.h (build_decl): Remove macro.
20098 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20100 * gimple.c (gimple_build_with_ops_stat): Adjust.
20101 (gimple_alloc_stat): Remove _stat from name.
20102 * gimple.h (gimple_alloc): Remove macro.
20104 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20106 * tree.c (make_tree_vec_stat): Remove _stat from name.
20107 (grow_tree_vec_stat): Likewise.
20108 * tree.h (make_tree_vec_stat): Adjust prototype.
20109 (grow_tree_vec_stat): Likewise.
20110 (make_tree_vec): Remove macro.
20111 (grow_tree_vec): Likewise.
20113 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20115 * fold-const.c (fold_build1_stat_loc): Adjust.
20116 (fold_build2_stat_loc): Likewise.
20117 (fold_build3_stat_loc): Likewise.
20118 * tree.c (build0_stat): Remove _stat from name.
20119 (build1_stat): Likewise.
20120 (build2_stat): Likewise.
20121 (build3_stat): Likewise.
20122 (build4_stat): Likewise.
20123 (build5_stat): Likewise.
20124 * tree.h (build1_loc): Remove macro, and rename _stat function
20126 (build2_loc): Likewise.
20127 (build3_loc): Likewise.
20128 (build4_loc): Likewise.
20129 (build5_loc): Likewise.
20131 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20133 * tree.c (make_int_cst_stat): Remove _stat from name.
20134 * tree.h (make_int_cst_stat): Adjust prototype.
20135 (make_int_cst): Remove macro.
20137 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20139 * tree.c (make_tre_binfo_stat): Remove _stat from name.
20140 * tree.h (make_tree_binfo_stat): Adjust prototype.
20141 (make_tree_binfo): Remove.
20143 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20145 * tree.c (copy_node_stat): Rename to copy_node.
20146 (build_distinct_type_copy): Adjust.
20147 * tree.h (copy_node_stat): Adjust prototype.
20148 (copy_node): Remove macro.
20150 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
20152 * tree.c (make_node_stat): rename to make_node.
20153 (build_tree_list_stat): Adjust.
20154 (build0_stat): Likewise.
20155 (build2_stat): Likewise.
20156 (build3_stat): Likewise.
20157 (build4_stat): Likewise.
20158 (build5_stat): Likewise.
20159 (build_decl_stat): Likewise.
20160 * tree.h (make_node_stat): Adjust prototype.
20161 (make_node): remove macro.
20163 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
20165 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
20166 (PPC_FEATURE2_SCV): Likewise.
20167 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
20169 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20171 * config/aarch64/aarch64.c
20172 (aarch64_internal_mov_immediate): Add new special pattern.
20173 * config/aarch64/aarch64.md (*movdi_aarch64):
20174 Add reg/32bit const mov case.
20176 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20177 Richard Sandiford <richard.sandiford@linaro.org>
20179 * config/aarch64/aarch64.md (mov<mode>): Generalize.
20180 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
20181 Add integer and movi cases.
20182 (movi-split-hf-df-sf split, fp16): New.
20183 (enabled): Added TARGET_FP_F16INST.
20184 * config/aarch64/iterators.md (GPF_HF): New.
20185 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
20187 2017-07-28 Tamar Christina <tamar.christina@arm.com>
20189 * config/aarch64/aarch64.c
20190 (aarch64_simd_container_mode): Add prototype.
20191 (aarch64_expand_mov_immediate): Add HI support.
20192 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
20193 (aarch64_can_const_movi_rtx_p): New.
20194 (aarch64_preferred_reload_class):
20195 Remove restrictions of using FP registers for certain SIMD operations.
20196 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
20197 (aarch64_valid_floating_const): Add integer move validation.
20198 (aarch64_simd_imm_scalar_p): Remove.
20199 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
20200 (aarch64_legitimate_constant_p): Expand list of supported cases.
20201 * config/aarch64/aarch64-protos.h
20202 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
20203 (aarch64_reinterpret_float_as_int): New.
20204 (aarch64_simd_imm_scalar_p): Remove.
20205 * config/aarch64/constraints.md (Uvi): New.
20206 (Dd): Split into Ds and new Dd.
20207 * config/aarch64/aarch64.md (*movsi_aarch64):
20209 (*movdi_aarch64): Add SIMD mov case.
20211 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20213 * tree-predcom.c: (struct chain): Handle store-store chain in which
20214 stores for elimination only store loop invariant values.
20215 (execute_pred_commoning_chain): Ditto.
20216 (prepare_initializers_chain_store_elim): Ditto.
20217 (prepare_finalizers): Ditto.
20218 (is_inv_store_elimination_chain): New function.
20219 (initialize_root_vars_store_elim_1): New function.
20221 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20223 * tree-predcom.c: Revise general description of the pass.
20224 (enum chain_type): New enum type for store elimination.
20225 (struct chain): New field supporting store elimination.
20226 (struct component): Ditto.
20227 (dump_chain): Dump store-stores chain.
20228 (release_chain): Release resources.
20229 (split_data_refs_to_components): Compute and create component
20230 contains only stores for elimination.
20231 (get_chain_last_ref_at): New function.
20232 (make_invariant_chain): Initialization.
20233 (make_rooted_chain): Specify chain type in parameter and record it.
20234 (add_looparound_copies): Skip for store-stores chain.
20235 (determine_roots_comp): Compute type of chain and pass it to
20237 (initialize_root_vars_store_elim_2): New function.
20238 (finalize_eliminated_stores): New function.
20239 (remove_stmt): Handle store for elimination.
20240 (execute_pred_commoning_chain): Execute predictive commoning on
20241 store-store chains.
20242 (determine_unroll_factor): Skip unroll for store-stores chain.
20243 (prepare_initializers_chain_store_elim): New function.
20244 (prepare_initializers_chain): Hanlde store-store chain.
20245 (prepare_finalizers_chain, prepare_finalizers): New function.
20246 (tree_predictive_commoning_loop): Return integer value indicating
20247 if loop is unrolled or lcssa form is corrupted.
20248 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
20250 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20252 * tree-predcom.c (initialize_root): Delete.
20253 (execute_pred_commoning_chain): Initialize root vars and replace
20254 reference of non-combined chain directly, rather than call above
20257 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20259 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
20260 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
20262 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20264 * tree-predcom.c (struct chain): New field init_seq.
20265 (release_chain): Release init_seq.
20266 (prepare_initializers_chain): Record intialization stmts in above
20268 (insert_init_seqs): New function.
20269 (tree_predictive_commoning_loop): Call insert_init_seqs.
20271 2017-07-28 Bin Cheng <bin.cheng@arm.com>
20273 * tree-predcom.c (determine_roots_comp): Skip trivial components.
20275 2017-07-28 Richard Biener <rguenther@suse.de>
20277 * match.pd: Remove superfluous :c.
20278 * genmatch.c (simplify::id): Add member.
20279 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
20281 (current_id): New global.
20282 (dt_node::parent): Move from ...
20283 (dt_operand::parent): ... here. Add for_id member.
20284 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
20285 (decision_tree::find_node): Relax order requirement when
20286 merging DT_TRUE nodes to ones inbetween the current simplify
20287 and the one we try to merge with. Add diagnostic whenever
20288 we need to enforce pattern order by not merging.
20289 (decision_tree::insert): Set current_id.
20290 (decision_tree::print_node): Dump parent node and for_id.
20291 (parser::last_id): Add member.
20292 (parser::push_simplify): Assign unique id.
20293 (parser::parser): Initialize last_id.
20295 2017-07-28 Martin Liska <mliska@suse.cz>
20298 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
20299 gimple_build_debug_bind.
20301 2017-07-28 Richard Biener <rguenther@suse.de>
20303 PR tree-optimization/81502
20304 * match.pd: Add pattern combining BIT_INSERT_EXPR with
20306 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
20308 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
20309 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
20310 for BIT_FIELD_REF args.
20311 * fold-const.c (make_bit_field_ref): Likewise.
20312 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
20314 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20317 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
20318 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
20319 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
20320 Or it into SANITIZER_UNDEFINED.
20321 * ubsan.c: Include gimple-fold.h and varasm.h.
20322 (ubsan_expand_ptr_ifn): New function.
20323 (instrument_pointer_overflow): New function.
20324 (maybe_instrument_pointer_overflow): New function.
20325 (instrument_object_size): Formatting fix.
20326 (pass_ubsan::execute): Call instrument_pointer_overflow
20327 and maybe_instrument_pointer_overflow.
20328 * internal-fn.c (expand_UBSAN_PTR): New function.
20329 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
20330 * sanitizer.def (__ubsan_handle_pointer_overflow,
20331 __ubsan_handle_pointer_overflow_abort): New builtins.
20332 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
20333 * internal-fn.def (UBSAN_PTR): New internal function.
20334 * opts.c (sanitizer_opts): Add pointer-overflow.
20335 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
20336 * fold-const.c (build_range_check): Compute pointer range check in
20337 integral type if pointer arithmetics would be needed. Formatting
20340 2017-07-28 Martin Liska <mliska@suse.cz>
20343 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
20344 parameters that are of a variable-length.
20346 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
20348 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
20350 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
20351 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
20352 (CRT_CALL_STATIC_FUNCTION): Likewise.
20353 (ASM_DEFAULT_SPEC): New define.
20354 (ASM_SPEC32): Likewise.
20355 (ASM_SPEC64): Likewise.
20356 (ASM_SPEC_COMMON): Likewise.
20357 (ASM_SPEC): Likewise.
20358 (INVALID_64BIT): Likewise.
20359 (LINK_OS_DEFAULT_SPEC): Likewise.
20360 (LINK_OS_SPEC32): Likewise.
20361 (LINK_OS_SPEC64): Likewise.
20362 (POWERPC_LINUX): Likewise.
20363 (PTRDIFF_TYPE): Likewise.
20364 (RESTORE_FP_PREFIX): Likewise.
20365 (RESTORE_FP_SUFFIX): Likewise.
20366 (SAVE_FP_PREFIX): Likewise.
20367 (SAVE_FP_SUFFIX): Likewise.
20368 (SIZE_TYPE): Likewise.
20369 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
20370 (TARGET_64BIT): Likewise.
20371 (TARGET_64BIT): Likewise.
20372 (TARGET_AIX): Likewise.
20373 (WCHAR_TYPE_SIZE): Likewise.
20374 (WCHAR_TYPE): Undefine.
20375 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
20376 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
20377 (CPP_OS_RTEMS_SPEC): Delete.
20378 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
20379 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
20381 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
20383 2017-07-28 Jakub Jelinek <jakub@redhat.com>
20385 PR tree-optimization/81578
20386 * tree-parloops.c (build_new_reduction): Bail out if
20387 reduction_code isn't one of the standard OpenMP reductions.
20388 Move the details printing after that decision.
20390 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20392 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
20393 related to reload_in_progress.
20394 (splat_input_operand): Likewise.
20395 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
20397 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
20399 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
20400 (TARGET_INSTANTIATE_DECLS): Likewise.
20401 (legitimate_indexed_address_p): Delete reload_in_progress code.
20402 (rs6000_debug_legitimate_address_p): Likewise.
20403 (rs6000_eliminate_indexed_memrefs): Likewise.
20404 (rs6000_emit_le_vsx_store): Likewise.
20405 (rs6000_emit_move_si_sf_subreg): Likewise.
20406 (rs6000_emit_move): Likewise.
20407 (register_to_reg_type): Likewise.
20408 (rs6000_pre_atomic_barrier): Likewise.
20409 (rs6000_machopic_legitimize_pic_address): Likewise.
20410 (rs6000_allocate_stack_temp): Likewise.
20411 (rs6000_address_for_fpconvert): Likewise.
20412 (rs6000_address_for_altivec): Likewise.
20413 (rs6000_secondary_memory_needed_rtx): Delete function.
20414 (rs6000_check_sdmode): Likewise.
20415 (rs6000_alloc_sdmode_stack_slot): Likewise.
20416 (rs6000_instantiate_decls): Likewise.
20417 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
20418 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
20419 Delete reload_in_progress.
20420 (*vec_reload_and_plus_<mptrsize>): Likewise.
20421 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
20422 (vsx_div_v2di): Likewise.
20423 (vsx_udiv_v2di): Likewise.
20425 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20427 * config/rs6000/rs6000.opt (mlra): Replace with stub.
20428 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
20429 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
20430 (rs6000_debug_reg_global): Delete print of LRA status.
20431 (rs6000_option_override_internal): Delete dead LRA related code.
20432 (rs6000_lra_p): Delete function.
20433 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
20435 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
20437 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
20438 * config/riscv/rtems.h: New file.
20440 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20441 Sudakshina Das <sudi.das@arm.com>
20443 * config/aarch64/aarch64.md
20444 (define_split for and<mode>3nr_compare): Move
20445 non aarch64_logical_operand to a register.
20446 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
20447 register immediate operand to a register.
20448 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
20450 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
20452 PR middle-end/81564
20453 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
20455 2017-07-27 Richard Biener <rguenther@suse.de>
20457 PR tree-optimization/81573
20458 PR tree-optimization/81494
20459 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
20460 multi defuse cycle case.
20462 2017-07-27 Richard Biener <rguenther@suse.de>
20464 PR tree-optimization/81571
20465 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
20468 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
20470 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
20471 earlier and only if MASK_FPU is set. Adjust formatting.
20473 2017-07-27 Martin Liska <mliska@suse.cz>
20475 * opt-functions.awk: Add validation of value of Init.
20476 * optc-gen.awk: Pass new argument.
20478 2017-07-27 Martin Liska <mliska@suse.cz>
20480 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
20481 Fix wrong condition.
20483 2017-07-27 Martin Liska <mliska@suse.cz>
20485 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
20486 BBs and edges seen by autoFDO.
20488 2017-07-27 Richard Biener <rguenther@suse.de>
20490 PR tree-optimization/81502
20491 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
20492 with incompatible but same sized type.
20493 (execute_update_addresses_taken): Likewise.
20495 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
20497 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
20498 flag_tree_loop_vectorize rather than flag_tree_vectorize.
20500 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20503 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
20504 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
20505 Change s_operand to memory_operand.
20507 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
20509 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
20510 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
20511 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
20512 Emit instructions rather than returning an expression. Handle TFmode
20513 and KFmode by casting to TImode.
20514 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
20515 (rs6000_emit_le_vsx_store): Likewise.
20516 * config/rs6000/vsx.md (VSX_TI): New iterator.
20517 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
20518 (*vsx_le_undo_permute_<mode>): Likewise.
20519 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
20520 emit the split sequence.
20521 (*vsx_le_perm_store_<mode>): Likewise.
20523 2017-07-27 Jakub Jelinek <jakub@redhat.com>
20525 PR tree-optimization/81555
20526 PR tree-optimization/81556
20527 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
20528 if true, force CHANGED for the recursive invocation.
20529 (reassociate_bb): Remember original length of ops array, pass
20530 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
20532 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
20533 attributes for noipa attribute. For naked attribute use
20534 lookup_attribute first before lookup_attribute_spec.
20535 * final.c (rest_of_handle_final): Disable IPA RA for functions with
20537 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
20538 for functions with noipa attribute.
20539 (cgraph_externally_visible_p): Return true for functions with noipa
20541 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
20542 for functions with noipa attribute.
20543 * doc/extend.texi: Document noipa function attribute.
20544 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
20545 also for functions with noipa attribute.
20546 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
20548 2017-07-26 Andrew Pinski <apinski@cavium.com>
20550 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
20551 vec_unalign_load_cost and vec_unalign_store_cost.
20553 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
20555 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
20556 -mvsx-small-integer option.
20557 (ISA_3_0_MASKS_IEEE): Likewise.
20558 (OTHER_VSX_VECTOR_MASKS): Likewise.
20559 (POWERPC_MASKS): Likewise.
20560 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
20561 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
20562 code, only testing for DImode being allowed in non-VSX floating
20564 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
20565 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
20567 (rs6000_option_override_internal): Delete -mvsx-small-integer.
20568 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
20569 TARGET_P8_VECTOR test.
20570 (rs6000_secondary_reload_simple_move): Likewise.
20571 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
20572 since TARGET_P9_VECTOR was already tested.
20573 (rs6000_opt_masks): Remove -mvsx-small-integer.
20574 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
20575 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20577 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
20578 test for TARGET_VEXTRACTUB was used, and that uses
20580 (p9 extract splitter): Likewise.
20581 (vsx_extract_<mode>_di_p9): Likewise.
20582 (vsx_extract_<mode>_store_p9): Likewise.
20583 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
20584 for TARGET_P9_VECTOR was used. Delete code that is now dead with
20585 the elimination of TARGET_VSX_SMALL_INTEGER.
20586 (vsx_extract_<mode>_p8): Likewise.
20587 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
20588 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
20589 (vsx_set_<mode>_p9): Likewise.
20590 (vsx_set_v4sf_p9): Likewise.
20591 (vsx_set_v4sf_p9_zero): Likewise.
20592 (vsx_insert_extract_v4sf_p9): Likewise.
20593 (vsx_insert_extract_v4sf_p9_2): Likewise.
20594 * config/rs6000/rs6000.md (sign extend splitter): Change
20595 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
20596 (floatsi<mode>2_lfiwax_mem): Likewise.
20597 (floatunssi<mode>2_lfiwzx_mem): Likewise.
20598 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
20599 since a test for TARGET_P9_VECTOR was used.
20600 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20601 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
20602 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20603 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20604 TARGET_P8_VECTOR test.
20605 (fix_trunc<mode>si2_stfiwx): Likewise.
20606 (fix_trunc<mode>si2_internal): Likewise.
20607 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
20608 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20610 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20611 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
20612 TARGET_P8_VECTOR test.
20613 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20614 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
20615 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
20617 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
20618 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
20619 since a test for TARGET_P9_VECTOR was used.
20620 (splitter for loading small constants): Likewise.
20622 2017-07-26 Andrew Pinski <apinski@cavium.com>
20624 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
20627 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
20630 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
20631 (fp_valid_at): Likewise.
20633 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20635 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
20636 (qdf24xx_addrcost_table): Likewise.
20637 (cortexa57_tunings): Update to use generic_branch_cost.
20638 (cortexa72_tunings): Likewise.
20639 (cortexa73_tunings): Likewise.
20640 (qdf24xx_tunings): Likewise.
20642 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
20644 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
20645 (thunderx2t99_branch_cost): Likewise.
20646 (cortexa35_tunings): Update to use generic_branch_cost.
20647 (cortexa53_tunings): Likewise.
20648 (cortexa57_tunings): Likewise.
20649 (cortexa72_tunings): Likewise.
20650 (cortexa73_tunings): Likewise.
20651 (thunderx2t99_tunings): Likewise.
20653 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20655 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
20656 (sparc_option_override): Honour MASK_FSMULD.
20657 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
20658 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
20659 * config/sparc/sparc.opt (mfsmuld): New option.
20660 * doc/invoke.texi (mfsmuld): Document option.
20662 2017-07-26 Marek Polacek <polacek@redhat.com>
20664 PR middle-end/70992
20665 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
20667 2017-07-26 Richard Biener <rguenther@suse.de>
20669 * gimple-match-head.c (do_valueize): Return OP if valueize
20671 (get_def): New helper to get at the def stmt of a SSA name
20672 if valueize allows.
20673 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
20674 do_valueize to get at the def stmt.
20675 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
20677 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
20679 PR middle-end/46932
20680 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
20682 2017-07-26 Martin Liska <mliska@suse.cz>
20685 * function.c (expand_function_start): Make expansion of
20686 nonlocal_goto_save_area after parm_birth_insn.
20688 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
20690 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
20691 from all CPU target flags enable members.
20693 2017-07-26 Richard Biener <rguenther@suse.de>
20695 * genmatch.c (dt_simplify::gen): Make iterator vars const.
20696 (decision_tree::gen): Make 'type' const.
20697 (write_predicate): Likewise.
20699 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20701 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
20702 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
20703 (rs6000_option_override_internal): Likewise.
20704 (rs6000_expand_vector_set): Likewise.
20705 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
20706 (TARGET_UPPER_REGS_SF): Likewise.
20707 (TARGET_UPPER_REGS_DI): Likewise.
20708 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
20709 (TARGET_DIRECT_MOVE_64BIT): Likewise.
20710 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
20711 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
20712 (Splitters for DI constants in Altivec registers): Likewise.
20713 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
20714 (vsx_set_v4sf_p9): Likewise.
20715 (vsx_set_v4sf_p9_zero): Likewise.
20716 (vsx_insert_extract_v4sf_p9): Likewise.
20717 (vsx_insert_extract_v4sf_p9_2): Likewise.
20719 2017-07-25 Carl Love <cel@us.ibm.com>
20721 * doc/extend.texi: Update the built-in documentation file for the
20722 existing built-in functions
20723 vector signed char vec_cnttz (vector signed char);
20724 vector unsigned char vec_cnttz (vector unsigned char);
20725 vector signed short vec_cnttz (vector signed short);
20726 vector unsigned short vec_cnttz (vector unsigned short);
20727 vector signed int vec_cnttz (vector signed int);
20728 vector unsigned int vec_cnttz (vector unsigned int);
20729 vector signed long long vec_cnttz (vector signed long long);
20730 vector unsigned long long vec_cnttz (vector unsigned long long);
20732 2017-07-25 Andrew Pinski <apinski@cavium.com>
20734 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
20735 accesses where the use is for the first operand of a BIT_INSERT.
20737 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20740 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20741 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20743 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
20745 * config/i386/gstabs.h: Delete.
20746 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
20748 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
20750 * config/i386/i386.c (ix86_decompose_address): Do not check for
20751 register RTX when looking at index_reg or base_reg.
20752 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
20754 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
20756 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
20757 to update EH info here.
20759 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20761 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
20763 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20765 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
20767 2017-07-25 Torsten Duwe <duwe@suse.de>
20769 * common.opt: Introduce -fpatchable-function-entry
20770 command line option, and its variables function_entry_patch_area_size
20771 and function_entry_patch_area_start.
20772 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
20773 including a two-value parser.
20774 * target.def (print_patchable_function_entry): New target hook.
20775 * targhooks.h (default_print_patchable_function_entry): New function.
20776 * targhooks.c (default_print_patchable_function_entry): Likewise.
20777 * toplev.c (process_options): Switch off IPA-RA if
20778 patchable function entries are being generated.
20779 * varasm.c (assemble_start_function): Look at the
20780 patchable-function-entry command line switch and current
20781 function attributes and maybe generate NOP instructions by
20782 calling the print_patchable_function_entry hook.
20783 * doc/extend.texi: Document patchable_function_entry attribute.
20784 * doc/invoke.texi: Document -fpatchable_function_entry
20785 command line option.
20786 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
20788 * doc/tm.texi: Re-generate.
20790 2017-07-25 Jakub Jelinek <jakub@redhat.com>
20793 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
20794 TARGET_AVX512DQ rather than TARGET_AVX512BW.
20796 2017-07-25 Tamar Christina <tamar.christina@arm.com>
20798 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
20800 2017-07-25 Richard Biener <rguenther@suse.de>
20802 PR tree-optimization/81455
20803 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
20804 not walk in cycles when looking for guards.
20806 2017-07-25 Richard Biener <rguenther@suse.de>
20808 PR tree-optimization/81529
20809 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
20810 when optimizing backedge uses.
20812 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
20814 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
20816 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
20817 to dl_section_ref. On AIX, append an expression to subtract
20818 the size of the section length to dl_section_ref.
20820 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
20822 * configure.ac: If any of the config.* scripts fail, exit 1.
20823 * configure: Regenerate.
20825 2017-07-25 Richard Biener <rguenther@suse.de>
20827 PR middle-end/81546
20828 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
20829 of immediate uses, be more verbose on errors.
20831 2017-07-25 Richard Biener <rguenther@suse.de>
20833 PR tree-optimization/81510
20834 * tree-vect-loop.c (vect_is_simple_reduction): When the
20835 reduction stmt is not inside the loop bail out.
20837 2017-07-25 Richard Biener <rguenther@suse.de>
20839 PR tree-optimization/81303
20840 * tree-vect-loop-manip.c (vect_loop_versioning): Build
20841 profitability check against LOOP_VINFO_NITERSM1.
20843 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
20845 * domwalk.c (cmp_bb_postorder): Simplify.
20846 (sort_bbs_postorder): New function. Use it...
20847 (dom_walker::walk): ...here to optimize common cases.
20849 2017-07-25 Martin Liska <mliska@suse.cz>
20852 * ipa-visibility.c (function_and_variable_visibility): Make the
20853 redirection just on target that supports aliasing.
20854 Fix GNU coding style.
20856 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
20859 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
20861 * config/arm/rtems.h: Likewise.
20862 * config/bfin/rtems.h: Likewise.
20863 * config/i386/rtemself.h: Likewise.
20864 * config/lm32/rtems.h: Likewise.
20865 * config/m32c/rtems.h: Likewise.
20866 * config/m68k/rtemself.h: Likewise.
20867 * config/microblaze/rtems.h: Likewise.
20868 * config/mips/rtems.h: Likewise.
20869 * config/moxie/rtems.h: Likewise.
20870 * config/nios2/rtems.h: Likewise.
20871 * config/powerpcspe/rtems.h: Likewise.
20872 * config/rs6000/rtems.h: Likewise.
20873 * config/rtems.h: Likewise.
20874 * config/sh/rtems.h: Likewise.
20875 * config/sh/rtemself.h: Likewise.
20876 * config/sparc/rtemself.h: Likewise.
20878 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20881 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
20882 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
20883 * tree-ssa-structalias.c (alias_get_name): Same.
20885 2017-07-25 Bin Cheng <bin.cheng@arm.com>
20888 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
20889 instructions if no du chain is found.
20891 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
20893 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
20895 2017-07-25 Richard Biener <rguenther@suse.de>
20897 PR middle-end/81505
20898 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
20901 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
20903 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
20904 upper-regs options.
20905 (ISA_2_7_MASKS_SERVER): Likewise.
20906 (ISA_3_0_MASKS_IEEE): Likewise.
20907 (OTHER_P8_VECTOR_MASKS): Likewise.
20908 (OTHER_VSX_VECTOR_MASKS): Likewise.
20909 (POWERPC_MASKS): Likewise.
20910 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
20911 duplicate list of options.
20912 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
20913 explicit -mupper-regs options.
20914 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
20915 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
20916 alias for -mupper-regs-df.
20917 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20918 (rs6000_init_hard_regno_mode_ok): Likewise.
20919 (rs6000_option_override_internal): Likewise.
20920 (rs6000_opt_masks): Likewise.
20921 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
20922 options in terms of whether -mvsx or -mpower8-vector was used.
20923 (TARGET_UPPER_REGS_DI): Likewise.
20924 (TARGET_UPPER_REGS_SF): Likewise.
20925 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
20926 -mupper-regs-* options.
20928 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20930 * passes.c (emergency_dump_function): Print some empty lines and a
20931 header before the RTL dump.
20933 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
20935 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
20937 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
20940 * config/aarch64/aarch64.c (aarch64_classify_symbol):
20941 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
20943 2017-07-24 Carl Love <cel@us.ibm.com>
20945 * config/rs6000/rs6000-c.c: Add support for built-in functions
20946 vector float vec_extract_fp32_from_shorth (vector unsigned short);
20947 vector float vec_extract_fp32_from_shortl (vector unsigned short);
20948 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
20949 vec_extract_fp_from_shortl): Add defines for the two builtins.
20950 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
20951 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
20953 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
20954 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
20955 * doc/extend.texi: Update the built-in documentation file for the
20956 new built-in function.
20958 2017-07-24 Jakub Jelinek <jakub@redhat.com>
20961 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
20963 * doc/generic.texi: Likewise.
20964 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
20965 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
20967 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
20969 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
20970 (aarch64_mls_elt_merge<mode>): Likewise.
20972 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
20974 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
20975 having __cxa_atexit.
20977 2017-07-23 Michael Collison <michael.collison@arm.com>
20979 * config/arm/arm.c (arm_option_override): Deprecate
20980 use of -mstructure-size-boundary.
20981 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
20982 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
20984 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20987 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
20988 Reduce cost estimate for direct moves.
20990 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
20993 * config/i386/i386.c (ix86_option_override_internal): Disable
20994 BMI, BMI2 and TBM instructions for -m16.
20996 2017-07-21 Carl Love <cel@us.ibm.com>
20998 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20999 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21000 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21001 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21002 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21003 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21004 VMULOSW): New enum "unspec" values.
21005 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21006 altivec_vmulosw): New patterns.
21007 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21008 VMULOSW): Add definitions.
21010 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
21012 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
21013 (qdf24xx): Likewise.
21014 * config/aarch64/aarch64-options-extensions.def (rdma); New.
21015 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
21016 (AARCH64_FL_V8_1): Renumber.
21017 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
21018 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
21019 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
21020 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
21021 rdma to feature modifiers list.
21023 2017-07-21 Yury Gribov <tetra2005@gmail.com>
21025 PR middle-end/56727
21026 * ipa-visibility (function_and_variable_visibility): Convert
21027 recursive PLT call to direct call if appropriate.
21029 2017-07-21 Andrew Pinski <apinski@cavium.com>
21031 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
21032 operand 1 to see if the types precision matches.
21033 * fold-const.c (operand_equal_p): Likewise.
21035 2017-07-21 Richard Biener <rguenther@suse.de>
21037 PR tree-optimization/81303
21038 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
21039 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
21040 (vect_peeling_hash_get_lowest_cost): Adjust.
21041 (vect_enhance_data_refs_alignment): Likewise. Use
21042 vect_get_peeling_costs_all_drs to compute the penalty for no
21043 peeling to match up costs.
21045 2017-07-21 Richard Biener <rguenther@suse.de>
21047 PR tree-optimization/81500
21048 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
21049 we didn't identify a reduction path.
21051 2017-07-21 Tom de Vries <tom@codesourcery.com>
21052 Cesar Philippidis <cesar@codesourcery.com>
21054 PR gcov-profile/81442
21055 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
21058 2017-07-21 Tom de Vries <tom@codesourcery.com>
21061 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
21063 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
21064 nvptx_override_options_after_change.
21066 2017-07-21 Ulrich Drepper <drepper@redhat.com>
21068 * dwarf2out.c (output_file_names): Avoid double testing for
21069 dwarf_version >= 5.
21071 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
21073 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
21075 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21077 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
21079 (try_crossjump_to_edge): Do not punt on partitioned functions.
21081 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
21083 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
21084 Put all BBs reachable only via paths crossing cold region to cold
21086 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
21088 2016-07-21 Richard Biener <rguenther@suse.de>
21090 PR tree-optimization/81303
21091 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
21092 into account prologue and epilogue iterations when raising
21093 min_profitable_iters to sth at least covering one vector iteration.
21095 2017-07-21 Tamar Christina <tamar.christina@arm.com>
21097 * config/arm/arm.c (arm_test_cpu_arch_dat):
21100 2017-07-20 Nathan Sidwell <nathan@acm.org>
21102 Remove TYPE_METHODS.
21103 * tree.h (TYPE_METHODS): Delete.
21104 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
21105 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
21106 (dbxout_type_methods): Scan TYPE_FIELDS.
21107 (dbxout_type): Don't check TYPE_METHODS here.
21108 * function.c (use_register_for_decl): Always ignore register for
21109 class types when not optimizing.
21110 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
21111 * tree.c (free_lang_data_in_type): Stitch out member functions and
21112 templates from TYPE_FIELDS.
21113 (build_distinct_type_copy, verify_type_variant,
21114 verify_type): Member fns are on TYPE_FIELDS.
21115 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
21116 * tree-pretty-print.c (dump_generic_node): Likewise.
21118 2017-07-20 Jakub Jelinek <jakub@redhat.com>
21121 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
21122 V2TImode and V4TImode.
21123 (ix86_expand_vector_extract): Likewise.
21124 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
21125 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
21126 (ssescalarmode): Handle V4TImode and V2TImode.
21127 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
21128 (*vec_extractv2ti, *vec_extractv4ti): New insns.
21129 (VEXTRACTI128_MODE): New mode iterator.
21130 (splitter for *vec_extractv?ti first element): New.
21131 (VEC_INIT_MODE): New mode iterator.
21132 (vec_init<mode>): Consolidate 3 expanders into one using
21133 VEC_INIT_MODE mode iterator.
21135 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21137 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
21138 non_spilled_static_chain_regno_p.
21140 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
21142 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
21144 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
21146 * bb-reorder.c (connect_traces): Allow copying of blocks within
21149 2017-07-20 Richard Biener <rguenther@suse.de>
21151 * gimple.h (gimple_phi_result): Add gphi * overload.
21152 (gimple_phi_result_ptr): Likewise.
21153 (gimple_phi_arg): Likewise. Adjust index assert to only
21154 allow actual argument accesses rather than all slots available
21156 (gimple_phi_arg_def): Add gphi * overload.
21157 * tree-phinodes.c (make_phi_node): Initialize only actual
21159 (resize_phi_node): Clear memory not covered by old node,
21160 do not initialize excess argument slots.
21161 (reserve_phi_args_for_new_edge): Initialize new argument slot
21164 2017-07-20 Bin Cheng <bin.cheng@arm.com>
21166 PR tree-optimization/81388
21168 2016-07-21 Bin Cheng <bin.cheng@arm.com>
21170 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
21171 by removing computation of may_be_zero.
21173 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21174 Tom de Vries <tom@codesourcery.com>
21176 PR middle-end/81030
21177 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
21178 when gimple level profile disagrees with what RTL expander did.
21180 2017-07-20 Richard Biener <rguenther@suse.de>
21182 PR tree-optimization/61171
21183 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
21184 (vect_analyze_stmt): Add slp instance parameter.
21185 (vectorizable_reduction): Likewise.
21186 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
21187 (vect_is_simple_reduction): Deal with chains not detected
21188 as SLP reduction chain, specifically not properly associated
21189 chains containing a mix of plus/minus.
21190 (get_reduction_op): Remove.
21191 (get_initial_defs_for_reduction): Simplify, pass in whether
21192 this is a reduction chain, pass in the SLP node for the PHIs.
21193 (vect_create_epilog_for_reduction): Get the SLP instance as
21195 (vectorizable_reduction): Get the SLP instance as arg.
21196 During analysis remember the SLP node with the PHIs in the
21197 instance. Simplify getting at the vectorized reduction PHIs.
21198 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
21199 through SLP instance.
21200 (vect_slp_analyze_operations): Likewise.
21201 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
21202 (vect_transform_stmt): Likewise.
21204 2017-07-20 Tom de Vries <tom@codesourcery.com>
21206 PR tree-optimization/81489
21207 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
21208 read of phi arg location to before loop that modifies phi.
21210 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21212 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
21215 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21217 PR middle-end/81331
21218 * except.c (execute): Fix ordering issue.
21220 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21222 PR rtl-optimization/81423
21223 * combine.c (make_compound_operation_int): Don't try to optimize
21224 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
21226 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
21228 PR rtl-optimization/81423
21229 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
21230 with a constant that is -1 in the truncated to mode.
21232 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21234 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
21235 (determine_unlikely_bbs): ... here.
21236 * predict.h (propagate_unlikely_bbs_forward): Declare.
21237 * cfgexpand.c (pass_expand::execute): Use it.
21238 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
21240 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
21241 propagate_unlikely_bbs_forward.
21243 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
21245 PR middle-end/81331
21246 * except.c (maybe_add_nop_after_section_switch): New function.
21249 2017-07-19 Tom de Vries <tom@codesourcery.com>
21251 * gimple.h (gimple_phi_set_arg): Make assert more strict.
21253 2017-07-19 Tom de Vries <tom@codesourcery.com>
21255 * gimple.h (gimple_phi_arg): Make assert more strict.
21257 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
21259 * config.gcc (powerpc*-*-*): Add mmintrin.h.
21260 * config/rs6000/mmintrin.h: New file.
21261 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
21263 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21265 PR tree-optimization/81346
21266 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
21268 2017-07-19 Tom de Vries <tom@codesourcery.com>
21270 * config/nvptx/nvptx.md (VECIM): Add V2DI.
21272 2017-07-19 Tom de Vries <tom@codesourcery.com>
21274 * config/nvptx/nvptx-modes.def: Add V2DImode.
21275 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
21276 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
21277 (nvptx_output_mov_insn): Handle lack of mov.b128.
21278 (nvptx_print_operand): Handle 'H' and 'L' codes.
21279 (nvptx_vector_mode_supported): Allow V2DImode.
21280 (nvptx_preferred_simd_mode): New function.
21281 (nvptx_data_alignment): New function.
21282 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
21283 nvptx_preferred_simd_mode.
21284 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
21286 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
21288 2017-07-19 Tom de Vries <tom@codesourcery.com>
21290 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
21291 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
21292 (nvptx_vector_mode_supported): New function. Allow V2SImode.
21293 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
21294 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
21295 (mov<VECIM>_insn): New define_insn.
21296 (define_expand "mov<VECIM>): New define_expand.
21298 2017-07-19 Tom de Vries <tom@codesourcery.com>
21300 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
21302 2017-07-19 Jakub Jelinek <jakub@redhat.com>
21304 PR tree-optimization/81346
21305 * fold-const.h (fold_div_compare, range_check_type): Declare.
21306 * fold-const.c (range_check_type): New function.
21307 (build_range_check): Use range_check_type.
21308 (fold_div_compare): No longer static, rewritten into
21309 a match.pd helper function.
21310 (fold_comparison): Don't call fold_div_compare here.
21311 * match.pd (X / C1 op C2): New optimization using fold_div_compare
21312 as helper function.
21314 2017-07-19 Nathan Sidwell <nathan@acm.org>
21316 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
21317 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
21318 * tree.c (find_decls_types_r, verify_type): Use
21319 TYPE_{MIN,MAX}_VALUE_RAW.
21320 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
21321 (hash_tree): Likewise.
21322 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21324 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21327 2017-07-18 Tom de Vries <tom@codesourcery.com>
21329 PR middle-end/81464
21330 * omp-expand.c (expand_omp_for_static_chunk): Handle
21331 equal-argument loop exit phi.
21333 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
21336 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
21337 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
21338 operand 2 predicate.
21339 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
21340 operand 2 predicate.
21341 (ror,rol -> rorx splitters): Use const_int_operand as
21342 operand 2 predicate.
21344 2017-06-18 Richard Biener <rguenther@suse.de>
21346 PR tree-optimization/81410
21347 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
21348 the gap in the ! slp_perm SLP case after each group.
21350 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21352 PR middle-end/81463
21353 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
21356 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
21358 PR middle-end/81462
21359 * predict.c (set_even_probabilities): Cleanup; do not affect
21360 probabilities that are already known.
21361 (combine_predictions_for_bb): Call even when count is set.
21363 2017-07-18 Nathan Sidwell <nathan@acm.org>
21365 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
21368 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21371 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
21372 optimization for loop niter analysis.
21374 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
21377 * config/avr/avr.c (avr_optimize_casesi): Don't use
21378 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
21380 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
21382 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
21383 body_cost_vec from _vect_peel_extended_info.
21384 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
21385 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
21388 2017-07-18 Bin Cheng <bin.cheng@arm.com>
21390 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
21392 2017-07-18 Richard Biener <rguenther@suse.de>
21394 PR tree-optimization/80620
21395 PR tree-optimization/81403
21396 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
21397 info when re-using a VN table entry.
21399 2017-07-18 Richard Biener <rguenther@suse.de>
21401 PR tree-optimization/81418
21402 * tree-vect-loop.c (vectorizable_reduction): Properly compute
21403 vectype_in. Verify that with lane-reducing reduction operations
21404 we have a single def-use cycle.
21406 2017-07-17 Carl Love <cel@us.ibm.com>
21408 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
21410 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21411 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
21412 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
21413 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
21414 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
21415 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
21416 VMULOSW): New enum "unspec" values.
21417 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21418 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
21419 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
21420 altivec_vmulosw): New patterns.
21421 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
21422 VMULOSW): Add definitions.
21424 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
21426 * config/alpha/alpha.c: Include predict.h.
21428 2017-07-17 Yury Gribov <tetra2005@gmail.com>
21430 * tree-vrp.c (compare_assert_loc): Fix comparison function
21431 to return predictable results.
21433 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21435 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
21437 (subdi3): Likewise.
21438 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
21439 * doc/invoke.texi (mexpand-adddi): Update text.
21441 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21443 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
21444 that also clobbers the CC register. The old expand code is moved
21446 (*arc_clzsi2): ... here.
21447 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
21448 the CC register. The old expand code is moved to ...
21449 (arc_ctzsi2): ... here.
21451 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21453 * config/arc/arc.opt (mindexed-loads): Use initial value
21454 TARGET_INDEXED_LOADS_DEFAULT.
21455 (mauto-modify-reg): Use initial value
21456 TARGET_AUTO_MODIFY_REG_DEFAULT.
21457 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21458 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21459 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
21460 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
21462 2017-07-17 Martin Liska <mliska@suse.cz>
21465 * opts.c (finish_options): Do not allow -fgnu-tm
21466 w/ -fsanitize={kernel-,}address. Say sorry.
21468 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21471 * tree-loop-distribution.c (classify_partition): Only assert on
21472 numer of iterations.
21473 (merge_dep_scc_partitions): Delete prameter. Update function call.
21474 (distribute_loop): Remove code handling loop with unknown niters.
21475 (pass_loop_distribution::execute): Skip loop with unknown niters.
21477 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21480 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
21481 function sort_partitions_by_post_order.
21483 2017-07-17 Bin Cheng <bin.cheng@arm.com>
21485 PR tree-optimization/81374
21486 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
21487 the max index of basic blocks, rather than number of basic blocks.
21489 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21491 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
21493 (arc_legitimate_pic_operand_p): Likewise.
21494 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
21496 (arc_needs_pcl_p): Likewise.
21497 (arc_legitimate_pc_offset_p): Likewise.
21498 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
21499 function is also used in constrains.md.
21500 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
21501 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
21502 PLUS. Only return true/false in known cases, otherwise assert.
21503 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
21504 is already called in arc_legitimate_constant_p.
21505 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
21507 (LEGITIMATE_PIC_OPERAND_P): Use
21508 arc_raw_symbolic_reference_mentioned_p function.
21509 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
21514 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
21515 Andrew Burgess <andrew.burgess@embecosm.com>
21517 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
21518 (arc_return_address_register): New function.
21519 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
21520 (arc_handle_fndecl_attribute): Add naked attribute.
21521 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
21522 (TARGET_WARN_FUNC_RETURN): Likewise.
21523 (arc_allocate_stack_slots_for_args): New function.
21524 (arc_warn_func_return): Likewise.
21525 (machine_function): Change type fn_type.
21526 (arc_compute_function_type): Consider new naked function type,
21527 change function return type.
21528 (arc_must_save_register): Adapt to handle new
21529 arc_compute_function_type's return type.
21530 (arc_expand_prologue): Likewise.
21531 (arc_expand_epilogue): Likewise.
21532 (arc_return_address_regs): Delete.
21533 (arc_return_address_register): New function.
21534 (arc_epilogue_uses): Use above function.
21535 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
21536 (arc_function_type): Change encoding, add naked type.
21537 (ARC_INTERRUPT_P): Change to handle the new encoding.
21538 (ARC_FAST_INTERRUPT_P): Likewise.
21539 (ARC_NORMAL_P): Define.
21540 (ARC_NAKED_P): Likewise.
21541 (arc_compute_function_type): Delete prototype.
21542 * config/arc/arc.md (in_ret_delay_slot): Use
21543 arc_return_address_register function.
21544 (simple_return): Likewise.
21545 (p_return_i): Likewise.
21547 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21549 PR tree-optimization/81428
21550 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
21551 can't be built for those types.
21553 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21555 Remove stuff dead since r239246.
21557 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
21558 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
21559 (avr_inform_devices): Remove dead stuff.
21561 2017-07-17 Tamar Christina <tamar.christina@arm.com>
21563 * config/arm/arm_neon.h: Fix softp typo.
21565 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21567 PR tree-optimization/81365
21568 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
21569 aggregate moves onto bb predecessor edges, make sure there are no
21570 loads that could alias the lhs in between the start of bb and the
21573 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
21576 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
21577 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
21578 [LSHIFTRT, outer_code = TRUNCATE]: Same.
21580 2017-07-17 Jakub Jelinek <jakub@redhat.com>
21582 PR tree-optimization/81396
21583 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
21584 (init_symbolic_number): Initialize it to 1.
21585 (perform_symbolic_merge): Add n_ops from both operands into the new
21587 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
21588 without base_addr as useless if they need more than one operation.
21589 (bswap_replace): Handle !bswap case for NULL base_addr.
21591 2017-07-17 Tom de Vries <tom@codesourcery.com>
21594 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
21597 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
21599 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
21600 conditional builtin define __FIX_LEON3FT_B2BST.
21602 2017-07-17 Daniel Cederman <cederman@gaisler.com>
21604 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
21605 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
21608 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21610 PR rtl-optimization/81424
21611 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
21612 to remove potential trapping from operands if -fnon-call-exceptions.
21614 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21616 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
21617 profile_proability for scalling.
21618 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
21620 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21622 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
21624 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21626 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
21627 fixpoint arithmetics.
21629 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21631 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
21632 fixpoint arithmetics.
21634 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21636 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
21637 fixpoint arithmetics.
21639 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
21641 * profile-count.h (profile_probability::from_reg_br_prob_note,
21642 profile_probability::to_reg_br_prob_note): New functions.
21643 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
21644 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
21645 * predict.c (probability_reliable_p): Update.
21646 (edge_probability_reliable_p): Update.
21647 (br_prob_note_reliable_p): Update.
21648 (invert_br_probabilities): Update.
21649 (add_reg_br_prob_note): New function.
21650 (combine_predictions_for_insn): Update.
21651 * asan.c (asan_clear_shadow): Update.
21652 * cfgbuild.c (compute_outgoing_frequencies): Update.
21653 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21654 (update_br_prob_note): Update.
21655 (rtl_verify_edges): Update.
21656 (purge_dead_edges): Update.
21657 (fixup_reorder_chain): Update.
21658 * emit-rtl.c (try_split): Update.
21659 * ifcvt.c (cond_exec_process_insns): Update.
21660 (cond_exec_process_if_block): Update.
21661 (dead_or_predicable): Update.
21662 * internal-fn.c (expand_addsub_overflow): Update.
21663 (expand_neg_overflow): Update.
21664 (expand_mul_overflow): Update.
21665 * loop-doloop.c (doloop_modify): Update.
21666 * loop-unroll.c (compare_and_jump_seq): Update.
21667 * optabs.c (emit_cmp_and_jump_insn_1): Update.
21668 * predict.h: Update.
21669 * reorg.c (mostly_true_jump): Update.
21671 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
21672 * config/alpha/alpha.c (emit_unlikely_jump): Update.
21673 * config/arc/arc.c: (emit_unlikely_jump): Update.
21674 * config/arm/arm.c: (emit_unlikely_jump): Update.
21675 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
21676 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
21677 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
21678 (ix86_print_operand): Update.
21679 (ix86_split_fp_branch): Update.
21680 (predict_jump): Update.
21681 * config/ia64/ia64.c (ia64_print_operand): Update.
21682 * config/mmix/mmix.c (mmix_print_operand): Update.
21683 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
21684 (rs6000_expand_split_stack_prologue): Update.
21685 * config/rs6000/rs6000.c: Update.
21686 * config/s390/s390.c (s390_expand_vec_strlen): Update.
21687 (s390_expand_vec_movstr): Update.
21688 (s390_expand_cs_tdsi): Update.
21689 (s390_expand_split_stack_prologue): Update.
21690 * config/sh/sh.c (sh_print_operand): Update.
21691 (expand_cbranchsi4): Update.
21692 (expand_cbranchdi4): Update.
21693 * config/sparc/sparc.c (output_v9branch): Update.
21694 * config/spu/spu.c (get_branch_target): Update.
21695 (ea_load_store_inline): Update.
21696 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
21697 * config/tilepro/tilepro.c: Update.
21699 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
21701 * gimplify.c (mostly_copy_tree_r): Revert latest change.
21702 (gimplify_save_expr): Likewise.
21704 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21706 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
21708 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21710 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
21712 * timevar.def (TV_IPA_FNSUMMARY): Define.
21714 2017-07-16 Daniel Cederman <cederman@gaisler.com>
21716 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
21717 to back store errata sensitive sequence from being generated.
21718 (sqrtdf2_fix): Likewise.
21720 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
21722 * tree-ssa-threadupdate.c (compute_path_counts,
21723 update_joiner_offpath_counts): Use profile_probability.
21725 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
21728 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21730 * config/arm/arm-c.c (arm_cpu_builtins): Define
21731 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21733 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
21735 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21736 array entries to represent __ieee128 versions of the
21737 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
21738 scalar_extract_sig, and scalar_insert_exp built-in functions.
21739 (altivec_resolve_overloaded_builtin): Add special case handling
21740 for the __builtin_scalar_insert_exp function, as represented by
21741 the P9V_BUILTIN_VEC_VSIEDP constant.
21742 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
21743 exponent support for __ieee128 argument.
21744 (VSESQP): Add scalar extract signature support for __ieee128
21746 (VSTDCNQP): Add scalar test negative support for __ieee128
21748 (VSIEQP): Add scalar insert exponent support for __int128 argument
21749 with __ieee128 result.
21750 (VSIEQPF): Add scalar insert exponent support for __ieee128
21751 argument with __ieee128 result.
21752 (VSTDCQP): Add scalar test data class support for __ieee128
21754 (VSTDCNQP): Add overload support for scalar test negative with
21755 __ieee128 argument.
21756 (VSTDCQP): Add overload support for scalar test data class
21757 __ieee128 argument.
21758 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
21759 UNSPEC_VSX_SXSIGDP.
21760 (UNSPEC_VSX_SIEXPQP): New constant.
21761 (xsxexpqp): New insn for VSX scalar extract exponent quad
21763 (xsxsigqp): New insn for VSX scalar extract significand quad
21765 (xsiexpqpf): New insn for VSX scalar insert exponent quad
21766 precision with floating point argument.
21767 (xststdcqp): New expand for VSX scalar test data class quad
21769 (xststdcnegqp): New expand for VSX scalar test negative quad
21771 (xststdcqp): New insn to match expansions for VSX scalar test data
21772 class quad precision and VSX scalar test negative quad precision.
21773 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
21774 special case operand checking to enforce that second operand of
21775 VSX scalar test data class with quad precision argument is a 7-bit
21777 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
21778 prototypes and descriptions of __ieee128 versions of
21779 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
21780 scalar_test_data_class, and scalar_test_neg built-in functions.
21782 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21784 PR tree-optimization/81162
21785 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
21786 replace a negate with an add.
21788 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
21790 * doc/invoke.texi (arm/-mcpu): Document +crypto.
21792 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21794 * config/arm/arm-c.c (arm_cpu_builtins): Define
21795 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
21797 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21799 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
21800 (armv8-r): Set ARM Cortex-R52 as default CPU.
21801 * config/arm/arm-tables.opt: Regenerate.
21802 * config/arm/arm-tune.md: Regenerate.
21803 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
21805 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
21806 extension for -mcpu=cortex-r52.
21808 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
21810 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
21811 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
21812 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
21813 (fp-armv8): Define it as FP_ARMv8 only.
21814 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
21815 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
21817 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
21818 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
21819 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
21820 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
21821 than TARGET_FPU_ARMV8.
21822 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
21823 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
21824 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
21826 * config/arm/neon.md (neon_vrint): Likewise.
21827 (neon_vcvt): Likewise.
21828 (neon_<fmaxmin_op><mode>): Likewise.
21829 (<fmaxmin><mode>3): Likewise.
21830 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
21831 * config/arm/predicates.md (arm_cond_move_operator): Check against
21832 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
21834 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
21836 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
21837 to top of function.
21839 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21841 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
21842 loop in comment with memset.
21844 2017-07-14 Martin Liska <mliska@suse.cz>
21846 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
21847 * dwarf2out.c (is_java): Remove the function.
21848 (output_pubname): Remove usage of the function.
21849 (lower_bound_default): Remove usage of DW_LANG_Java.
21850 (gen_compile_unit_die): Likewise.
21851 * gcc.c: Remove compiler defaults for .java and .zip files.
21852 * gimple-expr.c (remove_suffix): Change as there's no longer
21853 extension than 4-letter one.
21854 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
21855 (gimplify_save_expr): Likewise.
21856 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
21857 as it's possible even for other languages than Java.
21858 * langhooks.h (struct lang_hooks): Remove Java from a comment.
21859 * lto-opts.c (lto_write_options): Remove reference to Java.
21860 * opts.c (strip_off_ending): Update file extension handling.
21861 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
21862 * tree-eh.c (lower_resx): Likewise.
21863 * tree.c (free_lang_data_in_type): Remove dead code.
21864 (find_decls_types_r): Likewise.
21865 (build_common_builtin_nodes): Remove Java from a comment.
21866 (verify_type): Remove dead code.
21867 * varasm.c (assemble_external): Remove Java from a comment.
21869 2017-07-14 Martin Liska <mliska@suse.cz>
21871 * opts.c (finish_options): Add quotes.
21872 (common_handle_option): Likewise.
21874 2017-07-14 Martin Liska <mliska@suse.cz>
21876 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
21877 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
21878 Remove N_SO_PASCAL.
21879 * dwarf2out.c (lower_bound_default): Do not handle
21881 (gen_compile_unit_die): Likewise.
21882 * gcc.c: Remove default extension binding for GNU Pascal.
21883 * stmt.c: Remove Pascal language from a comment.
21884 * xcoffout.c: Likewise.
21886 2017-07-13 David Malcolm <dmalcolm@redhat.com>
21889 * diagnostic-show-locus.c (fixit_cmp): New function.
21890 (layout::layout): Sort m_fixit_hints.
21891 (column_range::column_range): Assert that the values are valid.
21892 (struct char_span): New struct.
21893 (correction::overwrite): New method.
21894 (struct source_line): New struct.
21895 (line_corrections::add_hint): Add assertions. Reimplement memcpy
21896 calls in terms of classes source_line and char_span, and
21897 correction::overwrite.
21898 (selftest::test_overlapped_fixit_printing_2): New function.
21899 (selftest::diagnostic_show_locus_c_tests): Call it.
21901 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
21903 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
21904 early if there is no lhs.
21906 2017-07-13 Martin Liska <mliska@suse.cz>
21908 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
21909 (gen_reference_type_die): Likewise.
21910 * stor-layout.c: Remove Pascal-related comment.
21912 2017-07-13 Martin Liska <mliska@suse.cz>
21914 * opts.c (finish_options): Add quotes to error messages.
21915 (parse_sanitizer_options): Likewise.
21917 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
21919 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
21921 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
21923 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
21925 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
21927 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
21929 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
21931 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
21934 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
21935 provides the hardware capability bits, define the macro
21936 __BUILTIN_CPU_SUPPORTS__.
21937 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
21938 if GLIBC does not provide the hardware capability bits. Add a
21939 gcc_unreachable call if the built-in cpu function is neither
21940 __builtin_cpu_is nor __builtin_cpu_supports.
21941 (rs6000_get_function_versions_dispatcher): Change the warning
21942 that an old GLIBC is used which does not export the capability
21943 bits to be an error.
21944 * doc/extend.texi (target_clones attribute): Document the
21945 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
21946 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
21947 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
21948 the macros defined by GCC if the newer GLIBC is available.
21950 2017-07-12 Jeff Law <law@redhat.com>
21952 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
21953 remaining includes slightly.
21954 * config/riscv/riscv-builtins.c: Include profile-count.h.
21956 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21959 * config/avr/avr.c (avr_set_current_function): In diagnostic
21960 messages: Quote keywords and (parts of) identifiers.
21961 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
21964 2017-07-12 Carl Love <cel@us.ibm.com>
21966 * config/rs6000/rs6000-c.c: Add support for built-in functions
21967 vector bool char vec_revb (vector bool char);
21968 vector bool short vec_revb (vector short char);
21969 vector bool int vec_revb (vector bool int);
21970 vector bool long long vec_revb (vector bool long long);
21971 * doc/extend.texi: Update the built-in documentation file for the
21972 new built-in functions.
21974 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21976 * config/s390/s390.md: Remove movcc splitter.
21978 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21980 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
21981 load/store on condition.
21983 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
21986 * config/avr/avr.c (avr_encode_section_info)
21987 [progmem && !TREE_READONLY]: Error if progmem object needs
21990 2017-07-11 Michael Collison <michael.collison@arm.com>
21992 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
21995 2017-07-11 Carl Love <cel@us.ibm.com>
21997 * config/rs6000/rs6000-c.c: Add support for builtins
21998 vector unsigned int vec_parity_lsbb (vector signed int);
21999 vector unsigned int vec_parity_lsbb (vector unsigned int);
22000 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
22001 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
22002 vector unsigned long long vec_parity_lsbb (vector signed long long);
22003 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
22004 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
22005 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
22006 * doc/extend.texi: Update the built-in documentation file for the
22007 new built-in functions.
22009 2017-07-11 David Malcolm <dmalcolm@redhat.com>
22011 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
22012 (layout::m_primary_loc): New field.
22013 (layout::layout): Initialize new field. Move location filtering
22014 logic from here to...
22015 (layout::maybe_add_location_range): ...this new method. Add
22016 support for filtering to just the lines already specified by other
22018 (layout::will_show_line_p): New method.
22019 (gcc_rich_location::add_location_if_nearby): New method.
22020 (selftest::test_add_location_if_nearby): New test function.
22021 (selftest::diagnostic_show_locus_c_tests): Call it.
22022 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
22025 2017-07-11 Tom de Vries <tom@codesourcery.com>
22027 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
22028 (bb_first_real_insn): New function.
22029 (nvptx_single): Add extra initialization of broadcasted condition
22032 2017-07-11 Nathan Sidwell <nathan@acm.org>
22034 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
22036 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
22038 * doc/extend.texi (AVR Function Attributes): Remove weblink to
22039 Binutils doc as TEXI will mess them up.
22040 * doc/invoke.texi (AVR Options): Same here.
22042 2017-07-11 Daniel Cederman <cederman@gaisler.com>
22044 * config/sparc/sparc.opt (mfix-ut700): New option.
22045 (mfix-gr712rc): Likewise.
22046 (sparc_fix_b2bst): New variable.
22047 * doc/invoke.texi (SPARC options): Document them.
22048 (ARM options): Fix warnings.
22049 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
22050 instructions to prevent sequences that can trigger the store-store
22051 errata for certain LEON3FT processors.
22052 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
22053 (sparc_option_override): Set sparc_fix_b2bst appropriately.
22054 * config/sparc/sparc.md (fix_b2bst): New attribute.
22055 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
22057 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
22060 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
22062 (*rsqrtsf2_sse): Ditto.
22064 (div<mode>3): Macroize insn from divdf3 and divsf3
22065 using MODEF mode iterator.
22067 2017-07-10 Martin Sebor <msebor@redhat.com>
22069 PR tree-optimization/80397
22070 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
22071 instead of testing for equality to INTEGER_TYPE.
22073 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
22075 * config.gcc: Remove uclibc from arc target spec.
22077 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
22079 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
22081 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
22084 * lto-wrapper.c (remove_option): New function.
22085 (merge_and_complain): Merge PIC/PIE options more realistically.
22087 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22089 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
22093 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
22094 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
22095 * config.in: Regenerate.
22096 * configure: Regenerate.
22097 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
22098 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
22099 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
22100 (TARGET_GASISR_PROLOGUES): ...target mask.
22101 * common/config/avr/avr-common.c
22102 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
22103 Set -mgas-isr-prologues.
22104 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
22105 INSERT_PASS_BEFORE for it.
22106 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
22107 * config/avr/avr.c (avr_option_override)
22108 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
22109 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
22110 (avr_attribute_table) <no_gccisr>: Add new function attribute.
22111 (avr_set_current_function) <is_no_gccisr>: Init machine field.
22112 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
22114 (make_avr_pass_pre_proep): New function.
22115 (emit_push_sfr) <treg>: Add argument to function and use it
22116 instead of TMP_REG.
22117 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
22118 and set machine->gasisr.yes.
22119 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
22120 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
22121 __gcc_isr.n_pushed to .L__stack_usage.
22122 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
22123 (avr_asm_final_postscan_insn): ...this new static function.
22124 * config/avr/avr.h (machine_function)
22125 <is_no_gccisr, use_L__stack_usage>: New fields.
22126 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
22127 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
22128 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
22129 (gasisr, *gasisr): New expander and insn.
22130 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
22131 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
22132 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
22134 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
22136 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
22139 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
22141 Move jump-tables out of .text again.
22144 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
22145 (ASM_OUTPUT_ADDR_VEC): New function.
22146 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
22147 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
22148 INSN_ADDRESSes as asm comment.
22149 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
22150 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
22151 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
22152 * config/avr/avr.md (*tablejump): Adjust comment.
22153 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
22154 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
22156 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
22157 (avr_output_addr_vec): New proto.
22158 (avr_log_t) <insn_addresses>: New field.
22160 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
22163 * config/i386/i386.c (ix86_function_arg_advance): Set
22164 outgoing_args_on_stack to true if there are outgoing arguments
22166 (ix86_function_arg): Likewise.
22167 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
22168 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
22169 * config/i386/i386.h (machine_function): Add
22170 outgoing_args_on_stack.
22172 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
22174 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
22175 supporting pthreds.
22176 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
22178 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
22180 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
22181 (REAL_H): Remove $(MACHMODE_H).
22182 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
22184 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
22185 $(MACHMODE_H) and double-int.h.
22186 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
22188 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
22191 2017-07-07 Andrew Pinski <apinski@cavium.com>
22193 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
22194 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
22196 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
22198 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
22199 Add warning if GCC was not configured to link against a GLIBC that
22200 exports the hardware capability bits.
22201 (make_resolver_func): Make resolver function private and not a
22202 COMDAT function. Create the name with clone_function_name instead
22203 of make_unique_name.
22206 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
22207 correct operand in doing the split.
22209 2017-07-07 Carl Love <cel@us.ibm.com>
22211 * config/rs6000/rs6000-c: Add support for built-in function
22212 vector unsigned short vec_pack_to_short_fp32 (vector float,
22214 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
22215 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
22216 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
22217 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
22218 (convert_4f32_8i16): Add define_expand.
22219 * doc/extend.texi: Update the built-in documentation file for the
22220 new built-in function.
22222 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22224 * config/sparc/m8.md: New file.
22225 * config/sparc/sparc.md: Include m8.md.
22227 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22229 * config/sparc/sparc.opt: New option -mvis4b.
22230 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
22231 (sparc_option_override): Handle VIS4B.
22232 (enum sparc_builtins): Define
22233 SPARC_BUILTIN_DICTUNPACK{8,16,32},
22234 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
22235 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
22236 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
22237 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
22238 (check_constant_argument): New function.
22239 (sparc_vis_init_builtins): Define builtins
22240 __builtin_vis_dictunpack{8,16,32},
22241 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
22242 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
22243 __builtin_vis_fpcmpde{8,16,32}shl and
22244 __builtin_vis_fpcmpur{8,16,32}shl.
22245 (sparc_expand_builtin): Check that the constant operands to
22246 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
22247 constant and in range.
22248 * config/sparc/sparc-c.c (sparc_target_macros): Handle
22250 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
22251 (SPARC_IMM5_P): Likewise.
22252 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
22253 (enabled): Handle vis4b.
22254 (UNSPEC_DICTUNPACK): New unspec.
22255 (UNSPEC_FPCMPSHL): Likewise.
22256 (UNSPEC_FPUCMPSHL): Likewise.
22257 (UNSPEC_FPCMPDESHL): Likewise.
22258 (UNSPEC_FPCMPURSHL): Likewise.
22259 (cpu_feature): New CPU feature `vis4b'.
22260 (dictunpack{8,16,32}): New insns.
22261 (FPCSMODE): New mode iterator.
22262 (fpcscond): New code iterator.
22263 (fpcsucond): Likewise.
22264 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
22265 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
22266 (fpcmpde{8,16,32}{si,di}shl): Likewise.
22267 (fpcmpur{8,16,32}{si,di}shl): Likewise.
22268 * config/sparc/constraints.md: Define constraints `q' for unsigned
22269 2-bit integer constants and `t' for unsigned 5-bit integer
22271 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
22273 (imm5_operand_dictunpack16): Likewise.
22274 (imm5_operand_dictunpack32): Likewise.
22275 (imm2_operand): Likewise.
22276 * doc/invoke.texi (SPARC Options): Document -mvis4b.
22277 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
22278 ditunpack* and fpcmp*shl builtins.
22280 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22282 * config.gcc: Handle m8 in --with-{cpu,tune} options.
22283 * config.in: Add HAVE_AS_SPARC6 define.
22284 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
22286 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
22288 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
22289 (CPP_CPU_SPEC): Handle m8.
22290 (ASM_CPU_SPEC): Likewise.
22291 * config/sparc/sparc-opts.h (enum processor_type): Add
22293 * config/sparc/sparc.c (m8_costs): New struct.
22294 (sparc_option_override): Handle TARGET_CPU_m8.
22295 (sparc32_initialize_trampoline): Likewise.
22296 (sparc64_initialize_trampoline): Likewise.
22297 (sparc_issue_rate): Likewise.
22298 (sparc_register_move_cost): Likewise.
22299 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
22300 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
22301 (ASM_CPU64_DEFAULT_SPEC): Likewise.
22302 (CPP_CPU_SPEC): Handle M8.
22303 (ASM_CPU_SPEC): Likewise.
22304 (AS_M8_FLAG): Define.
22305 * config/sparc/sparc.md: Add m8 to the cpu attribute.
22306 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
22307 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
22309 * configure: Regenerate.
22310 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
22313 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22315 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
22317 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
22318 ("*movdi_insn_sp32"): Do not set v3pipe.
22319 ("*movsi_insn"): Likewise.
22320 ("*movdi_insn_sp64"): Likewise.
22321 ("*movsf_insn"): Likewise.
22322 ("*movdf_insn_sp32"): Likewise.
22323 ("*movdf_insn_sp64"): Likewise.
22324 ("*zero_extendsidi2_insn_sp64"): Likewise.
22325 ("*sign_extendsidi2_insn"): Likewise.
22326 ("*mov<VM32:mode>_insn"): Likewise.
22327 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22328 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22329 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22330 ("<vlop:code><VL:mode>3"): Likewise.
22331 ("*not_<vlop:code><VL:mode>3"): Likewise.
22332 ("*nand<VL:mode>_vis"): Likewise.
22333 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
22334 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
22335 ("one_cmpl<VL:mode>2"): Likewise.
22336 ("faligndata<VM64:mode>_vis"): Likewise.
22337 ("alignaddrsi_vis"): Likewise.
22338 ("alignaddrdi_vis"): Likweise.
22339 ("alignaddrlsi_vis"): Likewise.
22340 ("alignaddrldi_vis"): Likewise.
22341 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22342 ("bmaskdi_vis"): Likewise.
22343 ("bmasksi_vis"): Likewise.
22344 ("bshuffle<VM64:mode>_vis"): Likewise.
22345 ("cmask8<P:mode>_vis"): Likewise.
22346 ("cmask16<P:mode>_vis"): Likewise.
22347 ("cmask32<P:mode>_vis"): Likewise.
22348 ("pdistn<P:mode>_vis"): Likewise.
22349 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22351 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22353 * config/sparc/sparc.md ("subtype"): New insn attribute.
22354 ("*wrgsr_sp64"): Set insn subtype.
22355 ("*rdgsr_sp64"): Likewise.
22356 ("alignaddrsi_vis"): Likewise.
22357 ("alignaddrdi_vis"): Likewise.
22358 ("alignaddrlsi_vis"): Likewise.
22359 ("alignaddrldi_vis"): Likewise.
22360 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
22361 ("fexpand_vis"): Likewise.
22362 ("fpmerge_vis"): Likewise.
22363 ("faligndata<VM64:mode>_vis"): Likewise.
22364 ("bshuffle<VM64:mode>_vis"): Likewise.
22365 ("cmask8<P:mode>_vis"): Likewise.
22366 ("cmask16<P:mode>_vis"): Likewise.
22367 ("cmask32<P:mode>_vis"): Likewise.
22368 ("fchksm16_vis"): Likewise.
22369 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
22370 ("fmean16_vis"): Likewise.
22371 ("fp<plusminus_insn>64_vis"): Likewise.
22372 ("<plusminus_insn>v8qi3"): Likewise.
22373 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
22374 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
22375 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
22376 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
22377 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
22378 ("*movqi_insn"): Likewise.
22379 ("*movhi_insn"): Likewise.
22380 ("*movsi_insn"): Likewise.
22381 ("movsi_pic_gotdata_op"): Likewise.
22382 ("*movdi_insn_sp32"): Likewise.
22383 ("*movdi_insn_sp64"): Likewise.
22384 ("movdi_pic_gotdata_op"): Likewise.
22385 ("*movsf_insn"): Likewise.
22386 ("*movdf_insn_sp32"): Likewise.
22387 ("*movdf_insn_sp64"): Likewise.
22388 ("*zero_extendhisi2_insn"): Likewise.
22389 ("*zero_extendqihi2_insn"): Likewise.
22390 ("*zero_extendqisi2_insn"): Likewise.
22391 ("*zero_extendqidi2_insn"): Likewise.
22392 ("*zero_extendhidi2_insn"): Likewise.
22393 ("*zero_extendsidi2_insn_sp64"): Likewise.
22394 ("ldfsr"): Likewise.
22395 ("prefetch_64"): Likewise.
22396 ("prefetch_32"): Likewise.
22397 ("tie_ld32"): Likewise.
22398 ("tie_ld64"): Likewise.
22399 ("*tldo_ldub_sp32"): Likewise.
22400 ("*tldo_ldub1_sp32"): Likewise.
22401 ("*tldo_ldub2_sp32"): Likewise.
22402 ("*tldo_ldub_sp64"): Likewise.
22403 ("*tldo_ldub1_sp64"): Likewise.
22404 ("*tldo_ldub2_sp64"): Likewise.
22405 ("*tldo_ldub3_sp64"): Likewise.
22406 ("*tldo_lduh_sp32"): Likewise.
22407 ("*tldo_lduh1_sp32"): Likewise.
22408 ("*tldo_lduh_sp64"): Likewise.
22409 ("*tldo_lduh1_sp64"): Likewise.
22410 ("*tldo_lduh2_sp64"): Likewise.
22411 ("*tldo_lduw_sp32"): Likewise.
22412 ("*tldo_lduw_sp64"): Likewise.
22413 ("*tldo_lduw1_sp64"): Likewise.
22414 ("*tldo_ldx_sp64"): Likewise.
22415 ("*mov<VM32:mode>_insn"): Likewise.
22416 ("*mov<VM64:mode>_insn_sp64"): Likewise.
22417 ("*mov<VM64:mode>_insn_sp32"): Likewise.
22419 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22421 * config/sparc/sparc.md ("type"): New insn type viscmp.
22422 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
22424 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
22425 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
22426 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
22427 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
22429 ("n7_vis_logical_11cycle"): Likewise.
22430 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
22431 * config/sparc/niagara2.md ("niag3_vis": Likewise.
22432 * config/sparc/niagara.md ("niag_vis"): Likewise.
22433 * config/sparc/ultra3.md ("us3_fga"): Likewise.
22434 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
22436 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
22438 * config/sparc/sparc.md: New instruction type `bmask'.
22439 (bmaskdi_vis): Use the `bmask' type.
22440 (bmasksi_vis): Likewise.
22441 * config/sparc/ultra3.md (us3_array): Likewise.
22442 * config/sparc/niagara7.md (n7_array): Likewise.
22443 * config/sparc/niagara4.md (n4_array): Likewise.
22444 * config/sparc/niagara2.md (niag2_vis): Likewise.
22445 (niag3_vis): Likewise.
22446 * config/sparc/niagara.md (niag_vis): Likewise.
22448 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22450 * ipa-comdats.c: Remove optimize check from gate.
22451 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
22452 for functions not optimized.
22453 (ipa_fn_summary_read): Skip optimize check.
22454 (ipa_fn_summary_write): Likewise.
22455 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
22457 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
22459 (can_inline_edge_p): Check flag_pcc_struct_return for match.
22460 (check_callers): Give up on caller which is not optimized.
22461 (inline_small_functions): Likewise.
22462 (ipa_inline): Do not give up when not optimizing.
22463 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
22464 away unoptimizes cdtors.
22465 (whole_program_function_and_variable_visibility): Do
22466 ipa_discover_readonly_nonaddressable_vars in LTO mode.
22467 * ipa.c (process_references): Do not check optimize.
22468 (symbol_table::remove_unreachable_nodes): Update optimize check.
22469 (set_writeonly_bit): Update optimize check.
22470 (pass_ipa_cdtor_merge::gate): Do not check optimize.
22471 (pass_ipa_single_use::gate): Remove.
22473 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22475 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
22476 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
22477 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
22478 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
22479 permute_load, permute_store, adjust_extract, adjust_splat,
22480 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
22481 replace_swap_with_copy, dump_swap_insn_table,
22482 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
22483 recombine_lvx_pattern, recombine_stvx_pattern,
22484 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
22485 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
22486 to file rs6000-p8swap.c.
22487 * config/rs6000/rs6000-p8swap.c: New file.
22488 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
22489 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
22490 and rs6000*-*-* targets.
22492 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22494 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
22496 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22498 * lto-wrapper.c (merge_and_complain): Do not merge
22499 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
22500 fsigned_zeros, ftrapping_math, fwrapv.
22501 (append_compiler_options): Do not track these options.
22502 (append_linker_options): Likewie
22504 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22506 * cgraphunit.c (cgraph_node::finalize_function): When
22507 !flag_toplevel_reorde set no_reorder flag.
22508 (varpool_node::finalize_decl): Likewise.
22509 (symbol_table::compile): Drop no toplevel reorder path.
22511 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22513 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
22514 edges; zero probability is not better than uninitialized.
22516 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22518 * asan.h (asan_sanitize_allocas_p): Declare.
22519 * asan.c (asan_sanitize_allocas_p): New function.
22520 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
22521 (handle_builtin_alloca): Likewise.
22522 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
22523 if !asan_sanitize_allocas_p.
22524 * params.def (asan-instrument-allocas): Add new option.
22525 * params.h (ASAN_PROTECT_ALLOCAS): Define.
22526 * opts.c (common_handle_option): Disable allocas sanitization for
22529 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
22531 * asan.c: Include gimple-fold.h.
22532 (get_last_alloca_addr): New function.
22533 (handle_builtin_stackrestore): Likewise.
22534 (handle_builtin_alloca): Likewise.
22535 (asan_emit_allocas_unpoison): Likewise.
22536 (get_mem_refs_of_builtin_call): Add new parameter, remove const
22537 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
22538 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
22539 (instrument_builtin_call): Pass gimple iterator to
22540 get_mem_refs_of_builtin_call.
22541 (last_alloca_addr): New global.
22542 * asan.h (asan_emit_allocas_unpoison): Declare.
22543 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
22544 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
22545 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
22546 if function calls alloca.
22547 * gimple-fold.c (replace_call_with_value): Remove static keyword.
22548 * gimple-fold.h (replace_call_with_value): Declare.
22549 * internal-fn.c: Include asan.h.
22550 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
22551 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
22553 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22555 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
22556 (C_SELFTEST_FLAGS): New.
22557 (CPP_SELFTEST_FLAGS): New.
22558 (SELFTEST_DEPS): New, from deps of s-selftest.
22559 (C_SELFTEST_DEPS): New, from deps of s-selftest.
22560 (CPP_SELFTEST_DEPS): New.
22561 (selftest): Add dependency on s-selftest-c++.
22562 (s-selftest): Rename to...
22563 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
22564 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
22565 than SELFTEST_FLAGS.
22566 (selftest-gdb): Rename to...
22567 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
22569 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
22570 (selftest-valgrind): Rename to...
22571 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
22573 (selftest-valgrind): Reintroduce as an alias for
22574 selftest-c-valgrind.
22575 (s-selftest-c++): New.
22576 (selftest-c++-gdb): New.
22577 (selftest-c++-valgrind): New.
22579 2017-07-06 Olivier Hainque <hainque@adacore.com>
22581 * gcc.c (process_command): When deciding if undefined variables
22582 should be ignored when processing specs, accept "gcc -v" as well.
22584 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
22586 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
22587 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
22589 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22591 * config/arm/arm-cpus.in (armv8-r): Add new entry.
22592 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
22593 * config/arm/arm-tables.opt: Regenerate.
22594 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
22596 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
22598 2017-07-06 Carl Love <cel@us.ibm.com>
22600 * ChangeLog: Clean up from mid air collision
22602 2017-07-06 Carl Love <cel@us.ibm.com>
22604 * config/rs6000/rs6000-c.c: Add support for built-in functions
22605 vector signed int vec_subc (vector signed int, vector signed int);
22606 vector signed __int128 vec_subc (vector signed __int128,
22607 vector signed __int128);
22608 vector unsigned __int128 vec_subc (vector unsigned __int128,
22609 vector unsigned __int128);
22610 vector signed int vec_sube (vector signed int, vector signed int,
22611 vector signed int);
22612 vector unsigned int vec_sube (vector unsigned int,
22613 vector unsigned int,
22614 vector unsigned int);
22615 vector signed __int128 vec_sube (vector signed __int128,
22616 vector signed __int128,
22617 vector signed__int128);
22618 vector unsigned __int128 vec_sube (vector unsigned __int128,
22619 vector unsigned __int128,
22620 vector unsigned __int128);
22621 vector signed int vec_subec (vector signed int, vector signed int,
22622 vector signed int);
22623 vector unsigned int vec_subec (vector unsigned int,
22624 vector unsigned int,
22625 vector unsigned int);
22626 vector signed __int128 vec_subec (vector signed __int128,
22627 vector signed __int128,
22628 vector signed__int128);
22629 vector unsigned __int128 vec_subec (vector unsigned __int128,
22630 vector unsigned __int128,
22631 vector unsigned __int128);
22632 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
22633 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
22634 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
22635 BU_ALTIVEC_OVERLOAD_X definitions.
22636 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
22637 * doc/extend.texi: Update the built-in documentation file for the new
22638 built-in functions.
22640 2017-07-06 David Malcolm <dmalcolm@redhat.com>
22643 * diagnostic-show-locus.c (layout::layout): Use start and finish
22644 spelling location for the start and finish of each range.
22645 * genmatch.c (linemap_client_expand_location_to_spelling_point):
22646 Add unused aspect param.
22647 * input.c (expand_location_1): Add "aspect" param, and use it
22648 to access the correct part of the location.
22649 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
22651 (expand_location_to_spelling_point): Likewise.
22652 (linemap_client_expand_location_to_spelling_point): Add "aspect"
22653 param, and pass it to expand_location_1.
22655 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
22657 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
22658 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
22659 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
22660 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
22661 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
22662 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
22663 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
22664 _mm_maskz_getmant_ss): New intrinsics.
22665 (__builtin_ia32_getexpss128_mask): Changed to ...
22666 __builtin_ia32_getexpss128_round ... this.
22667 (__builtin_ia32_getexpsd128_mask): Changed to ...
22668 __builtin_ia32_getexpsd128_round ... this.
22669 * config/i386/i386-builtin-types.def
22670 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
22671 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
22672 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
22673 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
22674 __builtin_ia32_getmantss_mask_round): New builtins.
22675 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
22676 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
22677 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
22678 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
22679 * config/i386/sse.md
22680 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
22681 avx512f_sgetexp<mode><mask_scalar_name>
22682 <round_saeonly_scalar_name> ... this.
22683 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
22684 %0, %1, %2<round_saeonly_op3>}): Changed to ...
22685 vgetexp<ssescalarmodesuffix>
22686 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22687 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
22688 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
22689 avx512f_vgetmant<mode><mask_scalar_name>
22690 <round_saeonly_scalar_name> ... this.
22691 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
22692 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
22693 vgetmant<ssescalarmodesuffix>
22694 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
22695 %0<mask_scalar_operand4>, %1, %2
22696 <round_saeonly_scalar_mask_op4>, %3} ... this.
22697 * config/i386/subst.md (mask_scalar_operand4,
22698 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
22699 round_saeonly_scalar_nimm_predicate): New subst attributes.
22701 2017-07-06 Julia Koval <julia.koval@intel.com>
22703 * config/i386/i386.c (ix86_erase_embedded_rounding):
22704 Remove code for old rounding pattern.
22706 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
22708 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
22710 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
22712 * doc/sourcebuild.texi (Test Directives, Variants of
22713 dg-require-support): Add documentation for dg-require-stack-check.
22715 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
22717 * config/i386/subst.md (mask_scalar, round_scalar,
22718 round_saeonly_scalar): New meta-templates.
22719 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
22720 round_scalar_mask_operand3, round_scalar_mask_op3,
22721 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
22722 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
22723 round_saeonly_scalar_constraint,
22724 round_saeonly_scalar_prefix): New subst attribute.
22725 * config/i386/sse.md
22726 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
22727 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
22728 <round_scalar_name> ... this.
22729 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
22730 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
22731 <round_scalar_name> ... this.
22732 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
22733 <sse>_vm<code><mode>3<mask_scalar_name>
22734 <round_saeonly_scalar_name> ... this.
22735 (v<plusminus_mnemonic><ssescalarmodesuffix>
22736 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22737 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22738 v<plusminus_mnemonic><ssescalarmodesuffix>
22739 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22740 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22741 (v<multdiv_mnemonic><ssescalarmodesuffix>
22742 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
22743 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
22744 v<multdiv_mnemonic><ssescalarmodesuffix>
22745 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22746 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
22747 (v<maxmin_float><ssescalarmodesuffix>
22748 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
22749 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
22750 v<maxmin_float><ssescalarmodesuffix>
22751 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
22752 %0<mask_scalar_operand3>, %1, %<iptr>2
22753 <round_saeonly_scalar_mask_op3>} ... this.
22755 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
22757 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
22758 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
22760 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
22761 Alan Hayward <alan.hayward@arm.com>
22762 David Sherwood <david.sherwood@arm.com>
22764 * combine.c (simplify_if_then_else): Remove "enum" before
22766 * compare-elim.c (can_eliminate_compare): Likewise.
22767 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
22769 (aarch64_lookup_simd_builtin_type): Likewise.
22770 (aarch64_simd_builtin_type): Likewise.
22771 (aarch64_init_simd_builtin_types): Likewise.
22772 (aarch64_simd_expand_args): Likewise.
22773 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
22775 (aarch64_reverse_mask): Likewise.
22776 (aarch64_simd_emit_reg_reg_move): Likewise.
22777 (aarch64_gen_adjusted_ldpstp): Likewise.
22778 (aarch64_ccmp_mode_to_code): Likewise.
22779 (aarch64_operands_ok_for_ldpstp): Likewise.
22780 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22781 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
22783 (aarch64_min_divisions_for_recip_mul): Likewise.
22784 (aarch64_reassociation_width): Likewise.
22785 (aarch64_get_condition_code_1): Likewise.
22786 (aarch64_simd_emit_reg_reg_move): Likewise.
22787 (aarch64_simd_attr_length_rglist): Likewise.
22788 (aarch64_reverse_mask): Likewise.
22789 (aarch64_operands_ok_for_ldpstp): Likewise.
22790 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
22791 (aarch64_gen_adjusted_ldpstp): Likewise.
22792 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
22794 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
22795 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
22796 (arm_lookup_simd_builtin_type): Likewise.
22797 (arm_simd_builtin_type): Likewise.
22798 (arm_init_simd_builtin_types): Likewise.
22799 (arm_expand_builtin_args): Likewise.
22800 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
22801 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
22802 (ft32_setup_incoming_varargs): Likewise.
22803 (ft32_function_arg): Likewise.
22804 (ft32_function_arg_advance): Likewise.
22805 (ft32_pass_by_reference): Likewise.
22806 (ft32_arg_partial_bytes): Likewise.
22807 (ft32_valid_pointer_mode): Likewise.
22808 (ft32_addr_space_pointer_mode): Likewise.
22809 (ft32_addr_space_legitimate_address_p): Likewise.
22810 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
22812 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
22813 (ix86_emit_outlined_ms2sysv_restore): Likewise.
22814 (iamcu_alignment): Likewise.
22815 (canonicalize_vector_int_perm): Likewise.
22816 (ix86_noce_conversion_profitable_p): Likewise.
22817 (ix86_mpx_bound_mode): Likewise.
22818 (ix86_operands_ok_for_move_multiple): Likewise.
22819 * config/microblaze/microblaze-protos.h
22820 (microblaze_expand_conditional_branch_reg): Likewise.
22821 * config/microblaze/microblaze.c
22822 (microblaze_expand_conditional_branch_reg): Likewise.
22823 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
22825 (rs6000_reassociation_width): Likewise.
22826 (rs6000_invalid_binary_op): Likewise.
22827 (fusion_p9_p): Likewise.
22828 (emit_fusion_p9_load): Likewise.
22829 (emit_fusion_p9_store): Likewise.
22830 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
22832 (riscv_hard_regno_mode_ok_p): Likewise.
22833 (riscv_address_insns): Likewise.
22834 (riscv_split_symbol): Likewise.
22835 (riscv_legitimize_move): Likewise.
22836 (riscv_function_value): Likewise.
22837 (riscv_hard_regno_nregs): Likewise.
22838 (riscv_expand_builtin): Likewise.
22839 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
22840 (riscv_build_integer): Likewise.
22841 (riscv_split_integer): Likewise.
22842 (riscv_legitimate_constant_p): Likewise.
22843 (riscv_cannot_force_const_mem): Likewise.
22844 (riscv_regno_mode_ok_for_base_p): Likewise.
22845 (riscv_valid_base_register_p): Likewise.
22846 (riscv_valid_offset_p): Likewise.
22847 (riscv_valid_lo_sum_p): Likewise.
22848 (riscv_classify_address): Likewise.
22849 (riscv_legitimate_address_p): Likewise.
22850 (riscv_address_insns): Likewise.
22851 (riscv_load_store_insns): Likewise.
22852 (riscv_force_binary): Likewise.
22853 (riscv_split_symbol): Likewise.
22854 (riscv_force_address): Likewise.
22855 (riscv_legitimize_address): Likewise.
22856 (riscv_move_integer): Likewise.
22857 (riscv_legitimize_const_move): Likewise.
22858 (riscv_legitimize_move): Likewise.
22859 (riscv_address_cost): Likewise.
22860 (riscv_subword): Likewise.
22861 (riscv_output_move): Likewise.
22862 (riscv_canonicalize_int_order_test): Likewise.
22863 (riscv_emit_int_order_test): Likewise.
22864 (riscv_function_arg_boundary): Likewise.
22865 (riscv_pass_mode_in_fpr_p): Likewise.
22866 (riscv_pass_fpr_single): Likewise.
22867 (riscv_pass_fpr_pair): Likewise.
22868 (riscv_get_arg_info): Likewise.
22869 (riscv_function_arg): Likewise.
22870 (riscv_function_arg_advance): Likewise.
22871 (riscv_arg_partial_bytes): Likewise.
22872 (riscv_function_value): Likewise.
22873 (riscv_pass_by_reference): Likewise.
22874 (riscv_setup_incoming_varargs): Likewise.
22875 (riscv_print_operand): Likewise.
22876 (riscv_elf_select_rtx_section): Likewise.
22877 (riscv_save_restore_reg): Likewise.
22878 (riscv_for_each_saved_reg): Likewise.
22879 (riscv_register_move_cost): Likewise.
22880 (riscv_hard_regno_mode_ok_p): Likewise.
22881 (riscv_hard_regno_nregs): Likewise.
22882 (riscv_class_max_nregs): Likewise.
22883 (riscv_memory_move_cost): Likewise.
22884 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
22885 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
22886 (rl78_addr_space_address_mode): Likewise.
22887 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22889 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
22890 (rs6000_reassociation_width): Likewise.
22891 (rs6000_invalid_binary_op): Likewise.
22892 (fusion_p9_p): Likewise.
22893 (emit_fusion_p9_load): Likewise.
22894 (emit_fusion_p9_store): Likewise.
22895 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
22896 (ok_for_simple_move_operands): Likewise.
22897 (ok_for_simple_move_strict_operands): Likewise.
22898 (ok_for_simple_arith_logic_operands): Likewise.
22899 (visium_legitimize_reload_address): Likewise.
22900 (visium_select_cc_mode): Likewise.
22901 (output_cbranch): Likewise.
22902 (visium_split_double_move): Likewise.
22903 (visium_expand_copysign): Likewise.
22904 (visium_expand_int_cstore): Likewise.
22905 (visium_expand_fp_cstore): Likewise.
22906 * config/visium/visium.c (visium_pass_by_reference): Likewise.
22907 (visium_function_arg): Likewise.
22908 (visium_function_arg_advance): Likewise.
22909 (visium_libcall_value): Likewise.
22910 (visium_setup_incoming_varargs): Likewise.
22911 (visium_legitimate_constant_p): Likewise.
22912 (visium_legitimate_address_p): Likewise.
22913 (visium_legitimize_address): Likewise.
22914 (visium_secondary_reload): Likewise.
22915 (visium_register_move_cost): Likewise.
22916 (visium_memory_move_cost): Likewise.
22917 (prepare_move_operands): Likewise.
22918 (ok_for_simple_move_operands): Likewise.
22919 (ok_for_simple_move_strict_operands): Likewise.
22920 (ok_for_simple_arith_logic_operands): Likewise.
22921 (visium_function_value_1): Likewise.
22922 (rtx_ok_for_offset_p): Likewise.
22923 (visium_legitimize_reload_address): Likewise.
22924 (visium_split_double_move): Likewise.
22925 (visium_expand_copysign): Likewise.
22926 (visium_expand_int_cstore): Likewise.
22927 (visium_expand_fp_cstore): Likewise.
22928 (visium_split_cstore): Likewise.
22929 (visium_select_cc_mode): Likewise.
22930 (visium_split_cbranch): Likewise.
22931 (output_cbranch): Likewise.
22932 (visium_print_operand_address): Likewise.
22933 * expmed.c (flip_storage_order): Likewise.
22934 * expmed.h (emit_cstore): Likewise.
22935 (flip_storage_order): Likewise.
22936 * genrecog.c (validate_pattern): Likewise.
22937 * hsa-gen.c (gen_hsa_addr): Likewise.
22938 * internal-fn.c (expand_arith_overflow): Likewise.
22939 * ira-color.c (allocno_copy_cost_saving): Likewise.
22940 * lra-assigns.c (find_hard_regno_for_1): Likewise.
22941 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
22942 (process_invariant_for_inheritance): Likewise.
22943 * lra-eliminations.c (move_plus_up): Likewise.
22944 * omp-low.c (lower_oacc_reductions): Likewise.
22945 * simplify-rtx.c (simplify_subreg): Likewise.
22946 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
22947 (TARGET_CHKP_BOUND_MODE): Likewise..
22948 * targhooks.c (default_chkp_bound_mode): Likewise.
22949 (default_setup_incoming_vararg_bounds): Likewise.
22950 * targhooks.h (default_chkp_bound_mode): Likewise.
22951 (default_setup_incoming_vararg_bounds): Likewise.
22952 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
22953 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
22954 (have_whole_vector_shift): Likewise.
22955 * tree-vect-stmts.c (vectorizable_load): Likewise.
22956 * doc/tm.texi: Regenerate.
22958 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22960 Graceful degrade if Binutils PR21472 is not available.
22963 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
22964 .rodata in flash test fails.
22965 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
22966 * confgure: Regenerate.
22967 * config.in: Regenerate.
22968 * config/avr/avr.c (avr_asm_named_section)
22969 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
22970 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
22971 (avr_asm_init_sections): Same.
22973 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22975 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
22976 (fma<VH:mode>4_intrinsic): Likewise.
22977 (*fmsub<VCVTF:mode>4): Likewise.
22978 (*fmsub<VH:mode>4_intrinsic): Likewise.
22980 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
22983 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
22984 Don't depend on "optimize > 0".
22985 (out_movhi_r_mr, out_movqi_mr_r): Same.
22986 (out_movhi_mr_r, out_movqi_r_mr): Same.
22987 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
22988 io_address_operand on "optimize > 0".
22990 2017-07-05 Bin Cheng <bin.cheng@arm.com>
22992 * tree-loop-distribution.c: Add general explanantion on the pass.
22993 (generate_loops_for_partition): Mark distributed loop.
22994 (pg_add_dependence_edges): New parameter. Handle alias data
22995 dependence specially and record it in the parameter if asked.
22996 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
22997 (init_partition_graph_vertices, add_partition_graph_edge): New.
22998 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
22999 (free_partition_graph_vdata, build_partition_graph): New.
23000 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
23001 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
23002 (data_ref_segment_size, latch_dominated_by_data_ref): New.
23003 (compute_alias_check_pairs, version_loop_by_alias_check): New.
23004 (version_for_distribution_p, finalize_partitions): New.
23005 (distribute_loop): Handle alias data dependence specially. Factor
23006 out loop fusion code as functions and call these functions.
23008 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23010 * tree-loop-distribution.c (classify_partition): New parameter and
23011 better handle reduction statement.
23012 (rdg_build_partitions): Revise comment.
23013 (distribute_loop): Compute statements in all partitions and pass it
23014 to classify_partition.
23016 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23018 * tree-loop-distribution.c (enum partition_type): New.
23019 (struct partition): New field type.
23020 (partition_merge_into): Add parameter. Update partition type.
23021 (data_dep_in_cycle_p, update_type_for_merge): New functions.
23022 (build_rdg_partition_for_vertex): Compute partition type.
23023 (rdg_build_partitions): Dump partition type.
23024 (distribute_loop): Update calls to partition_merge_into.
23026 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23028 * tree-loop-distribution.c (struct ddr_hasher): New.
23029 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
23031 (classify_partition): Call get_data_dependence.
23032 (pg_add_dependence_edges): Ditto.
23033 (distribute_loop): Release data dependence hash table.
23035 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23037 * tree-loop-distribution.c (ref_base_address): Delete.
23038 (similar_memory_accesses): Rename ...
23039 (share_memory_accesses): ... to this. Check if partitions access
23040 the same memory reference.
23041 (distribute_loop): Call share_memory_accesses.
23043 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23045 * tree-loop-distribution.c (struct partition): New field recording
23046 its data reference.
23047 (partition_alloc, partition_free): Init and release data refs.
23048 (partition_merge_into): Merge data refs.
23049 (build_rdg_partition_for_vertex): Collect data refs for partition.
23050 (pg_add_dependence_edges): Change parameters from vector to bitmap.
23052 (distribute_loop): Remve data refs from vertice data of partition
23055 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23057 * tree-loop-distribution.c (params.h): Include header file.
23058 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
23059 (datarefs_vec): New global var.
23060 (create_rdg_vertices): Use datarefs_vec directly.
23061 (free_rdg): Don't free data references.
23062 (build_rdg): Update use. Don't free data references.
23063 (distribute_loop): Compute global variable for data references.
23064 Bail out if there are too many data references.
23066 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23068 * tree-loop-distribution.c (loop_nest): New global var.
23069 (build_rdg): Use loop directly, rather than loop nest.
23070 (pg_add_dependence_edges): Remove loop nest parameter. Use global
23072 (distribute_loop): Compute global variable loop nest. Update use.
23074 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23076 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
23077 (partition_merge_into): New parameter. Dump reason for fusion.
23078 (distribute_loop): Update use of partition_merge_into.
23080 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23082 * tree-loop-distribution.c (bb_top_order_index): New.
23083 (bb_top_order_index_size, bb_top_order_cmp): New.
23084 (stmts_from_loop): Use topological order.
23085 (pass_loop_distribution::execute): Compute and release topological
23086 order for basic blocks.
23088 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23090 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
23093 2017-07-05 Bin Cheng <bin.cheng@arm.com>
23095 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
23096 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
23097 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
23098 * internal-fn.def (LOOP_DIST_ALIAS): New.
23099 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
23100 (fold_loop_internal_call): ... this.
23101 (vect_loop_dist_alias_call): New function.
23102 (set_uid_loop_bbs): Call fold_loop_internal_call.
23103 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
23106 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23109 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
23110 Require dead FLAGS_REG at the beginning of a peephole.
23112 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
23115 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
23116 arguments in the call to __builtin_ia32_sbb_u32.
23117 (_subborrow_u64): Swap _X and _Y arguments in the call to
23118 __builtin_ia32_sbb_u64.
23120 2017-07-04 Jakub Jelinek <jakub@redhat.com>
23123 * tree-vrp.c (compare_assert_loc): Turn into a function template
23124 with stable template parameter. Only test if a->e is NULL,
23125 !a->e == !b->e has been verified already. Use e == NULL or
23126 e != NULL instead of e or ! e tests. If stable is true, don't use
23127 iterative_hash_expr, on the other side allow a or b or both NULL
23128 and sort the NULLs last.
23129 (process_assert_insertions): Sort using compare_assert_loc<false>
23130 instead of compare_assert_loc, later sort using
23131 compare_assert_loc<true> before calling process_assert_insertions_for
23132 in a loop. Use break instead of continue once seen NULL pointer.
23134 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23136 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
23137 Cortex-R7 and Cortex-R8 processors.
23139 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23141 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
23142 uninitialized while src is not.
23144 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
23146 * common/config/arm/arm-common.c: Adjust include path for
23148 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
23149 (arm-cpu.h): Create in build directory. Adjust dependency rules.
23150 (arm-cpu-data.h): Likewise.
23151 (arm-cpu-cdata.h): Likewise.
23152 * config/arm/arm-cpu.h: Delete.
23153 * config/arm/arm-cpu-cdata.h: Delete.
23154 * config/arm/arm-cpu-data.h: Delete.
23156 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
23158 * config/arm/arm-cpus.in (cortex-a55): New.
23159 (cortex-a75): Likewise.
23160 (cortex-a75.cortex-a55): Likewise.
23161 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
23163 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
23164 * config/arm/arm-cpu-cdata.h: Regenerate.
23165 * config/arm/arm-cpu-data.h: Regenerate.
23166 * config/arm/arm-cpu.h: Regenerate.
23167 * config/arm/arm-tables.opt: Regenerate.
23168 * config/arm/arm-tune.md: Regenerate.
23170 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23172 * haifa-sched.c (sched_create_recovery_edges): Update profile.
23174 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
23176 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
23179 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
23181 PR tree-optimization/81292
23182 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
23183 full_string_p, also call adjust_related_strinfos if the adjustment
23184 is simple, otherwise invalidate related strinfos.
23186 2017-07-04 Martin Liska <mliska@suse.cz>
23189 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
23190 newly created variable as DECL_IGNORED_P.
23192 2017-07-04 Martin Liska <mliska@suse.cz>
23195 * ipa-inline.c (inline_small_functions):
23196 Use xstrdup_for_dump.
23198 2017-07-04 Tom de Vries <tom@codesourcery.com>
23200 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
23202 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23205 * config/darwin.c (darwin_function_switched_text_sections):
23208 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
23210 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
23212 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23214 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
23216 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23218 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
23219 min_profitable_iters, and th as inclusive lower bounds.
23220 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
23221 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
23222 for min_profitable_iters and min_profitable_estimate.
23223 (vect_transform_loop): Treat th as an inclusive lower bound.
23224 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
23226 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
23229 * config/darwin.c (darwin_function_switched_text_sections):
23230 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
23231 in two pieces, and suppress the use of buf.
23233 2017-07-03 Nathan Sidwell <nathan@acm.org>
23235 * hash-table.h (hash_table_mod1): Fix indentation.
23237 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23239 PR middle-end/81290
23240 * predict.c (force_edge_cold): Be more careful about propagation
23242 * profile-count.h (profile_probability::guessed,
23243 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
23245 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
23247 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
23249 * doc/invoke.texi (rcpc architecture extension): Document it.
23251 2017-07-03 Richard Biener <rguenther@suse.de>
23253 PR tree-optimization/60510
23254 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
23255 the scalar reduction PHI and use it.
23256 (vectorizable_reduction): Properly guard the single_defuse_cycle
23257 path for non-SLP reduction chains where we cannot use it.
23258 Rework reduc_def/index and vector type deduction. Rework
23259 vector operand gathering during reduction op code-gen.
23260 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
23261 chains dissolve the chain and leave it to non-SLP reduction
23264 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23266 * tree-data-ref.h (dr_alignment): Declare.
23267 * tree-data-ref.c (dr_alignment): New function.
23268 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
23269 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23271 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
23273 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23275 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
23276 and base_misalignment fields.
23277 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
23278 * tree-data-ref.c: Include builtins.h.
23279 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
23280 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
23281 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
23282 * tree-vect-data-refs.c: Include tree-cfg.h.
23283 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
23284 fields instead of calculating an alignment here.
23285 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
23286 innermost_loop_behavior fields.
23288 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23290 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
23292 (DR_STEP_ALIGNMENT): New macro.
23293 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
23294 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
23295 (create_data_ref): Print it.
23296 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
23297 to tell whether the step preserves vector (mis)alignment.
23298 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23299 Move the check for an integer step and generalise to all INTEGER_CST.
23300 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
23301 Print the outer step alignment.
23303 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23305 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
23306 with offset_alignment.
23307 (DR_ALIGNED_TO): Delete.
23308 (DR_OFFSET_ALIGNMENT): New macro.
23309 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
23310 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
23311 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
23312 (create_data_ref): Likewise.
23313 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
23314 (vect_analyze_data_refs): Likewise.
23315 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
23316 creating dummy innermost behavior.
23318 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23320 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
23321 with a "innermost_loop_behavior *" and refeence tree.
23322 * tree-data-ref.c (dr_analyze_innermost): Likewise.
23323 (create_data_ref): Update call accordingly.
23324 * tree-predcom.c (find_looparound_phi): Likewise.
23326 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23328 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
23329 fields with dr_wrt_vec_loop.
23330 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
23331 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
23332 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
23333 (vect_dr_behavior): New function.
23334 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23335 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
23336 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
23337 track whether the step preserves the misalignment.
23338 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
23339 Use vect_dr_behavior.
23340 (vect_setup_realignment): Update call accordingly.
23341 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
23342 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
23343 call to vect_create_addr_base_for_vector_ref.
23344 (vect_create_cond_for_align_checks): Likewise.
23345 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
23346 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
23347 (vect_recog_mask_conversion_pattern): Likewise.
23348 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
23349 (new_stmt_vec_info): Remove redundant zeroing.
23351 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
23353 * common/config/arm/arm-common.c (arm_be8_option): New function.
23354 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
23355 (ISA_ARMv6): Add isa_bit_be8.
23356 * config/arm/arm.h (arm_be8_option): Add prototype.
23357 (BE8_SPEC_FUNCTION): New define.
23358 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
23359 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
23360 (mlittle-endian): Similarly.
23361 (mbe8, mbe32): New options.
23362 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
23363 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
23365 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23367 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
23369 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23371 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
23372 (cleanup_tree_cfg_bb): Use it.
23373 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
23375 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
23377 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23380 * loop-doloop.c (add_test): Update profile.
23382 2017-07-03 Martin Liska <mliska@suse.cz>
23385 * sanopt.c (rewrite_usage_of_param): New function.
23386 (sanitize_rewrite_addressable_params): Likewise.
23387 (pass_sanopt::execute): Call rewrite_usage_of_param.
23389 2017-07-03 Richard Biener <rguenther@suse.de>
23391 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
23392 back to using VIEW_CONVERT_EXPR.
23394 2017-07-03 Martin Liska <mliska@suse.cz>
23397 * doc/extend.texi: Document when a resolver function is
23398 generated for target_clones.
23400 2017-07-03 Martin Liska <mliska@suse.cz>
23402 * asan.c (asan_emit_stack_protection): Unpoison just red zones
23403 and shadow memory of auto variables which are subject of
23404 use-after-scope sanitization.
23405 (asan_expand_mark_ifn): Add do set only when is_poison.
23407 2016-07-03 Richard Biener <rguenther@suse.de>
23409 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
23411 (vect_force_simple_reduction): Record reduction def -> phi mapping.
23412 (vectorizable_reduction): Perform reduction PHI creation when
23413 visiting a reduction PHI and adjust and simplify code generation
23414 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
23415 (vect_transform_loop): Visit reduction PHIs.
23416 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
23417 defs into the SLP tree.
23418 (vect_build_slp_tree): Reduction defs terminate the recursion.
23419 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
23421 (vect_get_vec_defs_for_stmt_copy): Export.
23422 (vect_get_vec_defs): Likewise.
23423 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
23425 (vect_get_vec_defs_for_stmt_copy): Declare.
23426 (vect_get_vec_defs): Likewise.
23428 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
23430 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
23431 parameter with a "loop" parameter and use it instead of the
23432 loop containing DR_STMT. Don't check simple_iv when doing
23433 BB analysis. Describe the two analysis modes in the comment.
23435 2017-07-03 Tom de Vries <tom@codesourcery.com>
23437 PR tree-optimization/69468
23438 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
23439 (find_same_succ_bb): Handle ignore_edge_flags.
23441 2017-07-03 Tom de Vries <tom@codesourcery.com>
23443 PR tree-optimization/81192
23444 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
23446 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
23448 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
23450 2017-07-03 Tom de Vries <tom@codesourcery.com>
23452 PR tree-optimization/81192
23453 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
23454 BB_SAME_SUCC (bb) == NULL.
23456 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23458 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
23461 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23463 * dumpfile.c: Include profile-count.h
23464 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
23466 (insert_cond_bb): Update profile.
23467 * tree-cfg.h (insert_cond_bb): Update prototype.
23468 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
23469 * tree-dump.c: Do not include tree-cfg.
23471 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23473 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
23475 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23477 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
23480 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
23482 * tree-complex.c (expand_complex_div_wide): update profile.
23484 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23485 Alan Hayward <alan.hayward@arm.com>
23486 David Sherwood <david.sherwood@arm.com>
23488 * Makefile.in (MACHMODE_H): Remove insn-modes.h
23489 (CORETYPES_H): New define.
23490 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
23491 (insn-modes-inline.h, s-modes-inline-h): New rules.
23492 (generated_files): Add insn-modes-inline.h.
23493 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
23494 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
23495 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
23496 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
23497 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
23498 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
23499 (build/gencodes.o, build/genconditions.o): Likewise.
23500 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
23501 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
23502 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
23503 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
23504 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
23505 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
23506 * coretypes.h: Include everything up to real.h for generators.
23507 Include insn-modes.h first. Include wide-int-print.h after
23508 wide-int.h. Include insn-modes-inline.h and then machmode.h.
23509 * machmode.h: Don't include insn-modes.h here.
23510 * function-tests.c: Remove includes of signop.h, machmode.h,
23511 double-int.h and wide-int.h.
23513 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
23515 * optc-save-gen.awk: Likewise.
23516 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
23517 * godump.c: Remove include of wide-int-print.h.
23518 * pretty-print.h: Likewise.
23519 * wide-int-print.cc: Likewise.
23520 * wide-int.cc: Likewise.
23521 * hash-map-tests.c: Remove include of signop.h.
23522 * hash-set-tests.c: Likewise.
23523 * rtl-tests.c: Likewise.
23524 * mkconfig.sh: Remove include of machmode.h.
23525 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
23527 (emit_insn_modes_inline_h): ...this new function. Emit the code
23528 into an insn-modes-inline.h header file, adding appropriate
23529 include guards and end comments.
23530 (emit_insn_modes_c_header): Remove include of machmode.h.
23531 (emit_min_insn_modes_c_header): Include coretypes.h rather than
23533 (main): Handle -i flag and call emit_insn_modes_inline_h when
23536 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23538 * tree-ssa-strlen.c (strinfo): Rename the length field to
23539 nonzero_chars. Add a full_string_p field.
23540 (compare_nonzero_chars, zero_length_string_p): New functions.
23541 (get_addr_stridx): Add an offset_out parameter.
23542 Use compare_nonzero_chars.
23543 (get_stridx): Update accordingly. Use compare_nonzero_chars.
23544 (new_strinfo): Update after above changes to strinfo.
23545 (set_endptr_and_length): Set full_string_p.
23546 (get_string_length): Update after above changes to strinfo.
23547 (unshare_strinfo): Update call to new_strinfo.
23548 (maybe_invalidate): Likewise.
23549 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
23550 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
23551 as a uhwi instead of an shwi. Update after above changes to
23552 strinfo and new_strinfo.
23553 (zero_length_string): Assert that chainsi contains full strings.
23554 Use zero_length_string_p. Update call to new_strinfo.
23555 (adjust_related_strinfos): Update after above changes to strinfo.
23556 Copy full_string_p from origsi.
23557 (adjust_last_stmt): Use zero_length_string_p.
23558 (handle_builtin_strlen): Update after above changes to strinfo and
23559 new_strinfo. Install the lhs as the string length if the previous
23560 entry didn't describe a full string.
23561 (handle_builtin_strchr): Update after above changes to strinfo
23563 (handle_builtin_strcpy): Likewise.
23564 (handle_builtin_strcat): Likewise.
23565 (handle_builtin_malloc): Likewise.
23566 (handle_pointer_plus): Likewise.
23567 (handle_builtin_memcpy): Likewise. Track nonzero characters
23568 that aren't necessarily followed by a nul terminator.
23569 (handle_char_store): Likewise.
23571 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23573 PR tree-optimization/80769
23574 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
23575 for malloc and calloc. Document the new invariant that all related
23576 strinfos have delayed lengths or none do.
23577 (verify_related_strinfos): Move earlier in file.
23578 (set_endptr_and_length): New function, split out from...
23579 (get_string_length): ...here. Also set the lengths of related
23581 (zero_length_string): Assert that chainsi has known (rather than
23583 (adjust_related_strinfos): Likewise.
23585 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
23587 PR tree-optimization/81136
23588 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
23589 assert that two references with the same misalignment have the same
23590 compile-time misalignment if those compile-time misalignments
23593 2017-07-01 Andi Kleen <ak@linux.intel.com>
23595 * print-tree.c (print_node): Print all attributes.
23597 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23599 * cfg.c (scale_bbs_frequencies): New function.
23600 * cfg.h (scale_bbs_frequencies): Declare it.
23601 * cfgloopanal.c (single_likely_exit): Cleanup.
23602 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
23604 (scale_loop_profile): Likewise.
23605 (loop_version): Likewise.
23606 (create_empty_loop_on_edge): Update.
23607 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
23608 scale_loop_frequencies, scale_loop_profile, loopify,
23609 loop_version): Update prototypes.
23610 * modulo-sched.c (sms_schedule): Update.
23611 * predict.c (unlikely_executed_edge_p): Also check probability.
23612 (probably_never_executed_edge_p): Fix typo.
23613 * tree-if-conv.c (version_loop_for_if_conversion): Update.
23614 * tree-parloops.c (gen_parallel_loop): Update.
23615 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
23616 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
23617 * tree-ssa-loop-split.c (split_loop): Update.
23618 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
23619 * tree-vect-loop-manip.c (vect_do_peeling): Update.
23620 (vect_loop_versioning): Update.
23621 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
23623 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23625 * trans-mem.c (split_bb_make_tm_edge): Update profile.
23627 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23629 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
23630 to keep profile consistent.
23632 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23634 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
23635 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
23636 * profile-count.h (max_safe_multiplier): Make unsigned.
23637 (profile_count::guessed_zero): New.
23639 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23641 * bb-reorder.c (fix_up_crossing_landing_pad,
23642 fix_crossing_conditional_branches): Use make_single_succ_edge
23643 to keep profile consistent.
23645 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
23647 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
23650 2017-07-01 Jakub Jelinek <jakub@redhat.com>
23653 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
23654 the right scopes, make sure cond_jump isn't preserved between multiple
23655 iterations. Search for fallthru edge whenever there are 3+ edges and
23656 use find_fallthru_edge for it.
23658 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23660 Patch by Alexander Monakov <amonakov@ispras.ru>
23661 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
23662 probabilities consistently.
23664 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23666 * pa.c (pa_expand_compare_and_swap_loop): Update call of
23667 emit_cmp_and_jump_insns.
23669 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23672 * tree-inline.c (expand_call_inline): Combine profile statuses.
23674 2017-06-30 Andrew Pinski <apinski@cavium.com>
23676 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
23677 fold_stmt returned true.
23679 2017-06-30 Nathan Sidwell <nathan@acm.org>
23681 * ggc.h (empty_string): Delete.
23682 * cfgexpand.c (expand_asm_stmt): Use plain "".
23683 * optabs.c (expand_asm_memory_barrier): Likewise.
23684 * stringpool.c (empty_string): Delete.
23685 (digit_vector, digit_string): Delete.
23686 (ggc_alloc_string): Use plain "", don't optimize single digit
23687 strings. Use ggc_alloc_atomic.
23689 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
23691 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
23692 comparison set and one other set, use the cost of the non-comparison
23695 2017-06-30 Nathan Sidwell <nathan@acm.org>
23697 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
23700 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
23702 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
23703 loops. Remove now unneeded calls to gimple_switch_set_label() that
23704 just set removed labels to NULL_TREE.
23706 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
23708 * tree-ssanames.c (set_range_info_raw): Abstract from ...
23709 (set_range_info): ...here. Only call set_range_info_raw if domain
23711 (set_nonzero_bits): Call set_range_info_raw.
23712 * tree-ssanames.h (set_range_info_raw): New.
23714 2017-06-30 Jakub Jelinek <jakub@redhat.com>
23717 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
23718 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
23719 of nonimmediate_operand and <store_mask_constraint> instead of m
23720 for the input operand. For V8FI iterator, always split if input
23721 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
23722 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
23723 <store_mask_predicate> instead of register_operand and
23724 <store_mask_constraint> instead of v for the input operand. Make
23725 sure both operands aren't MEMs for if not <mask_applied>.
23727 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
23729 * lto-wrapper.c (copy_file) Close both file descriptors before
23732 2017-06-30 Martin Liska <mliska@suse.cz>
23735 * multiple_target.c (create_dispatcher_calls): Make ifunc
23736 also for function that don't have calls or are not referenced.
23738 2017-06-30 Richard Biener <rguenther@suse.de>
23740 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
23741 analyze the first scalar stmt. Move vector type computation
23742 for the BB case here from ...
23743 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
23744 live operation processing in the SLP case properly.
23746 2017-06-30 Richard Biener <rguenther@suse.de>
23748 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
23750 2017-06-30 Martin Liska <mliska@suse.cz>
23753 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
23754 before BUILT_IN_UNWIND_RESUME when ASAN is used.
23756 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
23758 * doc/invoke.texi (AArch64): Add missing options and remove redundant
23761 2017-06-30 Richard Biener <rguenther@suse.de>
23763 PR tree-optimization/81249
23764 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
23765 condition reduction result to original scalar type.
23767 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23769 * profile-count.h (enum profile_quality): Fix typos and whitespace
23772 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23774 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
23775 type for branch probabilities.
23777 2017-06-29 Julian Brown <julian@codesourcery.com>
23778 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23780 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
23781 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
23782 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
23783 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
23785 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23787 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
23788 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
23789 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
23790 CC usage from generic code to here.
23791 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
23792 CC usage into the target macros.
23794 2017-06-29 Maya Rashish <coypu@sdf.org>
23796 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
23799 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23801 * arm/arm-builtins.c: Include profile-count.h
23802 * except.c (sjlj_emit_function_enter): Use
23803 profile_probability::unlikely.
23805 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23807 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
23808 and tocrel_offset be pointer args rather than implicitly using
23810 (legitimate_constant_pool_address_p, rs6000_emit_move,
23811 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
23812 tocrel_offset and use in toc_relative_expr_p call.
23813 (print_operand, print_operand_address): Use static tocrel_base_oac
23814 and tocrel_offset_oac.
23815 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
23818 2017-06-29 Maya Rashish <coypu@sdf.org>
23820 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
23822 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
23824 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
23825 objects, take into account only the alignment of 'op0' and 'mode1' if
23828 2017-06-29 Steve Ellcey <sellcey@cavium.com>
23830 * ccmp.c (ccmp_tree_comparison_p): New function.
23831 (ccmp_candidate_p): Update to use above function.
23832 (get_compare_parts): New function.
23833 (expand_ccmp_next): Update to use new functions.
23834 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
23836 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
23837 take mode as argument.
23838 * ccmp.h (expand_ccmp_expr): Add mode as argument.
23839 * expr.c (expand_expr_real_1): Pass mode as argument.
23841 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
23843 * combine.c (combine_instructions): Print insns to dump_file, together
23846 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
23848 * asan.c (asan_emit_stack_protection): Update.
23849 (create_cond_insert_point): Update.
23850 * auto-profile.c (afdo_propagate_circuit): Update.
23851 * basic-block.h (struct edge_def): Turn probability to
23852 profile_probability.
23853 (EDGE_FREQUENCY): Update.
23854 * bb-reorder.c (find_traces_1_round): Update.
23855 (better_edge_p): Update.
23856 (sanitize_hot_paths): Update.
23857 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
23858 (make_single_succ_edge): Update.
23859 (check_bb_profile): Update.
23860 (dump_edge_info): Update.
23861 (update_bb_profile_for_threading): Update.
23862 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
23863 probabilitycount to 0.
23864 * cfgbuild.c (compute_outgoing_frequencies): Update.
23865 * cfgcleanup.c (try_forward_edges): Update.
23866 (outgoing_edges_match): Update.
23867 (try_crossjump_to_edge): Update.
23868 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
23869 (expand_gimple_tailcall): Update.
23870 (construct_init_block): Use make_single_succ_edge.
23871 (construct_exit_block): Use make_single_succ_edge.
23872 * cfghooks.c (verify_flow_info): Update.
23873 (redirect_edge_succ_nodup): Update.
23874 (split_edge): Update.
23875 (account_profile_record): Update.
23876 * cfgloopanal.c (single_likely_exit): Update.
23877 * cfgloopmanip.c (scale_loop_profile): Update.
23878 (set_zero_probability): Remove.
23879 (duplicate_loop_to_header_edge): Update.
23880 * cfgloopmanip.h (loop_version): Update prototype.
23881 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
23882 (force_nonfallthru_and_redirect): Update.
23883 (update_br_prob_note): Update.
23884 (rtl_verify_edges): Update.
23885 (purge_dead_edges): Update.
23886 (rtl_lv_add_condition_to_bb): Update.
23887 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
23888 * cgraphunit.c (init_lowered_empty_function): Update.
23889 (cgraph_node::expand_thunk): Update.
23890 * cilk-common.c: Include profile-count.h
23891 * dojump.c (inv): Remove.
23892 (jumpifnot): Update.
23893 (jumpifnot_1): Update.
23894 (do_jump_1): Update.
23896 (do_jump_by_parts_greater_rtx): Update.
23897 (do_compare_rtx_and_jump): Update.
23898 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
23899 do_jump_1. do_compare_rtx_and_jump): Update prototype.
23900 * dwarf2cfi.c: Include profile-count.h
23901 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
23902 (sjlj_emit_dispatch_table): Likewise.
23903 * explow.c: Include profile-count.h
23904 * expmed.c (emit_store_flag_force): Update.
23905 (do_cmp_and_jump): Update.
23906 * expr.c (compare_by_pieces_d::generate): Update.
23907 (compare_by_pieces_d::finish_mode): Update.
23908 (emit_block_move_via_loop): Update.
23909 (store_expr_with_bounds): Update.
23910 (store_constructor): Update.
23911 (expand_expr_real_2): Update.
23912 (expand_expr_real_1): Update.
23913 * expr.h (try_casesi, try_tablejump): Update prototypes.
23914 * gimple-pretty-print.c (dump_probability): Update.
23915 (dump_profile): New.
23916 (dump_gimple_label): Update.
23917 (dump_gimple_bb_header): Update.
23918 * graph.c (draw_cfg_node_succ_edges): Update.
23919 * hsa-gen.c (convert_switch_statements): Update.
23920 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
23921 (find_if_case_1): Update.
23922 (find_if_case_2): Update.
23923 * internal-fn.c (expand_arith_overflow_result_store): Update.
23924 (expand_addsub_overflow): Update.
23925 (expand_neg_overflow): Update.
23926 (expand_mul_overflow): Update.
23927 (expand_vector_ubsan_overflow): Update.
23928 * ipa-cp.c (good_cloning_opportunity_p): Update.
23929 * ipa-split.c (split_function): Use make_single_succ_edge.
23930 * ipa-utils.c (ipa_merge_profiles): Update.
23931 * loop-doloop.c (add_test): Update.
23932 (doloop_modify): Update.
23933 * loop-unroll.c (compare_and_jump_seq): Update.
23934 (unroll_loop_runtime_iterations): Update.
23935 * lra-constraints.c (lra_inheritance): Update.
23936 * lto-streamer-in.c (input_cfg): Update.
23937 * lto-streamer-out.c (output_cfg): Update.
23938 * mcf.c (adjust_cfg_counts): Update.
23939 * modulo-sched.c (sms_schedule): Update.
23940 * omp-expand.c (expand_omp_for_init_counts): Update.
23941 (extract_omp_for_update_vars): Update.
23942 (expand_omp_ordered_sink): Update.
23943 (expand_omp_for_ordered_loops): Update.
23944 (expand_omp_for_generic): Update.
23945 (expand_omp_for_static_nochunk): Update.
23946 (expand_omp_for_static_chunk): Update.
23947 (expand_cilk_for): Update.
23948 (expand_omp_simd): Update.
23949 (expand_omp_taskloop_for_outer): Update.
23950 (expand_omp_taskloop_for_inner): Update.
23951 * omp-simd-clone.c (simd_clone_adjust): Update.
23952 * optabs.c (expand_doubleword_shift): Update.
23953 (expand_abs): Update.
23954 (emit_cmp_and_jump_insn_1): Update.
23955 (expand_compare_and_swap_loop): Update.
23956 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
23957 * predict.c (predictable_edge_p): Update.
23958 (edge_probability_reliable_p): Update.
23959 (set_even_probabilities): Update.
23960 (combine_predictions_for_insn): Update.
23961 (combine_predictions_for_bb): Update.
23962 (propagate_freq): Update.
23963 (estimate_bb_frequencies): Update.
23964 (force_edge_cold): Update.
23965 * profile-count.c (profile_count::dump): Add missing space into dump.
23966 (profile_count::debug): Add newline.
23967 (profile_count::differs_from_p): Explicitly convert to unsigned.
23968 (profile_count::stream_in): Update.
23969 (profile_probability::dump): New member function.
23970 (profile_probability::debug): New member function.
23971 (profile_probability::differs_from_p): New member function.
23972 (profile_probability::differs_lot_from_p): New member function.
23973 (profile_probability::stream_in): New member function.
23974 (profile_probability::stream_out): New member function.
23975 * profile-count.h (profile_count_quality): Rename to ...
23976 (profile_quality): ... this one.
23977 (profile_probability): New.
23978 (profile_count): Update.
23979 * profile.c (compute_branch_probabilities): Update.
23980 * recog.c (peep2_attempt): Update.
23981 * sched-ebb.c (schedule_ebbs): Update.
23982 * sched-rgn.c (find_single_block_region): Update.
23983 (compute_dom_prob_ps): Update.
23984 (schedule_region): Update.
23985 * sel-sched-ir.c (compute_succs_info): Update.
23986 * stmt.c (struct case_node): Update.
23987 (do_jump_if_equal): Update.
23988 (get_outgoing_edge_probs): Update.
23989 (conditional_probability): Update.
23990 (emit_case_dispatch_table): Update.
23991 (expand_case): Update.
23992 (expand_sjlj_dispatch_table): Update.
23993 (emit_case_nodes): Update.
23994 * targhooks.c: Update.
23995 * tracer.c (better_p): Update.
23996 (find_best_successor): Update.
23997 * trans-mem.c (expand_transaction): Update.
23998 * tree-call-cdce.c: Update.
23999 * tree-cfg.c (gimple_split_edge): Upate.
24000 (move_sese_region_to_fn): Upate.
24001 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
24002 * tree-eh.c (lower_resx): Upate.
24003 (cleanup_empty_eh_move_lp): Upate.
24004 * tree-if-conv.c (version_loop_for_if_conversion): Update.
24005 * tree-inline.c (copy_edges_for_bb): Update.
24006 (copy_cfg_body): Update.
24007 * tree-parloops.c (gen_parallel_loop): Update.
24008 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
24009 (gimple_gen_time_profiler): Update.
24010 * tree-ssa-dce.c (remove_dead_stmt): Update.
24011 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
24012 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
24013 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
24014 (unloop_loops): Update.
24015 (try_peel_loop): Update.
24016 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
24017 * tree-ssa-loop-split.c (connect_loops): Update.
24018 (split_loop): Update.
24019 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
24020 (hoist_guard): Update.
24021 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
24022 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
24023 (value_replacement): Update.
24024 * tree-ssa-reassoc.c (branch_fixup): Update.
24025 * tree-ssa-tail-merge.c (replace_block_by): Update.
24026 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
24027 (create_edge_and_update_destination_phis): Update.
24028 (compute_path_counts): Update.
24029 (recompute_probabilities): Update.
24030 (update_joiner_offpath_counts): Update.
24031 (freqs_to_counts_path): Update.
24032 (duplicate_thread_path): Update.
24033 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
24034 (struct switch_conv_info): Update.
24035 (gen_inbound_check): Update.
24036 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
24037 (vect_do_peeling): Update.
24038 (vect_loop_versioning): Update.
24039 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
24040 (optimize_mask_stores): Update.
24041 * ubsan.c (ubsan_expand_null_ifn): Update.
24042 * value-prof.c (gimple_divmod_fixed_value): Update.
24043 (gimple_divmod_fixed_value_transform): Update.
24044 (gimple_mod_pow2): Update.
24045 (gimple_mod_pow2_value_transform): Update.
24046 (gimple_mod_subtract): Update.
24047 (gimple_mod_subtract_transform): Update.
24048 (gimple_ic): Update.
24049 (gimple_stringop_fixed_value): Update.
24050 (gimple_stringops_transform): Update.
24051 * value-prof.h: Update.
24053 2017-06-29 Carl Love <cel@us.ibm.com>
24055 * config/rs6000/rs6000-c.c: Add support for built-in functions
24056 vector signed int vec_signed (vector float);
24057 vector signed long long vec_signed (vector double);
24058 vector signed int vec_signed2 (vector double, vector double);
24059 vector signed int vec_signede (vector double);
24060 vector signed int vec_signedo (vector double);
24061 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
24062 instruction generator.
24063 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24064 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
24065 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
24067 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
24068 vunsignede_v2df): Add define_expands.
24069 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
24070 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
24071 VEC_UNSIGNEDO): Add definitions.
24072 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
24073 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
24074 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
24075 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
24076 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
24077 * config/rs6000/altivec.h (vec_signed, vec_signed2,
24078 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
24079 vec_unsignede, vec_unsignedo): Add builtin defines.
24080 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
24082 * doc/extend.texi: Update the built-in documentation file for the
24083 new built-in functions.
24085 2017-06-29 Richard Biener <rguenther@suse.de>
24087 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
24088 reduction chains to LOOP_VINFO_REDUCTIONS.
24089 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
24090 SLP reductions after processing reduction chains.
24092 2017-06-29 Nathan Sidwell <nathan@acm.org>
24094 * builtins.c (fold_builtin_FUNCTION): Use
24095 lang_hooks.decl_printable_name.
24097 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
24099 PR middle-end/81194
24100 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
24101 with only one label.
24102 * stmt.c (expand_case): Assert NCASES is greater than one.
24104 2017-06-29 Richard Biener <rguenther@suse.de>
24106 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
24108 (group_case_labels): Likewise.
24109 (find_taken_edge): Push sanity checking on val to workers...
24110 (find_taken_edge_cond_expr): ... here
24111 (find_taken_edge_switch_expr): ... and here, handle cases
24112 with just a default label.
24113 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
24114 (group_case_labels): Likewise.
24115 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
24116 group_case_labels does anything cleanup the CFG again.
24118 2017-06-29 Bin Cheng <bin.cheng@arm.com>
24120 PR tree-optimization/81196
24121 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
24122 exit condition comparing two IVs.
24124 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
24126 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
24127 profile to the dummy entry at the end of the list of architectures.
24128 * config/arm/arm-cpu-cdata.h: Regenerated.
24130 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24131 Michael Collison <michael.collison@arm.com>
24134 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
24136 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
24137 (*aarch64_reg_<mode>3_minus_mask): New pattern.
24138 (*aarch64_<optab>_reg_di3_mask2): New pattern.
24139 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
24140 of shift when the shift amount is masked with constant equal to
24141 the size of the mode.
24142 * config/aarch64/predicates.md (subreg_lowpart_operator): New
24145 2017-06-29 Martin Liska <mliska@suse.cz>
24147 * config/i386/i386.opt: Change range from [1,5] to [0,5].
24149 2017-06-29 Yury Gribov <tetra2005@gmail.com>
24152 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
24155 (edge_growth_cache_entry::edge_growth_cache_entry): New
24157 (reset_edge_growth_cache): Update to use constructor.
24159 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24161 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
24162 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24163 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
24165 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
24167 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
24168 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
24170 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
24172 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
24173 (*-linux-uclibc*): Add t-uclibc tmake_file.
24174 * config/t-musl: New.
24175 * config/t-uclibc: New.
24177 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
24179 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
24181 (gen_comm_data): Emit architectural setting of arch_prof.
24182 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
24184 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
24185 (armv8-m.base, armv8-m.main): Likewise.
24186 * arm-protos.h (arm_build_target): Add profile field.
24187 (arch_option): Likewise.
24188 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
24190 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
24191 arm_active_target.profile.
24193 2017-06-28 Richard Biener <rguenther@suse.de>
24195 PR middle-end/81227
24196 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
24197 TYPE_OVERFLOW_WRAPS.
24198 * match.pd (negate_expr_p): Likewise.
24199 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
24200 fold_build2, not fold_binary.
24202 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24204 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
24205 Convert memory address to Pmode.
24206 (aarch64_print_operand): Assert MEM operands are always Pmode.
24208 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
24211 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
24212 Remove redundant if.
24213 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
24214 * config/arm/aarch-common-protos.h
24215 (aarch_forward_to_shift_is_not_shifted_re): Remove.
24216 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
24218 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
24221 * multiple_target.c (create_dispatcher_calls): Set the default
24222 clone to be static, not public.
24224 2017-06-28 Richard Biener <rguenther@suse.de>
24226 * tree-vect-loop.c (vectorizable_reduction): Move special
24227 cond reduction IV var creation ...
24228 (vect_create_epilog_for_reduction): ... here. Remove induction_index
24229 parameter. Use STMT_VINFO_VECTYPE.
24230 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
24233 2017-06-28 Martin Liska <mliska@suse.cz>
24236 * ipa-visibility.c (non_local_p): Handle visibility.
24238 2017-06-28 Martin Liska <mliska@suse.cz>
24241 * common.opt: Add IntegerRange to various options.
24242 * opt-functions.awk (integer_range_info): New function.
24243 * optc-gen.awk: Add integer_range_info to cl_options struct.
24244 * opts-common.c (decode_cmdline_option): Handle
24245 CL_ERR_INT_RANGE_ARG.
24246 (cmdline_handle_error): Likewise.
24247 * opts.c (print_filtered_help): Show valid interval in
24248 when --help is provided.
24249 * opts.h (struct cl_option): Add range_min and range_max fields.
24250 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
24252 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
24254 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
24255 (x * C EQ/NE y * C): New transformation.
24257 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
24259 * genmultilib (combination_space): Accept '+' in option names.
24261 2017-06-28 Martin Liska <mliska@suse.cz>
24264 * asan.c (instrument_derefs): Bail out inner references
24265 that are hard register variables.
24267 2017-06-28 Jakub Jelinek <jakub@redhat.com>
24270 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
24271 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
24273 2017-06-28 Richard Biener <rguenther@suse.de>
24275 * tree-vectorizer.h (vect_get_vec_defs): Remove.
24276 (vect_get_slp_defs): Adjust.
24277 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
24279 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
24281 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
24282 get_initial_defs_for_reduction instead of vect_get_vec_defs.
24283 (vectorizable_reduction): Adjust.
24284 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
24286 (vect_get_slp_defs): Likewise.
24287 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
24288 (vectorizable_bswap): Adjust.
24289 (vectorizable_call): Likewise.
24290 (vectorizable_conversion): Likewise.
24291 (vectorizable_assignment): Likewise.
24292 (vectorizable_shift): Likewise.
24293 (vectorizable_operation): Likewise.
24294 (vectorizable_store): Likewise.
24295 (vectorizable_condition): Likewise.
24296 (vectorizable_comparison): Likewise.
24298 2017-06-28 Michael Collison <michael.collison@arm.com>
24301 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
24303 (arm_gen_movmemqi): Removed unused variable 'i'.
24304 Convert 'for' loop into 'while' loop.
24305 (arm_expand_prologue): Remove last unnecessary set of insn.
24306 (thumb_pop): Remove unused variable 'pushed_words'.
24307 (thumb_exit): Remove last unnecessary set of regs_to_pop.
24309 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24311 * config/s390/predicates.md: Use s390_rel_address_ok_p.
24312 * config/s390/s390-protos.h: Add prototype of
24313 s390_rel_address_ok_p.
24314 * config/s390/s390.c (s390_got_symbol): New function.
24315 (s390_rel_address_ok_p): New function.
24316 (legitimize_pic_address): Use s390_rel_address_ok_p.
24317 (s390_load_got): Use s390_got_symbol.
24318 (s390_option_override): Issue error if
24319 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
24320 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
24322 * config/s390/s390.opt: New option mpic-data-is-text-relative.
24324 2017-06-27 Andrew Pinski <apinski@cavium.com>
24326 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
24327 (X * copysign (1.0, X)): New pattern.
24328 (X * copysign (1.0, -X)): New pattern.
24329 (copysign (-1.0, CST)): New pattern.
24331 2017-06-27 Joseph Myers <joseph@codesourcery.com>
24333 * genmultilib (combination_space): Remove variable.
24334 Validate reuse rules against regular expression for any sequence
24335 of multilib options in any order.
24337 2017-06-27 Michael Collison <michael.collison@arm.com>
24339 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
24340 call aarch64_split_simd_combine.
24341 * (aarch64_combine_internal<mode>): Delete pattern.
24342 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
24343 Allow register and subreg operands.
24345 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24347 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
24348 specific need, just fallback on defaults.
24349 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
24351 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24352 Olivier Hainque <hainque@adacore.com>
24354 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
24356 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
24357 targets. Pick a default if no particular attempt applied.
24358 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
24361 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24363 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
24364 (x86_64-wrs-vxworks7): Likewise.
24366 2017-06-27 Marek Polacek <polacek@redhat.com>
24369 * ubsan.c (instrument_null): Check get_base_address's result for null.
24371 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24373 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
24375 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
24377 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
24378 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
24379 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
24380 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
24381 New function types.
24382 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
24383 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
24384 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
24385 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
24386 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
24387 BUILT_IN_FEUPDATEENV): New builtins.
24388 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
24389 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
24390 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
24391 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
24393 (builtin_structptr_types): Adjust size.
24394 * tree.c (builtin_structptr_types): Add four entries.
24396 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24397 Olivier Hainque <hainque@adacore.com>
24399 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
24400 (TLS_SYM): New local macro, forcing reference to __tls__ on
24401 link command lines for VxWorks 7 RTPs, triggering initialization
24403 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
24404 OS features TLS support, true for RTPs on VxWorks 7.
24405 * config/vxworks.c (vxworks_override_options): Setup emutls
24408 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24410 * predict.c (test_prediction_value_range): Use -1U instead of -1
24411 to avoid narrowing conversion warning.
24412 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
24413 to avoid narrowing conversion warning.
24414 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
24416 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
24418 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
24420 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
24421 64bit configurations.
24422 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
24423 (SIZE_TYPE): Likewise.
24424 * config/vxworks.c (vxworks_emutls_var_fields): Use
24425 long_unsigned_type_node instead of unsigned_type_node as the offset
24426 field type, which is "pointer" mode in emutls.c.
24428 2017-06-27 Jakub Jelinek <jakub@redhat.com>
24431 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
24433 PR middle-end/81207
24434 * gimple-fold.c (replace_call_with_call_and_fold): Handle
24435 gimple_vuse copying separately from gimple_vdef copying.
24437 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24439 * value-prof.c (free_hist): Remove call to memset and the enclosing if
24442 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24443 Olivier Hainque <hainque@adacore.com>
24445 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
24446 for all vxworks7 targets.
24447 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
24448 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
24449 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
24450 variations for VX6/VX7 and 32/64bits later on in ...
24451 (VXWORKS_LIB_SPEC): Leverage new macros.
24452 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
24453 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
24455 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
24457 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
24458 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
24460 2017-06-26 Carl Love <cel@us.ibm.com>
24462 * config/rs6000/rs6000-c.c: Add support for built-in functions
24463 vector bool char vec_reve (vector bool char);
24464 vector signed char vec_reve (vector signed char);
24465 vector unsigned char vec_reve (vector unsigned char);
24466 vector bool int vec_reve (vector bool int);
24467 vector signed int vec_reve (vector signed int);
24468 vector unsigned int vec_reve (vector unsigned int);
24469 vector bool long long vec_reve (vector bool long long);
24470 vector signed long long vec_reve (vector signed long long);
24471 vector unsigned long long vec_reve (vector unsigned long long);
24472 vector bool short vec_reve (vector bool short);
24473 vector signed short vec_reve (vector signed short);
24474 vector double vec_reve (vector double);
24475 vector float vec_reve (vector float);
24476 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
24477 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
24478 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
24479 (altivec_vreve): New pattern.
24480 * config/rs6000/altivec.h (vec_reve): New define.
24481 * doc/extend.texi (vec_rev): Update the built-in documentation file
24482 for the new built-in functions.
24484 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24486 PR tree-optimization/71815
24487 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
24489 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
24491 (slsr_process_phi): Likewise.
24492 (replace_uncond_cands_and_profitable_phis): Don't replace a
24493 multiply candidate with a stride of 1 (copy or cast).
24494 (phi_incr_cost): Call uses_consumed_by_stmt rather than
24496 (lowest_cost_path): Likewise.
24497 (total_savings): Likewise.
24499 2017-06-26 Richard Biener <rguenther@suse.de>
24502 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
24503 Use def_builtin_pure for all gather builtins.
24505 2017-06-26 Richard Biener <rguenther@suse.de>
24507 PR tree-optimization/81203
24508 * tree-tailcall.c (find_tail_calls): Do not move stmts into
24509 non-dominating BBs.
24511 2017-06-26 Marek Polacek <polacek@redhat.com>
24514 * doc/invoke.texi: Document -Wmultistatement-macros.
24516 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
24518 * doc/sourcebuild.texi (ARM-specific attributes): Document new
24519 arm_neon_ok_no_float_abi effective target.
24521 2017-06-26 Richard Biener <rguenther@suse.de>
24523 PR tree-optimization/80928
24524 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
24525 (copy_bbs): Set BB_DUPLICATED flag early.
24526 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
24528 (execute_on_shrinking_pred): Likewise.
24529 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
24530 BB_DUPLICATED blocks.
24531 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
24532 iterate over all PHIs considering removal of *gsi.
24534 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
24536 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
24539 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24541 * config/rs6000/rs6000-string.c: (expand_block_clear,
24542 do_load_for_compare, select_block_compare_mode,
24543 compute_current_alignment, expand_block_compare,
24544 expand_strncmp_align_check, expand_strn_compare,
24545 expand_block_move, rs6000_output_load_multiple)
24546 Move functions related to string/block move/compare
24547 to a separate file.
24548 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
24549 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
24550 for this function which is now used in two files.
24551 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
24552 * config.gcc: Add rs6000-string.o to extra_objs for
24553 targets powerpc*-*-* and rs6000*-*-*.
24555 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
24558 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
24559 32-bit, since indexed is not valid for DImode.
24560 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
24561 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
24562 (define_peephole2 for Altivec d-form load): Add 32-bit support.
24563 (define_peephole2 for Altivec d-form store): Likewise.
24566 * multiple_target.c (create_dispatcher_calls): Only create the
24567 dispatcher call if the function is the default clone of a
24568 versioned function.
24570 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
24572 PR middle-end/80902
24573 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
24574 a call, force the call to not be a tail call.
24576 2017-06-23 Jeff Law <law@redhat.com>
24578 * doc/contrib.texi: Add entry for Steven Pemberton's work on
24581 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
24583 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
24584 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
24585 handling for early expansion of vector shifts (sl,sr,sra,rl).
24586 (builtin_function_type): Add vector shift right instructions
24587 to the unsigned argument list.
24589 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
24591 rtl-optimizatoin/79286
24592 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
24593 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
24594 trap. PIC register plus a const unspec without offset can never trap.
24596 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
24598 * tree.h (builtin_structptr_type): New type.
24599 (builtin_structptr_types): Declare new array.
24600 * tree.c (builtin_structptr_types): New array.
24601 (free_lang_data, build_common_tree_nodes): Use it.
24603 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
24606 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
24609 2017-06-22 Matt Turner <mattst88@gmail.com>
24611 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
24612 Lake models to skylake case. Assume skylake for unknown
24613 models with clflushopt.
24615 2017-06-22 Jeff Law <law@redhat.com>
24617 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
24618 frame sizes that do not satisfy aarch64_uimm12_shift.
24620 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
24622 * profile-count.h (apply_probability,
24623 apply_scale, probability_in): Fix checks for zero.
24625 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24627 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
24628 * doc/cppdiropts.texi (-I @var{dir}): Document it.
24630 2016-06-22 Richard Biener <rguenther@suse.de>
24632 * tree-vect-loop.c (vect_model_reduction_cost): Handle
24633 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
24634 REDUC_MAX_EXPR support.
24635 (vectorizable_reduction): Likewise.
24636 (vect_create_epilog_for_reduction): Likewise.
24638 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
24640 * match.pd (A / (1 << B) -> A >> B): New.
24641 * generic-match-head.c: Include optabs-tree.h.
24642 * gimple-match-head.c: Likewise.
24643 * optabs-tree.h (target_supports_op_p): New.
24644 * optabs-tree.c (target_supports_op_p): New.
24646 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24648 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
24649 $gcc_cv_ld --help output.
24650 (gcc_cv_ld_demangle): Likewise.
24651 (gcc_cv_ld_eh_frame_hdr): Likewise.
24652 (gcc_cv_ld_pie): Likewise.
24653 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
24654 (gcc_cv_ld_buildid): Likewise.
24655 (gcc_cv_ld_sysroot): Likewise.
24656 (ld_bndplt_support): Likewise.
24657 (ld_pushpopstate_support): Likewise.
24658 * configure: Regenerate.
24659 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
24661 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24664 * config/i386/sse.md (round<mode>2): Renumber match_dup and
24665 operands indexes to avoid gap between operands and match_dups.
24667 2017-06-21 Andrew Pinski <apinski@cavium.com>
24669 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
24670 Increment Arith_shift and Arith_shift_reg by 1.
24671 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
24673 * config/aarch64/aarch64.c (thunderx_tunings): Enable
24674 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
24675 (aarch64_strip_extend): Add new argument and test for it.
24676 (aarch64_cheap_mult_shift_p): New function.
24677 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
24678 add a cost if it is true.
24679 Update calls to aarch64_strip_extend.
24680 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
24682 2017-06-21 Andrew Pinski <apinski@cavium.com>
24684 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
24686 (thunderxt88): Likewise.
24687 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
24688 (thunderx_prefetch_tune): New variable.
24689 (thunderx2t99_prefetch_tune): Update for the correct values.
24690 (thunderxt88_tunings): New variable.
24691 (thunderx_tunings): Use thunderx_prefetch_tune instead of
24692 generic_prefetch_tune.
24693 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
24695 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24697 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
24698 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
24699 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
24700 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
24701 (aarch64_atomic_cas<mode>, GPI): Likewise.
24703 2017-06-21 Martin Liska <mliska@suse.cz>
24705 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
24706 statements on cold and hot labels.
24707 * predict.c (tree_estimate_probability_bb): Remove the
24708 prediction from this place.
24710 2017-06-21 Martin Liska <mliska@suse.cz>
24712 PR tree-optimization/79489
24713 * gimplify.c (maybe_add_early_return_predict_stmt): New
24715 (gimplify_return_expr): Call the function.
24716 * predict.c (tree_estimate_probability_bb): Remove handling
24718 * predict.def: Update comment about early return predictor.
24719 * gimple-predict.h (is_gimple_predict): New function.
24720 * predict.def: Change default value of early return to 66.
24721 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
24723 * passes.def: Put pass_strip_predict_hints to the beginning of
24726 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
24728 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
24729 FUNCTION_DECL declarations.
24730 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
24732 (dwaf2out_decl): Likewise.
24733 * godump.c (go_early_global_decl): Skip call to the real debug hook
24734 for FUNCTION_DECL declarations.
24735 * passes.c (rest_of_decl_compilation): Skip call to the
24736 early_global_decl debug hook for FUNCTION_DECL declarations, unless
24737 -fdump-go-spec is passed.
24739 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24741 * config/i386/i386.c (struct builtin_isa): New field pure_p.
24742 Reorder for compactness.
24743 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
24744 (def_builtin_pure, def_builtin_pure2): New functions.
24745 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
24747 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
24749 * match.pd (nop_convert): New predicate.
24750 ((A +- CST1) +- CST2): Allow some NOP conversions.
24752 2017-06-21 Jakub Jelinek <jakub@redhat.com>
24755 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
24756 with ctors/dtors if GOVD_SHARED is set.
24758 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24760 * config/aarch64/aarch64.md (movti_aarch64):
24761 Emit mov rather than orr.
24762 (movtf_aarch64): Likewise.
24763 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
24764 Emit mov rather than orr.
24766 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24768 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
24769 Swap alternatives, make integer dup more expensive.
24771 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
24773 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
24774 Return true for non-tls symbols.
24776 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
24778 * config/aarch64/aarch64-cores.def (cortex-a55): New.
24779 (cortex-a75): Likewise.
24780 (cortex-a75.cortex-a55): Likewise.
24781 * config/aarch64/aarch64-tune.md: Regenerate.
24782 * doc/invoke.texi (-mtune): Document new values for -mtune.
24784 2017-06-21 Tom de Vries <tom@codesourcery.com>
24786 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
24787 stack_size feature.
24788 (Effective-Target Keywords, Other attributes): Suggest using
24789 dg-add-options stack_size feature to get stack limit in stack_size
24790 effective target documentation.
24792 2017-06-21 Julian Brown <julian@codesourcery.com>
24793 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24795 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
24796 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
24797 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
24799 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
24800 attribute type list for neon_multiply.
24801 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
24802 attribute type list for neon_multiply.
24803 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
24804 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
24805 attribute type list for neon_multiply.
24806 * config/arm/types.md (crypto_pmull): Add.
24807 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
24808 attribute type list.
24810 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
24812 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
24815 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24817 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
24818 to make sure not to dereference a NULL cost_classes_ptr pointer.
24820 2017-06-20 Carl Love <cel@us.ibm.com>
24822 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24823 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24824 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24825 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24826 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24827 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24828 VMULOSW): New enum "unspec" values.
24829 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24830 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24831 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24832 altivec_vmulosw): New patterns.
24833 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24834 VMULOSW): Add definitions.
24836 2017-06-20 Julia Koval <julia.koval@intel.com>
24838 * config/i386/i386.c: Fix rounding expand for new pattern.
24839 * config/i386/subst.md: Fix pattern (parallel -> unspec).
24841 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24843 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
24844 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
24846 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24848 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
24851 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
24853 * config/aarch64/aarch64-cores.def: Rearrange to sort by
24854 architecture, then by implementer ID.
24855 * config/aarch64/aarch64-tune.md: Regenerate.
24857 2017-06-20 Richard Biener <rguenther@suse.de>
24859 PR middle-end/81097
24860 * fold-const.c (split_tree): Fold to type before negating.
24862 2017-06-20 David Malcolm <dmalcolm@redhat.com>
24864 * diagnostic-show-locus.c
24865 (selftest::test_fixit_deletion_affecting_newline): New function.
24866 (selftest::diagnostic_show_locus_c_tests): Call it.
24868 2017-06-20 Andreas Schwab <schwab@suse.de>
24871 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
24874 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24876 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
24877 __ARM_FEATURE_COPROC according to support.
24879 2017-06-20 Jakub Jelinek <jakub@redhat.com>
24881 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
24882 Rewritten to avoid overflow for > 32-bit pointers.
24885 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
24886 by removing enum keyword.
24887 (ubsan_type_descriptor): Likewise. Formatting fix.
24890 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
24891 splitter): Require TARGET_SSE2 in the condition.
24893 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
24896 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
24897 for doing vector set of SFmode on ISA 3.0.
24898 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
24899 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
24901 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
24902 SFmode value into a V4SF variable that was extracted from another
24903 V4SF variable without converting the element to double precision
24904 and back to single precision vector format.
24905 (vsx_insert_extract_v4sf_p9_2): Likewise.
24907 2017-06-19 Jakub Jelinek <jakub@redhat.com>
24909 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
24910 in UWHI to avoid undefined overflow.
24913 * ubsan.h (enum ubsan_encode_value_phase): New.
24914 (ubsan_encode_value): Change second argument to
24915 enum ubsan_encode_value_phase with default value of
24916 UBSAN_ENCODE_VALUE_GENERIC.
24917 * ubsan.c (ubsan_encode_value): Change second argument to
24918 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
24919 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
24920 create_tmp_var_raw instead of create_tmp_var and use a
24922 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
24923 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
24924 ubsan_encode_value callers.
24927 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
24928 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
24929 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
24931 2017-06-19 Richard Biener <rguenther@suse.de>
24933 PR middle-end/81118
24934 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
24935 estimates if we changed anything.
24937 2017-06-19 Richard Biener <rguenther@suse.de>
24939 PR tree-optimization/80887
24940 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
24941 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
24942 simplified lookups, then reset mprts_hook.
24943 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
24945 (try_to_simplify): Likewise.
24947 2017-06-19 Martin Liska <mliska@suse.cz>
24950 * gimplify.c (gimplify_switch_expr):
24951 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
24953 2017-06-19 Martin Liska <mliska@suse.cz>
24955 * doc/install.texi: Document that PGO runs in 4 stages.
24957 2017-06-19 Martin Liska <mliska@suse.cz>
24960 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
24961 to dispatcher function name.
24962 * multiple_target.c (replace_function_decl): New function.
24963 (create_dispatcher_calls): Redirect both edges and references.
24965 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
24967 * profile-count.c (profile_count::dump): Dump quality.
24968 (profile_count::differs_from_p): Update for unsigned val.
24969 * profile-count.h (profile_count_quality): New enum.
24970 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
24972 2017-06-19 Richard Biener <rguenther@suse.de>
24974 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
24975 struct function as arg.
24976 (estimate_numbers_of_iterations): Export overload with loop arg.
24977 (free_numbers_of_iterations_estimates_loop): Use an overload of
24978 free_numbers_of_iterations_estimates instead.
24979 * tree-cfg.c (remove_bb): Adjust.
24980 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
24981 * tree-parloops.c (gen_parallel_loop): Likewise.
24982 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
24984 (tree_unroll_loops_completely): Likewise.
24985 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
24986 Use an overload instead and export.
24987 (estimated_loop_iterations): Adjust.
24988 (max_loop_iterations): Likewise.
24989 (likely_max_loop_iterations): Likewise.
24990 (estimate_numbers_of_iterations): Take struct function as arg
24992 (loop_exits_before_overflow): Adjust.
24993 (free_numbers_of_iterations_estimates_loop): Use an overload.
24994 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
24995 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
24997 2017-06-19 Richard Biener <rguenther@suse.de>
25000 * ipa-prop.c (find_constructor_constant_at_offset): Handle
25001 RANGE_EXPR conservatively.
25003 2017-06-16 Carl Love <cel@us.ibm.com>
25005 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
25006 definitions for vec_float, vec_float2, vec_floato,
25007 vec_floate built-ins.
25008 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
25009 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
25011 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
25012 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
25013 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
25014 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
25015 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
25016 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
25017 vec_floato): Add builtin defines.
25018 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
25019 Update the built-in documentation file for the new built-in
25022 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25024 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
25025 (mthumb): Mark as the negative of -marm.
25027 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25029 * doc/invoke.texi (ARM Options, -mcpu): Document supported
25031 (ARM Options, -mtune): Document that this accepts the same
25032 extension options as -mcpu.
25033 (ARM Options, -mfpu): Document addition of -mfpu=auto.
25035 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25037 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
25038 permitted extensions.
25040 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25042 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
25043 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
25044 (armv8-m.main): Add option +nodsp.
25045 * config/arm/arm-cpu-cdata.h: Regenerated.
25047 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25049 * config/arm/t-fuchsia: New file.
25050 * config.gcc (arm*-*-fuchsia*): Use it.
25052 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25054 * config/arm/t-symbian: Rewrite for new option infrastructure.
25056 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25058 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
25059 (MULTILIB_REQUIRED): Likewise.
25061 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25063 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
25064 (MULTILIB_RESUE): Likewise.
25065 (MULTILIB_MATCHES): Likewise.
25066 (MULTLIB_REQUIRED): Likewise.
25068 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25070 * config/arm/t-rtems: Rewrite for new option framework.
25072 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25074 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
25075 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
25076 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
25077 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
25078 * config/arm/t-multilib: ... here.
25079 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
25080 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
25081 armv7-a and armv8*-a when A-profile libraries have not been built.
25082 * config/arm/t-rmprofile: Rewrite.
25084 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25086 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
25087 with a backslash. Remove the backslash after substituting unescaped
25089 * doc/fragments.texi (MULTILIB_REUSE): Document it.
25091 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25093 * config.gcc: (arm*-*-*): When building a-profile libraries, force
25094 the driver to pass through the default setting of -mfloat-abi.
25095 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
25097 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
25098 (all_feat_combs): New rule.
25099 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
25101 * config/arm/t-aprofile: Rewrite.
25103 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25105 * config/arm/arm.h (FPUTYPE_AUTO): Define.
25106 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
25107 fpu is not specified by the user/command-line.
25108 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
25109 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
25110 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
25111 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
25112 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
25113 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
25115 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25117 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
25118 * config/arm/t-arm-elf: Rewritten.
25120 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25122 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
25123 have some floating-point instructions.
25124 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
25125 (TARGET_MAYBE_HARD_FLOAT): New macro.
25126 * config/arm/arm-builtins.c (arm_init_builtins): Use
25127 TARGET_MAYBE_HARD_FLOAT.
25128 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
25130 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25132 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
25133 (configargs.h): Include it.
25134 (arm_print_hint_for_fpu_option): New function.
25135 (arm_parse_fpu_option): New function.
25136 (candidate_extension): New class.
25137 (arm_canon_for_multilib): New function.
25138 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
25139 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
25140 (ARCH_CANONICAL_SPECS): New macro.
25141 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
25143 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25145 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
25146 are set after handling multilib fragments. Set target_cpu_default2
25149 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25151 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
25153 (arm*-*-*): Set target_cpu_default2 to a quoted string.
25154 * config/arm/parsecpu.awk (check_cpu): Validate any extension
25156 (check_arch): Likewise.
25157 * config/arm/arm.c (arm_configure_build_target): Handle
25158 TARGET_CPU_DEFAULT being a string constant. Scan any feature
25159 options in the default.
25161 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25163 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
25164 when an option is an alias of another.
25165 * config/arm/parsecpu.awk (optalias): New parser token.
25166 (gen_comm_data): Mark non-alias options as such. Emit entries
25167 for extension aliases.
25168 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
25169 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
25170 (armv6kz, armv6zk, armv6t2): Likewise.
25171 (armv7): Make vfpv3-d16 an alias.
25172 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
25174 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
25175 Sort in canonical order.
25176 (armv8-a): Sort in canonical order.
25177 (armv8.1-a, armv8.2-a): Likewise.
25178 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
25180 (cortex-a9): Sort in canonical order.
25181 * config/arm/arm.c (selftests.h): Include it.
25182 (arm_test_cpu_arch_data): New function.
25183 (arm_run_self_tests): New function.
25184 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
25185 (targetm): Move declaration to the end of the file.
25186 * arm-cpu-cdata.h: Regenerated.
25188 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25190 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
25191 call to target_mode_check describing the type of option passed.
25192 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
25193 (arm_target_thumb_only): Use arm_parse_arch_option_name or
25194 arm_parse_cpu_option_name to match parameters against list of
25196 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
25197 arm_arch_core_flags data structure.
25198 * config/arm/arm-cpu_cdata.h: Regenerated.
25200 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25202 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
25204 (arm_print_hint_for_cpu_option): Likewise.
25205 (arm_print_hint_for_arch_option): Likewise.
25206 (arm_parse_cpu_option_name): Likewise.
25207 (arm_parse_arch_option_name): Likewise.
25208 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
25209 of entries in the all_fpus list.
25210 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
25211 (arm_parse_cpu_option_name): Declare.
25212 (arm_parse_arch_option_name): Declare.
25213 (arm_parse_option_features): Declare.
25214 (arm_intialize_isa): Declare.
25215 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
25217 (gen_comm_data): ... here. Make definitions non-static.
25218 * config/arm/arm-cpu-data.h: Regenerated.
25219 * config/arm/arm-cpu-cdata.h: Regenerated.
25221 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25223 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
25224 (cpu_arch_extension): New structure.
25225 (cpu_arch_option, arch_option, cpu_option): New structures.
25226 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
25227 architecture types.
25228 (gen_data): Generate new format data tables.
25229 * config/arm/arm.c (cpu_tune): New structure.
25230 (cpu_option, processors): Delete.
25231 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
25232 (arm_print_hint_for_cpu_option): ... this and ...
25233 (arm_print_hint_for_arch_option): ... this.
25234 (arm_parse_arch_cpu_name): Delete. Replace with ...
25235 (arm_parse_cpu_option_name): ... this and ...
25236 (arm_parse_arch_option_name): ... this.
25237 (arm_unrecognized_feature): Change type of target parameter to
25239 (arm_parse_arch_cpu_features): Delete. Replace with ...
25240 (arm_parse_option_features): ... this.
25241 (arm_configure_build_target): Rework to use new configuration data
25243 (arm_print_tune_info): Rework for new configuration data tables.
25244 * config/arm/arm-cpu-data.h: Regenerated.
25245 * config/arm/arm-cpu.h: Regenerated.
25247 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25249 * Makefile.in (OBJS): Move sbitmap.o from here ...
25250 (OBJS-libcommon): ... to here.
25252 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25254 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
25255 (ISA_ALL_CRYPTO): New macro.
25256 (ISA_ALL_SIMD): New macro
25257 (ISA_ALL_FP): New macro.
25258 * config/arm/arm.c (fpu_bitlist): Update initializer.
25259 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
25261 (arm9e): Add fpu. Add option for nofp
25262 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
25263 (arm926ej-s, arm1026ej-s): Likewise.
25264 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
25265 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
25266 neon-fp16, neon-vfpv4, nofp and nosimd.
25267 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
25268 (cortex-a8): Add fpu. Add option for nofp.
25269 (cortex-a9): Add fpu. Add options for nosimd and nofp.
25270 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
25271 (cortex-r4f): Add fpu.
25272 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
25273 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
25275 (cortex-r8): Likewise.
25276 (cortex-m4): Add fpu. Add option for nofp.
25277 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
25278 (cortex-a17.cortex-a7): Likewise.
25279 (cortex-a32): Add fpu. Add options for crypto and nofp.
25280 (cortex-a35, cortex-a53): Likewise.
25281 (cortex-a57): Add fpu. Add option for crypto.
25282 (cortex-a72, cortex-a73): Likewise.
25283 (exynos-m1): Likewise.
25284 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
25285 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
25286 (cortex-m33): Add fpu. Add option for nofp.
25287 * config/arm/arm-cpu-cdata.h: Regenerated
25288 * config/arm/arm-cpu-data.h: Regenerated.
25290 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25292 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
25293 (armv5te, armv5tej): Likewise.
25294 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
25295 (armv7): Add options fp and vfpv3-d16.
25296 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
25297 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
25299 (armv7ve): Likewise.
25300 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
25301 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
25302 (armv8-a): Add nocrypto option.
25303 (armv8.1-a, armv8.2-a): Likewise.
25304 (armv8-m.main): add options fp, fp.dp and nofp.
25306 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25308 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
25310 (armv8-a+crc): Delete.
25311 (armv8.1-a): Add options simd, crypto and nofp.
25312 (armv8.2-a): Add options fp16, simd, crypto and nofp.
25313 (armv8.2-a+fp16): Delete.
25314 (armv8-m.main): Add option dsp.
25315 (armv8-m.main+dsp): Delete.
25316 (cortex-a8): Add fpu. Add nofp option.
25317 (cortex-a9): Add fpu. Add nofp and nosimd options.
25318 * config/arm/parsecpu.awk (gen_data): Generate option tables and
25319 link to main cpu and architecture data structures.
25320 (gen_comm_data): Only put isa attributes from the main architecture
25322 (option): New statement for architecture and CPU entries.
25323 * arm.c (struct cpu_option): New structure.
25324 (struct processors): Add entry for options.
25325 (arm_unrecognized_feature): New function.
25326 (arm_parse_arch_cpu_name): Ignore any characters after the first
25328 (arm_parse_arch_cpu_feature): New function.
25329 (arm_configure_build_target): Separate out any CPU and architecture
25330 features and parse separately. Don't error out if -mfpu=auto is
25331 used with only an architecture string.
25332 (arm_print_asm_arch_directives): New function.
25333 (arm_file_start): Call it.
25334 * config/arm/arm-cpu-cdata.h: Regenerated.
25335 * config/arm/arm-cpu-data.h: Likewise.
25336 * config/arm/arm-tables.opt: Likewise.
25338 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25340 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
25341 assembler when it is not -mfpu=auto.
25343 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25345 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
25346 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
25347 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
25348 (ASM_CPU_SPEC): Rewrite.
25349 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
25350 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
25351 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
25352 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
25353 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
25354 copied string is NUL-terminated. Also strip any characters prefixed
25356 (arm_rewrite_selected_arch): New function.
25357 (arm_rewrite_march): New function.
25359 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
25361 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
25362 (x_arm_cpu_string, x_arm_tune_string): Likewise.
25363 (march, mcpu, mtune): Convert to string-based options.
25364 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
25365 (arm_parse_arch_cpu_name): New function.
25366 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
25367 identify selected architecture or CPU.
25368 (arm_option_save): New function.
25369 (TARGET_OPTION_SAVE): Redefine.
25370 (arm_option_restore): Restore string options.
25371 (arm_option_print): Print string options.
25373 2017-06-16 Martin Sebor <msebor@redhat.com>
25375 PR tree-optimization/80933
25376 PR tree-optimization/80934
25377 * builtins.c (fold_builtin_3): Do not handle bcmp here.
25378 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
25379 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
25380 (gimple_fold_builtin): Call them.
25382 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25384 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
25385 as unlikely; update profile.
25387 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25389 * predict.c (force_edge_cold): Handle declaring edges impossible
25392 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25394 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
25396 (try_unroll_loop_completely): Fix reporting.
25398 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25400 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
25402 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
25405 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
25406 if given a non-constant argument for an intrinsic which requires a
25409 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25411 * profile.c (compare_freqs): New function.
25412 (branch_prob): Sort edge list.
25413 (find_spanning_tree): Assume that the list is priority sorted.
25415 2017-06-16 Richard Biener <rguenther@suse.de>
25417 PR tree-optimization/81090
25418 * passes.def (pass_record_bounds): Remove.
25419 * tree-pass.h (make_pass_record_bounds): Likewise.
25420 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
25421 make_pass_record_bounds): Likewise.
25422 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
25423 not free niter estimates at the beginning but at the end.
25424 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
25426 2017-06-16 Richard Biener <rguenther@suse.de>
25428 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
25429 initializer to workaround ICE in host GCC 4.8.
25431 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
25433 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
25435 (clone_inlined_nodes): Update.
25437 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25439 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
25440 prefetch settings, and enable prefetching by default at -O3.
25442 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25444 * config/aarch64/aarch64.c (aarch64_override_options_internal):
25445 Set flag_prefetch_loop_arrays according to tuning data.
25447 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
25449 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
25450 New tune structure.
25451 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
25452 [Unrelated to main purpose of the patch] Place the pointer field last
25453 to enable type checking errors when tune structure are wrongly merged.
25454 * config/aarch64/aarch64.c (generic_prefetch_tune,)
25455 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
25456 (thunderx2t99_prefetch_tune): New tune constants.
25457 (tune_params *_tunings): Update all tunings (no functional change).
25458 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
25459 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
25460 from tunings structures.
25462 2017-06-16 Jakub Jelinek <jakub@redhat.com>
25465 * ubsan.c (instrument_null): Add T argument, use it instead
25466 of computing it based on IS_LHS.
25467 (instrument_object_size): Likewise.
25468 (pass_ubsan::execute): Adjust instrument_null and
25469 instrument_object_size callers to pass gimple_get_lhs or
25470 gimple_assign_rhs1 result to it. Use instrument_null instead of
25471 calling get_base_address and instrument_mem_ref. Handle
25472 aggregate call arguments for object-size sanitization.
25474 2017-06-16 Yury Gribov <tetra2005@gmail.com>
25476 PR tree-optimization/81089
25477 * tree-vrp.c (is_masked_range_test): Validate operands of
25480 2017-06-15 Martin Sebor <msebor@redhat.com>
25483 * dumpfile.c (dump_register): Avoid calling memset to initialize
25484 a class with a default ctor.
25485 * gcc.c (struct compiler): Remove const qualification.
25486 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
25487 * hash-table.h: Ditto.
25488 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
25490 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
25491 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
25493 * params.h (struct param_info): Make struct members non-const.
25494 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
25495 with default initialization.
25496 * vec.h (vec_copy_construct, vec_default_construct): New helper
25498 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
25499 with vec_copy_construct.
25500 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
25501 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
25502 * doc/invoke.texi (-Wclass-memaccess): Document.
25504 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25506 * emit-rtl.h (is_leaf): Update comment about local
25507 register allocator.
25509 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
25512 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
25513 for a variable to have a section before checking if the section has a
25515 Set section to.persistent if persistent attribute is set.
25516 Warn if .persistent attribute is used on an automatic variable.
25518 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
25520 PR rtl-optimization/80474
25521 * reorg.c (update_block): Do not ignore instructions in a delay slot.
25523 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
25525 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
25528 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
25530 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
25531 (casesi): Emit bounds checking as RTL.
25532 (casesi_internal_mips16_<mode>): Remove bounds checking.
25534 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
25536 * config/xtensa/xtensa.c (xtensa_option_override): Append
25537 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
25538 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
25539 xtensa_doloop_hooks): Define unconditionally.
25540 (xtensa_reorg_loops): Only call reorg_loops in the presence of
25542 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
25543 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
25544 for it in xtensa_option_override.
25545 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
25546 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
25548 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
25550 * doc/cppopts.texi: Document '-' special value to -MF.
25552 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
25554 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
25555 (cortex_a53_fconst): Likewise.
25556 (cortex_a53_fpmul): Likewise.
25557 (cortex_a53_f_load_64): Likewise.
25558 (cortex_a53_f_load_many): Likewise.
25559 (cortex_a53_advsimd_alu): Likewise.
25560 (cortex_a53_advsimd_alu_q): Likewise.
25561 (cortex_a53_advsimd_mul): Likewise.
25562 (cortex_a53_advsimd_mul_q): Likewise.
25563 (fpmac bypass): Add new bypass for fpmac-fpmac case.
25564 Add missing fmul, r2f_cvt and fconst cases.
25566 2017-06-14 Richard Biener <rguenther@suse.de>
25568 PR middle-end/81088
25569 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
25571 (fold_binary_loc): When associating do not treat pre-existing
25572 TREE_OVERFLOW on literal constants as a reason to allow
25573 TREE_OVERFLOW on associated literal constants.
25575 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25577 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
25578 (MASK_FEATURES): New macro.
25579 * config/sparc/sparc.c (sparc_option_override): Remove the special
25580 handling of -mfpu and generalize it to all MASK_FEATURES switches.
25582 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25584 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
25585 a division of 0 if non-call exceptions are enabled.
25587 2017-06-14 Andrew Pinski <apinski@cavium.com>
25588 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25591 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
25592 Improve vector initialization code gen for only variable case.
25594 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
25596 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
25598 2017-06-14 Richard Biener <rguenther@suse.de>
25600 PR tree-optimization/81083
25601 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
25604 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25606 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
25607 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
25608 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
25609 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
25610 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
25611 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
25613 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25615 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
25616 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
25618 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25620 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
25622 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25624 * config/rs6000/t-rtems: Don't handle SPE.
25626 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25628 * config/rs6000/t-linux: Don't handle SPE.
25630 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25632 * config/rs6000/eabispe.h: Delete file.
25634 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25636 * config/rs6000/t-spe: Delete file.
25638 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25640 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
25641 (rs6000_legitimate_offset_address_p): Return false for anything in
25642 V2SImode or V2SFmode.
25644 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
25646 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
25647 except V2SF and V2SI. Rearrange the vector modes, and add comments.
25648 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
25650 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
25651 (rs6000_legitimate_offset_address_p): Ditto.
25652 (rs6000_emit_move): Ditto.
25653 (rs6000_init_builtins): Remove V4HI_type_node.
25655 2017-06-13 Martin Liska <mliska@suse.cz>
25658 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
25659 (gate_asan): Likewise.
25660 * asan.h (asan_no_sanitize_address_p): Remove the function.
25661 (sanitize_flags_p): New function.
25662 * builtins.def: Fix coding style.
25663 * common.opt: Use renamed enum value.
25664 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
25665 * doc/extend.texi: Document no_sanitize attribute.
25666 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
25667 to SANITIZE_UNDEFINED_NONDEFAULT.
25668 * gcc.c (sanitize_spec_function): Use the renamed enum value.
25669 * gimple-fold.c (optimize_atomic_compare_exchange_p):
25670 Use sanitize_flags_p.
25671 * gimplify.c (gimplify_function_tree): Likewise.
25672 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
25673 * opts.c (parse_no_sanitize_attribute): New function.
25674 (common_handle_option): Use renamed enum value.
25675 * opts.h (parse_no_sanitize_attribute): Declare.
25676 * tree.c (sanitize_flags_p): New function.
25677 * tree.h: Declared here.
25678 * tsan.c: Use sanitize_flags_p.
25679 * ubsan.c (ubsan_expand_null_ifn): Likewise.
25680 (instrument_mem_ref): Likewise.
25681 (instrument_bool_enum_load): Likewise.
25682 (do_ubsan_in_current_function): Remove the function.
25683 (pass_ubsan::execute): Use sanitize_flags_p.
25684 * ubsan.h: Remove do_ubsan_in_current_function
25685 * tree-cfg.c (print_no_sanitize_attr_value): New function.
25686 (dump_function_to_file): Use it here.
25688 2017-06-13 Martin Jambor <mjambor@suse.cz>
25690 PR tree-optimization/80803
25691 PR tree-optimization/81063
25692 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
25693 (propagate_subaccesses_across_link): Enqueue subtree whenever
25694 necessary instead of relying on the caller.
25696 2017-06-13 Martin Jambor <mjambor@suse.cz>
25698 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
25699 that have a first_link.
25700 (sort_and_splice_var_accesses): Do not check first_link before
25702 (subtree_mark_written_and_enqueue): Likewise.
25703 (propagate_all_subaccesses): Likewise and do not stop at first
25704 parent with a first_link.
25706 2017-06-13 Martin Jambor <mjambor@suse.cz>
25708 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
25711 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25713 * match.pd: New pattern.
25715 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25717 * tree-vrp.c (is_masked_range_test): New function.
25718 (register_edge_assert_for): Determine ranges for
25721 2017-06-13 Yury Gribov <tetra2005@gmail.com>
25723 PR tree-optimization/67328
25724 * fold-const.c (maskable_range_p): New function.
25725 (build_range_check): Generate bittests if possible.
25727 2017-06-13 Martin Liska <mliska@suse.cz>
25729 * gimple-pretty-print.c (dump_probability): Add new argument.
25730 (dump_edge_probability): Dump both probability and count.
25731 (dump_gimple_label): Likewise.
25732 (dump_gimple_bb_header): Likewise.
25734 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
25737 * config/avr/avr-devices.c: Fix indentation.
25738 * config/avr/gen-avr-mmcu-specs.c: Dito.
25740 2017-06-13 Richard Biener <rguenther@suse.de>
25742 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
25743 instead get vector type from stmt_info.
25744 (vectorizable_reduction): Adjust. Remove dead code.
25746 2017-06-13 Richard Biener <rguenther@suse.de>
25748 PR middle-end/81065
25749 * fold-const.c (extract_muldiv_1): Remove bogus distribution
25750 case of C * (x * C2 + C3).
25751 (fold_addr_of_array_ref_difference): Properly fold index difference.
25753 2017-06-12 David S. Miller <davem@davemloft.net>
25756 * config/sparc/sparc.md (return expander): Emit frame blockage if
25757 function uses alloca.
25759 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
25761 * combine.c (make_field_assignment): Check len rather than the mode
25762 precision when calling force_to_mode.
25764 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
25766 Support multilibs and devices that see flash in RAM address range.
25769 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
25770 (avr_mcu_t) <flash_pm_offset>: New field.
25771 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
25772 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
25773 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
25774 (AVR_TINY_PM_OFFSET): Remove macro.
25775 * config/avr/avr.opt (-mshort-calls): New option.
25776 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
25777 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
25778 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
25779 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
25780 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
25781 instead of avr_arch->have_jmp_call.
25782 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
25783 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
25784 avr_arch->flash_pm_offset to define.
25785 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
25786 new field flash_pm_offset. Add entry for avrxmega3.
25787 (avr_texinfo): Add entry for avrxmega3.
25788 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
25789 attiny212, attiny214,
25790 attiny412, attiny414, attiny416, attiny417,
25791 attiny814, attiny816, attiny817,
25792 attiny1614, attiny1616, attiny1617,
25793 attiny3214, attiny3216, attiny3217.
25794 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
25795 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
25796 (avr_print_operand_address) [AVR_TINY]: Same.
25797 (avr_asm_init_sections) <readonly_data_section>: Only patch
25798 callback if avr_arch->flash_pm_offset = 0.
25799 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
25800 for rodata if avr_arch->flash_pm_offset != 0.
25801 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
25802 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
25803 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
25804 (m_options): Append opt_rcall.
25805 (m_dirnames): Append dir_rcall.
25806 * config/avr/t-multilib: Regenerate.
25808 * configure.ac [target=avr]: Check whether avrxmega3 default
25809 linker description file works as needed.
25810 * configure: Regenerate.
25811 * doc/avr-mmcu.texi: Regenerate.
25812 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
25813 <__AVR_ARCH__>: Document avrxmega3 and 103.
25814 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
25815 <__AVR_SHORT_CALLS__>: Document it.
25816 <__AVR_PM_BASE_ADDRESS__>: Document it.
25817 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
25818 (AVR Variable Attributes) <progmem>: Document this is
25819 not needed for avrxmega3.
25820 (AVR Named Address Spaces) <__flash>: Dito.
25822 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
25824 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
25826 2017-06-12 Doug Rupp <rupp@adacore.com>
25828 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
25829 Append vxworks-stdint.h to the tm_file list.
25830 * config/vxworks-stdint.h: New file.
25832 2017-06-12 Martin Liska <mliska@suse.cz>
25834 PR tree-optimization/81041
25835 * tree-profile.c (gimple_gen_ic_func_profiler):
25836 Create an extra BB in profile-generate
25837 (gimple_gen_time_profiler): Likewise.
25839 2017-06-12 Jakub Jelinek <jakub@redhat.com>
25841 PR tree-optimization/81003
25842 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
25843 (update_range_test): Use it instead of force_gimple_operand_gsi.
25845 2017-06-12 Richard Biener <rguenther@suse.de>
25847 PR tree-optimization/81053
25848 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
25849 with backedge value not defined in loop. Simplify def stmt
25852 2017-06-11 Tom de Vries <tom@codesourcery.com>
25855 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
25857 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
25858 nvptx_cannot_force_const_mem.
25860 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25862 * opts.c (finish_options): Move test for flag_split_stack after
25863 it has been initialized.
25865 2017-06-11 Jason Merrill <jason@redhat.com>
25867 * tree.h (id_equal): New.
25868 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
25869 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
25870 instead of strcmp of IDENTIFIER_POINTER.
25872 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25874 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
25875 (mark_all_inlined_calls_cdtor): Fix formating.
25876 (inline_transform): Rescale profile before inlining.
25878 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25880 * cgraph.h (cgraph_edge::clone): Update prototype.
25881 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
25882 (cgraph_node::create_clone): Update.
25883 (cgraph_node::create_version_clone): Update.
25884 * tree-inline.c (copy_bb): Update.
25885 (expand_call_inline): Update.
25887 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25889 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
25890 factored out from ...
25891 (rs6000_emit_prologue): ... here.
25893 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
25895 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
25896 factored out from ...
25897 (rs6000_emit_prologue): ... here.
25899 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25901 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
25903 (handle_missing_profiles): Fix computation of tp_first_run.
25904 (counts_to_freqs): Do not touch freqs when count is 0.
25906 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
25908 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
25911 2017-06-10 Tom de Vries <tom@codesourcery.com>
25913 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
25914 attributes): Document signal effective target.
25916 2017-06-10 Tom de Vries <tom@codesourcery.com>
25918 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
25919 Document effective target stack_size.
25921 2017-06-09 David Malcolm <dmalcolm@redhat.com>
25923 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
25924 to the edit_context if they can be auto-applied.
25926 2017-06-9 Ian Lance Taylor <iant@golang.org>
25928 * opts.c (finish_options): If -fsplit-stack, disable implicit
25929 -forder-blocks-and-partition.
25930 * doc/invoke.texi (Optimize Options): Document that when using
25931 -fsplit-stack -forder-blocks-and-partition is not implicitly
25934 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25936 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
25937 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
25938 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
25939 * builtins.def (abort, trap, unreachable): Declare cold.
25940 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
25941 * tree-core.h (ECF_COLD): New.
25942 * tree.c (set_call_expr_flags): Handle ECF_COLD.
25943 (build_common_builtin_nodes): Mark unreachable and abort as cold.
25945 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25947 * predict.c (unlikely_executed_stmt_p): Cleanup.
25949 2017-06-09 Richard Biener <rguenther@suse.de>
25951 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
25952 model if the ref is always written to.
25954 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25956 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
25958 2017-06-09 Tamar Christina <tamar.christina@arm.com>
25960 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
25963 2017-06-09 Tom de Vries <tom@codesourcery.com>
25966 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
25967 "target cannot support label values" when encountering LABEL_REF.
25969 2017-06-09 Martin Liska <mliska@suse.cz>
25971 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
25972 (gimple_gen_ic_func_profiler): Emit direct comparison
25973 of __gcov_indirect_call_callee with NULL.
25974 (gimple_gen_time_profiler): Change probability from
25975 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
25977 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25979 * profile.c (edge_gcov_counts): Turn to pointer.
25980 (compute_branch_probabilities, compute_branch_probabilities): Update.
25981 (branch_prob): Do not clear edge_gcov_count.
25982 * profile.h (edge_gcov_counts): Turn to pointer.
25983 (edge_gcov_count): Update.
25985 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
25987 * gimple.h (gimple_check_failed): Mark cold.
25989 2017-06-09 Richard Biener <rguenther@suse.de>
25991 PR tree-optimization/66623
25992 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
25993 refactor check_reduction into two parts, properly computing
25994 whether we have to check reduction validity for outer loop
25997 2017-06-09 Richard Biener <rguenther@suse.de>
25999 PR tree-optimization/79483
26000 * graphite-scop-detection.c (order): New global.
26001 (get_order): Compute bb to order mapping that satisfies code
26002 generation constraints.
26003 (cmp_pbbs): New helper.
26004 (build_scops): Start domwalk at entry block, sort generated
26007 2017-06-09 Richard Biener <rguenther@suse.de>
26009 PR middle-end/81007
26010 * ipa-polymorphic-call.c
26011 (ipa_polymorphic_call_context::restrict_to_inner_class):
26012 Skip FIELD_DECLs with error_mark_node type.
26013 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
26016 2017-06-09 Martin Liska <mliska@suse.cz>
26018 * predict.c (struct branch_predictor): New struct.
26019 (test_prediction_value_range): New test.
26020 (predict_c_tests): New function.
26021 * selftest-run-tests.c (selftest::run_tests): Run the function.
26022 * selftest.h: Declare new tests.
26024 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
26027 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
26028 gen_add3_insn did not fail.
26029 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
26030 r0, construct that number in a temporary reg and add that reg to r0.
26031 If asked to put the result in r0 as well, fail.
26033 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
26035 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
26036 for early expansion of vec_eqv.
26038 2017-06-08 Jakub Jelinek <jakub@redhat.com>
26040 PR middle-end/81005
26041 * ubsan.c (instrument_null): Avoid pointless code temporary.
26042 (pass_ubsan::execute): Instrument aggregate arguments of calls.
26044 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
26048 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
26051 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
26054 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26056 * predict.c (unlikely_executed_edge_p): Move ahead.
26057 (probably_never_executed_edge_p): Use it.
26059 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26061 PR middle-end/79988
26062 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
26063 gimple_call_builtin_p call.
26065 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26067 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
26068 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
26069 rtl_check_failed_type2, rtl_check_failed_code1,
26070 rtl_check_failed_code2, rtl_check_failed_code_mode,
26071 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
26072 rtvec_check_failed_bounds, rtl_check_failed_flag,
26073 _fatal_insn_not_found, _fatal_insn): Likewise.
26074 * tree.h (tree_contains_struct_check_failed,
26075 tree_check_failed, tree_not_check_failed,
26076 tree_class_check_failed, tree_range_check_failed,
26077 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
26078 tree_vec_elt_check_failed, phi_node_elt_check_failed,
26079 tree_operand_check_failed, omp_clause_check_failed,
26080 omp_clause_operand_check_failed, omp_clause_range_check_failed):
26083 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26085 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
26086 flag_branch_probabilities.
26087 * ipa-inline.c (edge_badness): Likewise.
26088 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
26089 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
26090 * predict.c (maybe_hot_frequency_p): Likewise.
26091 (probably_never_executed): Likewise.
26092 * sched-ebb.c (schedule_ebbs): Likewise.
26093 * sched-rgn.c (find_single_block_region): Likewise.
26094 * tracer.c (tail_duplicate): Likewise.
26096 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26098 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
26099 longer requires x_flag_profile_use.
26101 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26103 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
26104 instead of flag_reorder_blocks_and_partition.
26105 * dbxout.c (dbxout_function_end): Likewise.
26106 * dwarf2out.c (gen_subprogram_die): Likewise.
26107 * haifa-sched.c (sched_create_recovery_edges): Likewise.
26108 * hw-doloop.c (reorg_loops): Likewise.
26109 * varasm.c (assemble_start_function,
26110 assemble_end_function): Likewise.
26111 (decide_function_section): Do not check for
26112 flag_reorder_blocks_and_partition.
26114 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
26116 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
26118 (chkp_get_hard_register_fake_addr_expr): Ditto.
26119 (chkp_build_addr_expr): Add check for hard reg case.
26120 (chkp_parse_array_and_component_ref): Ditto.
26121 (chkp_find_bounds_1): Ditto.
26122 (chkp_process_stmt): Don't generate bounds store for
26125 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26127 * predict.c (maybe_hot_bb_p): Do not check profile status.
26128 (maybe_hot_edge_p): Likewise.
26129 (probably_never_executed): Check for zero counts even if profile
26131 (unlikely_executed_edge_p): New function.
26132 (unlikely_executed_stmt_p): New function.
26133 (unlikely_executed_bb_p): New function.
26134 (set_even_probabilities): Use unlikely predicates.
26135 (combine_predictions_for_bb): Likewise.
26136 (predict_paths_for_bb): Likewise.
26137 (predict_paths_leading_to_edge): Likewise.
26138 (determine_unlikely_bbs): New function.
26139 (estimate_bb_frequencies): Use it.
26140 (compute_function_frequency): Use zero counts even if profile is
26142 * profile-count.h: Fix typo.
26144 2017-08-08 Julia Koval <julia.koval@intel.com>
26146 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
26147 _mm512_mask_cvtsepi16_storeu_epi8,
26148 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
26149 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
26150 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
26151 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
26152 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
26153 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
26154 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
26155 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
26156 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
26157 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
26158 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
26159 __builtin_ia32_pmovuswb256mem_mask,
26160 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
26161 __builtin_ia32_pmovwb512mem_mask): New builtins.
26163 2017-08-08 Julia Koval <julia.koval@intel.com>
26165 PR target/73350,80862
26166 * config/i386/subst.md (round): Fix round pattern.
26167 * config/i386/i386.c (ix86_erase_embedded_rounding):
26168 Fix erasing rounding for the fixed pattern.
26170 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
26172 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
26174 2017-06-08 Martin Liska <mliska@suse.cz>
26176 PR gcov-profile/80911
26177 * gcov.c (block_info::block_info): New constructor.
26179 2017-06-07 Carl Love <cel@us.ibm.com>
26181 * config/rs6000/rs6000-c: The return type of the following
26182 built-in functions was implemented as int not long long. Fix sign
26183 of return value for the unsigned version of vec_mulo and vec_mule.
26184 vector unsigned long long vec_bperm (vector unsigned long long,
26185 vector unsigned char)
26186 vector signed long long vec_mule (vector signed int,
26188 vector unsigned long long vec_mule (vector unsigned int,
26189 vector unsigned int)
26190 vector signed long long vec_mulo (vector signed int,
26192 vector unsigned long long vec_mulo (vector unsigned int,
26193 vector unsigned int)
26194 * doc/extend.texi: Fix the documentation for the built-in
26197 2017-06-07 Carl Love <cel@us.ibm.com>
26200 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
26203 2017-06-07 Carl Love <cel@us.ibm.com>
26205 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
26206 support, Generate doublehv for signed int/float for BE case only.
26208 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
26210 * doc/invoke.texi (mcx16): Rewrite.
26212 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26214 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
26215 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
26216 *mov<mode>_softfloat, and an anonymous splitter): Use
26217 nonimmediate_operand instead of rs6000_nonimmediate_operand.
26219 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26221 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
26223 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26224 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
26225 (rs6000_debug_reg_global): Adjust.
26226 (rs6000_init_hard_regno_mode_ok): Adjust.
26227 (rs6000_dbx_register_number): Adjust.
26228 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
26229 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26230 Remove SPE_ACC and SPEFSCR.
26231 (REG_ALLOC_ORDER): Ditto.
26232 (FRAME_POINTER_REGNUM): Change to 111.
26233 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
26234 (REG_CLASS_NAMES): Ditto.
26235 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
26236 (REGISTER_NAMES): Ditto.
26237 (ADDITIONAL_REG_NAMES): Ditto.
26238 (rs6000_reg_names): Ditto.
26239 * config/rs6000/rs6000.md: Renumber some register number
26242 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26244 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
26246 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
26247 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
26249 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
26250 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
26251 Delete the SPE high registers.
26252 (REG_ALLOC_ORDER): Ditto.
26253 (enum reg_class): Remove SPE_HIGH_REGS.
26254 (REG_CLASS_NAMES): Ditto.
26255 (REG_CLASS_CONTENTS): Delete the SPE high registers.
26256 (REGISTER_NAMES): Ditto.
26257 (rs6000_reg_names): Ditto.
26258 * doc/tm.texi.in: Remove SPE as example.
26259 * doc/tm.texi: Regenerate.
26261 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26263 * config/rs6000/8540.md (ppc8540_brinc): Delete.
26264 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
26265 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
26266 * config/rs6000/rs6000.md (type): Remove "brinc".
26268 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26270 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
26271 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
26272 * config/rs6000/linuxspe.h: Delete file.
26273 * config/rs6000/rs6000.md: Don't include spe.md.
26274 * config/rs6000/spe.h: Delete file.
26275 * config/rs6000/spe.md: Delete file.
26276 * config/rs6000/t-rs6000: Remove spe.md.
26278 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26280 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
26281 (reg_or_none500mem_operand): Delete.
26282 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
26283 instead of reg_or_none500mem_operand.
26285 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26287 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26288 handling of SPE flags.
26289 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
26291 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26293 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
26295 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
26296 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
26297 paired_divv2sf3): Similar.
26298 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
26299 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
26300 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
26302 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
26303 Rename the paired_* instruction patterns.
26304 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
26306 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
26307 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
26308 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
26309 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
26310 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
26311 PAIRED_VECTOR_MODE.
26312 (struct machine_function): Delete field spe_insn_chain_scanned_p.
26313 (spe_func_has_64bit_regs_p): Delete.
26314 (spe_expand_predicate_builtin): Delete.
26315 (spe_expand_evsel_builtin): Delete.
26316 (TARGET_DWARF_REGISTER_SPAN): Do not define.
26317 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
26318 (invalid_e500_subreg): Delete.
26319 (rs6000_legitimize_address): Always force_reg op2 as well, for
26320 paired single memory accesses.
26321 (rs6000_member_type_forces_blk): Delete.
26322 (rs6000_spe_function_arg): Delete.
26323 (rs6000_expand_unop_builtin): Delete SPE handling.
26324 (rs6000_expand_binop_builtin): Ditto.
26325 (spe_expand_stv_builtin): Delete.
26326 (bdesc_2arg_spe): Delete.
26327 (spe_expand_builtin): Delete.
26328 (spe_expand_predicate_builtin): Delete.
26329 (spe_expand_evsel_builtin): Delete.
26330 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
26331 (spe_init_builtins): Delete.
26332 (spe_func_has_64bit_regs_p): Delete.
26333 (savres_routine_name): Delete "info" parameter. Adjust callers.
26334 (rs6000_emit_stack_reset): Ditto.
26335 (rs6000_dwarf_register_span): Delete.
26336 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
26337 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
26338 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
26340 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
26342 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
26343 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
26344 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
26345 mulv2sf3, divv2sf3): Delete expanders.
26347 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26349 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
26351 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26353 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
26354 * config/rs6000/rs6000.c: Ditto.
26356 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26358 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
26359 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
26360 comparison_operator.
26362 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26364 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
26365 * config/rs6000/rs6000.opt: Ditto.
26366 * config/rs6000/t-rtems: Ditto.
26368 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26370 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
26371 TARGET_E500_SINGLE by 0, simplify.
26372 * config/rs6000/rs6000.c: Ditto.
26373 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
26374 (spe_build_register_parallel): Delete.
26375 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
26376 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
26377 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
26378 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
26379 (E500_CONVERT): Delete.
26380 * config/rs6000/spe.md: Remove many patterns and all define_constants.
26382 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
26384 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
26385 * config/rs6000/dfp.md: Ditto.
26386 (negdd2, *negdd2_fpr): Merge.
26387 (absdd2, *absdd2_fpr): Merge.
26388 (negtd2, *negtd2_fpr): Merge.
26389 (abstd2, *abstd2_fpr): Merge.
26390 * config/rs6000/e500.h: Delete file.
26391 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
26392 TARGET_FPRS by 1 and simplify.
26393 * config/rs6000/rs6000-c.c: Ditto.
26394 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
26395 TARGET_DF_SPE by 0.
26396 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
26398 * config/rs6000/rs6000.md: Ditto.
26399 (floatdidf2, *floatdidf2_fpr): Merge.
26400 (move_from_CR_gt_bit): Delete.
26401 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
26402 (E500_CR_IOR_COMPARE): Delete.
26403 (All patterns that require !TARGET_FPRS): Delete.
26404 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
26406 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26408 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
26410 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26412 * graphds.c (add_edge): Intitialize edge's attached data.
26413 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
26414 pointer parameter. Call pointed function on each edge during
26415 graph traversing. Skip traversing the edge when the function
26417 (graphds_dfs, graphds_scc): Ditto.
26418 (for_each_edge): New parameter. Pass the new parameter to callback
26420 * graphds.h (skip_edge_callback): New function pointer type.
26421 (graphds_dfs, graphds_scc): New function pointer parameter.
26422 (graphds_edge_callback, for_each_edge): New parameter.
26424 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26426 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
26427 out code checking if runtime alias check is possible to below ...
26428 Call the new function.
26429 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
26430 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
26432 2017-06-07 Marek Polacek <polacek@redhat.com>
26435 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
26436 TYPE_OVERFLOW_WRAPS check.
26438 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26440 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
26441 if versioning is required.
26442 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
26443 peeling with the check for versioning.
26445 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26447 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
26448 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
26449 Set true to new parameter if new ssa variable is defined.
26450 (vect_gen_vector_loop_niters): Refactor. Set range information
26451 for the new vector loop bound variable.
26452 (vect_do_peeling): Ditto.
26454 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26456 * tree-affine.c (ssa.h): Include header file.
26457 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
26458 has wrapping overflow behavior.
26460 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26462 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
26464 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26466 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
26467 (tree_to_aff_combination): ... here.
26469 2017-06-07 Bin Cheng <bin.cheng@arm.com>
26471 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
26472 reg_pressure model function.
26473 (ivopts_global_cost_for_size): Delete.
26474 (determine_set_costs, iv_ca_recount_cost): Call new model function
26475 ivopts_estimate_reg_pressure.
26477 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26479 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
26480 expensive than udiv. Remove floating point cases from mod.
26482 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26484 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
26485 Increase idiv cost.
26487 2017-06-07 Tamar Christina <tamar.christina@arm.com>
26489 * config/aarch64/aarch64.md
26490 (copysignsf3): Fix mask generation.
26492 2017-06-07 Jakub Jelinek <jakub@redhat.com>
26494 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
26496 (class dump_manager): Add register_dumps method.
26497 * dumpfile.c: Include langhooks.h.
26498 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
26499 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
26500 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
26501 (dump_manager::dump_register): Start with 512 entries instead of 32.
26502 (dump_manager::register_dumps): New method.
26503 * toplev.c (general_init): Instead of invoking register_dumps
26504 langhook, invoke register_dumps method on the dump manager.
26505 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
26508 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
26510 * doc/md.texi: Clarify the restrictions on a define_insn condition.
26511 Say that # requires an associated define_split to exist, and that
26512 the define_split must be suitable for use after register allocation.
26514 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26516 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
26517 (compute_outgoing_frequencies): Also initialize zero counts.
26518 (find_many_sub_basic_blocks): Do not produce uninitialized profile
26519 around loops; preserve more of profile when nothing changes.
26521 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
26523 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
26525 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
26526 * config/arm/arm-cpu-cdata.h: Regenerate.
26527 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
26528 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
26529 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
26530 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
26531 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
26533 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
26535 * config/arm/t-rmprofile: Likewise.
26536 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
26538 2017-06-06 David S. Miller <davem@davemloft.net>
26541 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
26542 blockage if function uses alloca.
26544 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26546 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
26547 New "uid" fields to hold pretty-print IDs of group and ref.
26548 Memory references are now identified as <group_id>:<ref_id>
26549 instead of using [random] addresses.
26550 (dump_mem_details): Simplify, no functional change.
26551 (dump_mem_ref): Simplify and make output more concise.
26552 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
26553 (find_or_create_group): Initialize group uid.
26554 (record_ref): Initialize ref uid. Improve debug output.
26555 (prune_group_by_reuse, should_issue_prefetch_p,)
26556 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
26557 (mark_nontemporal_store, determine_loop_nest_reuse):
26558 Improve debug output.
26560 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
26562 * dbgcnt.def (prefetch): New debug counter.
26563 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
26564 (schedule_prefetches): Stop issueing prefetches if debug counter
26567 2017-06-06 Tom de Vries <tom@codesourcery.com>
26569 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
26570 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
26572 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26574 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
26575 Use aarch64_reg_or_zero predicate for operand 4.
26576 (aarch64_compare_and_swap<mode> define_insn_and_split):
26577 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
26578 (aarch64_store_exclusive<mode>): Likewise for operand 2.
26580 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
26582 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
26583 (arm_compute_save_core_reg_mask): This.
26584 (thumb1_compute_save_reg_mask): Rename into ...
26585 (thumb1_compute_save_core_reg_mask): This.
26586 (arm_compute_save_reg0_reg12_mask): Adapt comment.
26587 (arm_compute_frame_layout): Likewise.
26589 2017-06-06 Richard Biener <rguenther@suse.de>
26591 PR tree-optimization/80974
26592 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
26593 keep or clear leaders SSA info.
26595 2017-06-06 Tom de Vries <tom@codesourcery.com>
26597 * config/nvptx/nvptx.c (split_mode_p): New function.
26598 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
26600 2017-06-06 Tom de Vries <tom@codesourcery.com>
26602 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
26604 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26607 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
26610 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26612 * shrink-wrap.c (handle_simple_exit): Update profile.
26613 (try_shrink_wrapping): Upate profile.
26615 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26617 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
26618 (tree_guess_outgoing_edge_probabilities): New.
26619 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
26620 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
26622 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26624 * ipa-split.c (split_function): Initialize return bb profile.
26626 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
26628 * profile.c (compute_branch_probabilities): Also initialize
26629 EXIT_BLOCK profile.
26631 2017-06-06 Richard Biener <rguenther@suse.de>
26633 PR tree-optimization/80928
26634 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
26635 (vect_analyze_loop_operations): Properly guard analysis for
26637 (vect_transform_loop): Likewise.
26638 (vect_analyze_loop_2): Also reset SLP type on PHIs.
26639 (vect_model_induction_cost): Do not cost for pure SLP.
26640 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
26641 of induction in inner loop vectorization.
26642 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
26643 (vect_get_and_check_slp_defs): Handle vect_induction_def.
26644 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
26646 (vect_analyze_slp_cost_1): Cost induction.
26647 (vect_detect_hybrid_slp_stmts): Handle PHIs.
26648 (vect_get_slp_vect_defs): Likewise.
26649 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
26650 (vect_transform_stmt): Handle SLP reductions.
26651 * tree-vectorizer.h (vectorizable_induction): Adjust.
26653 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26655 * config/rs6000/rs6000.c (make_resolver_func): Update
26656 init_lowered_empty_function call.
26658 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
26660 * doc/invoke.texi: Document the -fprofile-abs-path option.
26661 * common.opt (fprofile-abs-path): New option.
26662 * gcov-io.h (gcov_write_filename): Declare.
26663 * gcov-io.c (gcov_write_filename): New function.
26664 * coverage.c (coverage_begin_function): Use gcov_write_filename.
26665 * profile.c (output_location): Likewise.
26667 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26669 * shring-wrap.c: Revert accidental commit.
26671 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
26673 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
26675 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26677 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
26679 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
26680 profile in callgraph edge.
26681 * profile-count.h (apply_probability): If THIS is 0, then result is 0
26682 (apply_scale): Likewise.
26683 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
26684 Also scale profile when inlining function with zero profile.
26685 (initialize_cfun): Update exit block profile even when it is zero.
26686 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
26687 when profile is read.
26689 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
26691 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
26692 (CLONE_*): New constants to define the processors we can generate
26693 code for with the target_clone attribute.
26694 (rs6000_clone_map): New array to identify which clone processors
26695 the current program is running on.
26696 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
26697 target_clone attribute.
26698 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
26699 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
26700 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
26701 (cpu_expand_builtin): Add support for target_clone attribute.
26702 (rs6000_valid_attribute_p): Allow "default" attribute.
26703 (get_decl_name): New debug function to simplify printing the
26704 current function name in debugging statements.
26705 (rs6000_clone_priority): New functions to support the target_clone
26706 attribute, and be able to generate code to switch between ISA 2.05
26707 through ISA 3.0 (power6 through power9).
26708 (rs6000_compare_version_priority): Likewise.
26709 (rs6000_get_function_versions_dispatcher): Likewise.
26710 (make_resolver_func): Likewise.
26711 (add_condition_to_bb): Likewise.
26712 (dispatch_function_versions): Likewise.
26713 (rs6000_generate_version_dispatcher_body): Likewise.
26714 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
26715 (fusion_gpr_load_p): Fix a spacing issue.
26716 * doc/extend.texi (Common Function Attributes): Document that the
26717 PowerPC supports the target_clone attribute.
26719 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
26721 * config/arm/arm.h: explain F symbol found in description of ARM
26722 register allocation in its legend.
26724 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26726 * config/mips/frame-header-opt.c: Include profile-count.h.
26727 * config/riscv/riscv.c: Include profile-count.h
26729 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
26731 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
26733 (sm_set_flag_if_changed): Add bbs field.
26734 (execute_sm_if_changed_flag_set): Pass BBS.
26735 (execute_sm): Update.
26737 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26739 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
26742 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26744 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
26745 (peephole2): New peephole2 to emit the above.
26746 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
26748 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26750 * config/aarch64/aarch64.c (define_peephole2 above
26751 *sub_<shift>_<mode>): New peephole.
26753 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
26755 * config/i386/i386.c (make_resolver_func): Update.
26756 * Makefile.in: Add profile-count.h and profile-count.o
26757 * auto-profile.c (afdo_indirect_call): Update to new API.
26758 (afdo_set_bb_count): Update.
26759 (afdo_propagate_edge): Update.
26760 (afdo_propagate_circuit): Update.
26761 (afdo_calculate_branch_prob): Update.
26762 (afdo_annotate_cfg): Update.
26763 * basic-block.h: Include profile-count.h
26764 (struct edge_def): Turn count to profile_count.
26765 (struct basic_block_def): Likewie.
26766 (REG_BR_PROB_BASE): Move to profile-count.h
26767 (RDIV): Move to profile-count.h
26768 * bb-reorder.c (max_entry_count): Turn to profile_count.
26769 (find_traces): Update.
26770 (rotate_loop):Update.
26771 (connect_traces):Update.
26772 (sanitize_hot_paths):Update.
26773 * bt-load.c (migrate_btr_defs): Update.
26774 * cfg.c (RDIV): Remove.
26775 (init_flow): Use alloc_block.
26776 (alloc_block): Uninitialize count.
26777 (unchecked_make_edge): Uninitialize count.
26778 (check_bb_profile): Update.
26779 (dump_edge_info): Update.
26780 (dump_bb_info): Update.
26781 (update_bb_profile_for_threading): Update.
26782 (scale_bbs_frequencies_int): Update.
26783 (scale_bbs_frequencies_gcov_type): Update.
26784 (scale_bbs_frequencies_profile_count): New.
26785 * cfg.h (update_bb_profile_for_threading): Update.
26786 (scale_bbs_frequencies_profile_count): Declare.
26787 * cfgbuild.c (compute_outgoing_frequencies): Update.
26788 (find_many_sub_basic_blocks): Update.
26789 * cfgcleanup.c (try_forward_edges): Update.
26790 (try_crossjump_to_edge): Update.
26791 * cfgexpand.c (expand_gimple_tailcall): Update.
26792 (construct_exit_block): Update.
26793 * cfghooks.c (verify_flow_info): Update.
26794 (dump_bb_for_graph): Update.
26795 (split_edge): Update.
26796 (make_forwarder_block): Update.
26797 (duplicate_block): Update.
26798 (account_profile_record): Update.
26799 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
26800 (get_estimated_loop_iterations): Update.
26801 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
26802 (single_likely_exit): Update.
26803 * cfgloopmanip.c (scale_loop_profile): Update.
26805 (set_zero_probability): Update.
26806 (lv_adjust_loop_entry_edge): Update.
26807 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
26808 (purge_dead_edges): Update.
26809 (rtl_account_profile_record): Update.
26810 * cgraph.c (cgraph_node::create): Uninitialize count.
26811 (symbol_table::create_edge): Uninitialize count.
26812 (cgraph_update_edges_for_call_stmt_node): Update.
26813 (cgraph_edge::dump_edge_flags): Update.
26814 (cgraph_node::dump): Update.
26815 (cgraph_edge::maybe_hot_p): Update.
26816 * cgraph.h: Include profile-count.h
26817 (create_clone), create_edge, create_indirect_edge): Update.
26818 (cgraph_node): Turn count to profile_count.
26819 (cgraph_edge0: Likewise.
26820 (make_speculative, clone): Update.
26821 (create_edge): Update.
26822 (init_lowered_empty_function): Update.
26823 * cgraphclones.c (cgraph_edge::clone): Update.
26824 (duplicate_thunk_for_node): Update.
26825 (cgraph_node::create_clone): Update.
26826 * cgraphunit.c (cgraph_node::analyze): Update.
26827 (cgraph_node::expand_thunk): Update.
26828 * final.c (dump_basic_block_info): Update.
26829 * gimple-streamer-in.c (input_bb): Update.
26830 * gimple-streamer-out.c (output_bb): Update.
26831 * graphite.c (print_global_statistics): Update.
26832 (print_graphite_scop_statistics): Update.
26833 * hsa-brig.c: Include basic-block.h.
26834 * hsa-dump.c: Include basic-block.h.
26835 * hsa-gen.c (T sum_slice): Update.
26836 (convert_switch_statements):Update.
26837 * hsa-regalloc.c: Include basic-block.h.
26838 * ipa-chkp.c (chkp_produce_thunks): Update.
26839 * ipa-cp.c (struct caller_statistics): Update.
26840 (init_caller_stats): Update.
26841 (gather_caller_stats): Update.
26842 (ipcp_cloning_candidate_p): Update.
26843 (good_cloning_opportunity_p): Update.
26844 (get_info_about_necessary_edges): Update.
26845 (dump_profile_updates): Update.
26846 (update_profiling_info): Update.
26847 (update_specialized_profile): Update.
26848 (perhaps_add_new_callers): Update.
26849 (decide_about_value): Update.
26850 (ipa_cp_c_finalize): Update.
26851 * ipa-devirt.c (struct odr_type_warn_count): Update.
26852 (struct decl_warn_count): Update.
26853 (struct final_warning_record): Update.
26854 (possible_polymorphic_call_targets): Update.
26855 (ipa_devirt): Update.
26856 * ipa-fnsummary.c (redirect_to_unreachable): Update.
26857 * ipa-icf.c (sem_function::merge): Update.
26858 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
26859 * ipa-inline.c (compute_uninlined_call_time): Update.
26860 (compute_inlined_call_time): Update.
26861 (want_inline_small_function_p): Update.
26862 (want_inline_self_recursive_call_p): Update.
26863 (edge_badness): Update.
26864 (lookup_recursive_calls): Update.
26865 (recursive_inlining): Update.
26866 (inline_small_functions): Update.
26867 (dump_overall_stats): Update.
26868 (dump_inline_stats): Update.
26869 * ipa-profile.c (ipa_profile_generate_summary): Update.
26870 (ipa_propagate_frequency): Update.
26871 (ipa_profile): Update.
26872 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
26873 * ipa-utils.c (ipa_merge_profiles): Update.
26874 * loop-doloop.c (doloop_modify): Update.
26875 * loop-unroll.c (report_unroll): Update.
26876 (unroll_loop_runtime_iterations): Update.
26877 * lto-cgraph.c (lto_output_edge): Update.
26878 (lto_output_node): Update.
26879 (input_node): Update.
26880 (input_edge): Update.
26881 (merge_profile_summaries): Update.
26882 * lto-streamer-in.c (input_cfg): Update.
26883 * lto-streamer-out.c (output_cfg): Update.
26884 * mcf.c (create_fixup_graph): Update.
26885 (adjust_cfg_counts): Update.
26886 (sum_edge_counts): Update.
26887 * modulo-sched.c (sms_schedule): Update.
26888 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
26889 * predict.c (maybe_hot_count_p): Update.
26890 (probably_never_executed): Update.
26891 (dump_prediction): Update.
26892 (combine_predictions_for_bb): Update.
26893 (propagate_freq): Update.
26894 (handle_missing_profiles): Update.
26895 (counts_to_freqs): Update.
26896 (rebuild_frequencies): Update.
26897 (force_edge_cold): Update.
26898 * predict.h: Include profile-count.h
26899 (maybe_hot_count_p, counts_to_freqs): UPdate.
26900 * print-rtl-function.c: Do not include cfg.h
26901 * print-rtl.c: Include basic-block.h
26902 * profile-count.c: New file.
26903 * profile-count.h: New file.
26904 * profile.c (is_edge_inconsistent): Update.
26905 (correct_negative_edge_counts): Update.
26906 (is_inconsistent): Update.
26907 (set_bb_counts): Update.
26908 (read_profile_edge_counts): Update.
26909 (compute_frequency_overlap): Update.
26910 (compute_branch_probabilities): Update; Initialize and deinitialize
26912 (branch_prob): Update.
26913 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
26914 (edge_gcov_count): New.
26915 (bb_gcov_count): New.
26916 * shrink-wrap.c (try_shrink_wrapping): Update.
26917 * tracer.c (better_p): Update.
26918 * trans-mem.c (expand_transaction): Update.
26919 (ipa_tm_insert_irr_call): Update.
26920 (ipa_tm_insert_gettmclone_call): Update.
26921 * tree-call-cdce.c: Update.
26922 * tree-cfg.c (gimple_duplicate_sese_region): Update.
26923 (gimple_duplicate_sese_tail): Update.
26924 (gimple_account_profile_record): Update.
26925 (execute_fixup_cfg): Update.
26926 * tree-inline.c (copy_bb): Update.
26927 (copy_edges_for_bb): Update.
26928 (initialize_cfun): Update.
26929 (freqs_to_counts): Update.
26930 (copy_cfg_body): Update.
26931 (expand_call_inline): Update.
26932 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26933 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
26934 (try_unroll_loop_completely): Update.
26935 (try_peel_loop): Update.
26936 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26937 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
26938 * tree-ssa-loop-split.c (connect_loops): Update.
26939 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
26940 * tree-ssa-reassoc.c (branch_fixup): Update.
26941 * tree-ssa-tail-merge.c (replace_block_by): Update.
26942 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
26943 (compute_path_counts): Update.
26944 (update_profile): Update.
26945 (recompute_probabilities): Update.
26946 (update_joiner_offpath_counts): Update.
26947 (estimated_freqs_path): Update.
26948 (freqs_to_counts_path): Update.
26949 (clear_counts_path): Update.
26950 (ssa_fix_duplicate_block_edges): Update.
26951 (duplicate_thread_path): Update.
26952 * tree-switch-conversion.c (case_bit_test_cmp): Update.
26953 (struct switch_conv_info): Update.
26954 * tree-tailcall.c (decrease_profile): Update.
26955 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26956 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26957 * value-prof.c (check_counter): Update.
26958 (gimple_divmod_fixed_value): Update.
26959 (gimple_mod_pow2): Update.
26960 (gimple_mod_subtract): Update.
26961 (gimple_ic_transform): Update.
26962 (gimple_stringop_fixed_value): Update.
26963 * value-prof.h (gimple_ic): Update.
26965 2017-06-02 Carl Love <cel@us.ibm.com>
26967 * config/rs6000/rs6000-c: Add support for built-in functions
26968 vector double vec_doublee (vector signed int);
26969 vector double vec_doublee (vector unsigned int);
26970 vector double vec_doublee (vector float);
26971 vector double vec_doubleh (vector signed int);
26972 vector double vec_doubleh (vector unsigned int);
26973 vector double vec_doubleh (vector float);
26974 vector double vec_doublel (vector signed int);
26975 vector double vec_doublel (vector unsigned int);
26976 vector double vec_doublel (vector float);
26977 vector double vec_doubleo (vector signed int);
26978 vector double vec_doubleo (vector unsigned int);
26979 vector double vec_doubleo (vector float);.
26980 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
26981 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
26983 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
26984 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
26985 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
26987 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
26988 vec_doublel, vec_doubleh.
26989 * doc/extend.texi: Update the built-in documentation file for the
26990 new built-in functions.
26992 2017-06-02 David Malcolm <dmalcolm@redhat.com>
26995 * ipa-inline-analysis.c (free_growth_caches): Set
26996 edge_removal_hook_holder to NULL after removing it.
26998 2017-06-02 Sudakshina Das <sudi.das@arm.com>
27000 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
27001 comparision with zero.
27003 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27004 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27005 for early expansion of vec_min and vec_max builtins.
27006 (builtin_function_type): Add min/max unsigned variants to those
27007 identified as having unsigned arguments.
27009 2017-06-02 Olivier Hainque <hainque@adacore.com>
27011 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
27013 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27015 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
27016 Use VALL_F16 iterator rather than VALL.
27018 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27020 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
27021 Emit CBNZ inside loop when doing a strong exchange and comparing
27022 against zero. Generate the CC flags after the loop.
27024 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
27026 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
27027 (dl_section_ref): New.
27028 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
27029 On AIX, append an expression to subtract the size of the
27030 section length to dl_section_ref.
27032 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
27034 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
27035 for early expansion of vector absolute builtins.
27037 2017-06-02 Richard Biener <rguenther@suse.de>
27039 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
27040 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
27042 2017-06-02 Richard Biener <rguenther@suse.de>
27044 PR tree-optimization/80948
27045 * tree-tailcall.c (find_tail_calls): Track stmts to move in
27046 stmt order as well.
27048 2017-06-02 Richard Biener <rguenther@suse.de>
27050 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
27052 * tree-vect-stmts.c (process_use): Do not mark backedge defs
27053 for inductions as relevant.
27055 2017-06-02 Richard Biener <rguenther@suse.de>
27057 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
27058 (vectorizable_induction): ... this. Remove dead code.
27060 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
27062 * builtins. (expand_builtin_alloca): Remove second parameter and
27063 infer its value from the first parameter instead.
27064 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
27066 2017-06-02 Jakub Jelinek <jakub@redhat.com>
27068 PR rtl-optimization/80903
27069 * loop-doloop.c (add_test): Unshare sequence.
27071 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
27073 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
27075 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
27077 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
27079 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
27080 xlogue_layout::get_instance, logue_layout::xlogue_layout,
27081 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
27082 (xlogue_layout::get_stub_rtx): Make static.
27083 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
27084 (xlogue_layout::compute_stub_managed_regs): Rename to...
27085 (xlogue_layout::count_stub_managed_regs): ...this.
27086 (xlogue_layout::is_stub_managed_reg): New function.
27087 (xlogue_layout::m_stub_names): Rename to...
27088 (xlogue_layout::s_stub_names): ...this, make static.
27089 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
27090 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
27091 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
27092 xlogue_layout::s_stub_names): Instantiate statics.
27093 (stub_managed_regs): Remove.
27094 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
27095 (disable_call_ms2sysv_xlogues): Rename to...
27096 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
27097 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
27099 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
27100 change after reload_completed.
27101 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
27103 (ix86_expand_prologue): Likewise.
27104 (ix86_expand_epilogue): Likewise.
27105 (ix86_expand_split_stack_prologue): Likewise.
27106 (ix86_compute_frame_layout): Remove frame parameter ...
27107 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
27108 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
27110 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
27111 (ix86_frame): Move from here ...
27112 * config/i386/i386.h (ix86_frame): ... to here.
27113 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
27114 complete ix86_frame data structure instead. Remove some_ld_name.
27116 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27118 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
27119 symbols that hold a DECL_VALUE_EXPR.
27121 2017-06-01 Martin Jambor <mjambor@suse.cz>
27123 PR tree-optimization/80898
27124 * tree-sra.c (process_subtree_disqualification): Removed.
27125 (disqualify_candidate): Do not acll
27126 process_subtree_disqualification.
27127 (subtree_mark_written_and_enqueue): New function.
27128 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
27129 RHS has been disqualified and re-queue LHS if necessary. Apart
27130 from that, ignore disqualified RHS.
27132 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27134 * config/s390/s390.c (s390_emit_epilogue): Disable early return
27135 address fetch for z10 or later.
27137 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27139 * config/arc/arc.md (tst_movb): Add guard when splitting.
27141 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27143 * config/arc/arc.c (arc_can_eliminate): Test against
27144 arc_frame_pointer_needed.
27146 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27148 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
27149 to prevent store reordering.
27150 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
27151 (type): Add block type.
27152 (stack_tie): Define special instruction to be used in
27155 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27157 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
27158 constraint. It is not valid for the pattern.
27159 (noncommutative_binary_comparison): Likewise.
27161 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27163 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
27166 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27168 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
27169 be used by the reg-alloc.
27171 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27173 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
27174 reg-alloc when having mul64 or mul32x16 instructions.
27175 (mulsidi3): Likewise.
27176 (umulsidi3): Likewise.
27177 (mulsi32x16): New pattern.
27178 (mulsi64): Likewise.
27179 (mulsidi64): Likewise.
27180 (umulsidi64): Likewise.
27181 (MUL32x16_REG): Define.
27182 (mul64_600): Use MUL32x16_REG.
27183 (mac64_600): Likewise.
27184 (umul64_600): Likewise.
27185 (umac64_600): Likewise.
27187 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
27189 * config/arc/arc.md (mulsi3_700): Make it commutative.
27191 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
27193 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
27195 (*sign_extendsidi2_insn): Likewise for movstosw.
27197 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
27199 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
27200 the type of the input discriminant value. Convert the
27201 discriminant value of signedness vary.
27203 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
27205 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
27206 Add to -Wall section.
27208 2017-06-01 Richard Biener <rguenther@suse.de>
27210 PR middle-end/66313
27211 * fold-const.c (fold_plusminus_mult_expr): If the factored
27212 factor may be zero use a wrapping type for the inner operation.
27213 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
27214 and handle moved defs.
27215 (process_assignment): Properly guard the unary op case. Return a
27216 tri-state indicating that moving the stmt before the call may allow
27217 to continue. Pass through to_move.
27218 (find_tail_calls): Handle moving unrelated defs before
27221 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
27224 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
27225 splitter result in the canonical way.
27227 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
27229 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
27230 also for 32bit target. Update insn attributes.
27231 (zero-extendsidi2 splitter): Allow all registers for operand 1.
27233 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
27235 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
27236 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
27237 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
27238 (_mm_maskz_min_ss): New intrinsics.
27240 2017-05-31 Martin Liska <mliska@suse.cz>
27242 * tree-vect-loop.c (vect_create_epilog_for_reduction):
27243 Change comment style to one we normally use.
27244 (vectorizable_reduction): Likewise.
27245 (vectorizable_induction): Likewise.
27246 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
27247 (vectorizable_call): Likewise.
27248 (vectorizable_simd_clone_call): Likewise.
27249 (vectorizable_conversion): Likewise.
27250 (vectorizable_assignment): Likewise.
27251 (vectorizable_shift): Likewise.
27252 (vectorizable_operation): Likewise.
27253 (vectorizable_store): Likewise.
27254 (vectorizable_load): Likewise.
27255 * tree-vectorizer.h: Likewise.
27257 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
27259 * passes.c (emergency_dump_function): New.
27260 * tree-pass.h (emergency_dump_function): Declare.
27261 * plugin.c (plugins_internal_error_function): Remove.
27262 * plugin.h (plugins_internal_error_function): Remove declaration.
27263 * toplev.c (internal_error_function): New static function. Use it...
27264 (general_init): ...here.
27266 2017-05-31 Graham Markall <graham.markall@embecosm.com>
27268 * config/arc/arc.c (arc_print_operand): Handle constant operands.
27269 (arc_rtx_costs): Add costs for new patterns.
27270 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
27271 * config/arc/predicates.md: Add _1_2_3_operand predicate.
27273 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27275 * tree-ssa-strlen.c (get_next_strinfo): New function.
27276 (get_stridx_plus_constant): Use it.
27277 (zero_length_string): Likewise.
27278 (adjust_related_strinfos): Likewise.
27279 (adjust_last_stmt): Likewise.
27281 2017-05-31 Richard Biener <rguenther@suse.de>
27284 * config/i386/i386.c (ix86_expand_builtin): Remove assert
27285 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
27287 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
27289 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
27290 loop_vinfo argument and use of dependence distance vectors.
27291 Check instead whether the two references differ only in their
27292 initial value and assume that they have the same alignment if the
27293 difference is a multiple of the vector alignment.
27294 (vect_analyze_data_refs_alignment): Update call accordingly.
27296 2017-05-31 Martin Liska <mliska@suse.cz>
27299 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
27301 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27303 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
27304 (create_intersect_range_checks): Move from ...
27305 * tree-data-ref.c (create_intersect_range_checks_index)
27306 (create_intersect_range_checks): ... to here.
27307 (create_runtime_alias_checks): New function factored from ...
27308 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
27309 here. Call above function.
27310 * tree-data-ref.h (create_runtime_alias_checks): New function.
27312 2017-05-31 Bin Cheng <bin.cheng@arm.com>
27314 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
27315 segment length for dr_b and compute it in wide_int.
27317 2017-05-31 Richard Biener <rguenther@suse.de>
27319 PR tree-optimization/80906
27320 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
27321 and pass through iv_map.
27322 (copy_bb_and_scalar_dependences): Adjust.
27323 (translate_pending_phi_nodes): Likewise.
27324 (copy_loop_close_phi_args): Handle code-generating IVs instead
27327 2017-05-30 David Malcolm <dmalcolm@redhat.com>
27329 * diagnostic-color.c (color_dict): Add "type-diff".
27330 (parse_gcc_colors): Update comment.
27331 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
27332 -fdiagnostics-show-template-tree and -fno-elide-type.
27333 (GCC_COLORS): Add type-diff to example.
27335 (-fdiagnostics-show-template-tree): New.
27336 (-fno-elide-type): New.
27337 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
27338 the pp_format_decoder callback. Call any m_format_postprocessor's
27340 (pretty_printer::pretty_printer): Initialize
27341 m_format_postprocessor.
27342 (pretty_printer::~pretty_printer): Delete any
27343 m_format_postprocessor.
27344 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
27345 (class format_postprocessor): New class.
27346 (struct pretty_printer::format_decoder): Document the new parameters.
27347 (struct pretty_printer::m_format_postprocessor): New field.
27348 * tree-diagnostic.c (default_tree_printer): Update for new
27349 bool and const char ** params.
27350 * tree-diagnostic.h (default_tree_printer): Likewise.
27352 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
27354 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
27355 (lwa_operand): Delete rs6000_gen_cell_microcode test.
27356 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
27357 rs6000_gen_cell_microcode code.
27358 (rs6000_final_prescan_insn): Delete.
27359 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
27360 "warn-cell-microcode" entries.
27361 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
27362 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
27363 throughout. Change cc_reg_not_micro_cr0_operand to
27364 cc_reg_not_cr0_operand throughout.
27365 (*extendhi<mode>2_noload): Delete.
27366 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
27367 (mwarn-cell-microcode): Delete.
27368 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
27369 -mgen-cell-microcode and -mwarn-cell-microcode.
27371 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
27374 * config/i386/constraints.md (Yd): New constraint.
27376 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
27377 and (?Yd, r) alternatives. Update insn attributes.
27378 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
27379 and (?*Yd, r) alternatives. Update insn attributes.
27380 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
27382 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
27384 * gimplify.c (gimplify_modify_expr): Don't create a
27385 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
27388 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
27390 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
27392 2017-05-30 Richard Biener <rguenther@suse.de>
27394 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
27395 and reduc_def fields.
27396 (STMT_VINFO_REDUC_TYPE): New define.
27397 (STMT_VINFO_REDUC_DEF): Likewise.
27398 (vect_force_simple_reduction): Adjust prototype.
27399 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
27400 (vect_is_simple_reduction): Remove check_reduction argument.
27401 (vect_force_simple_reduction): Adjust and set
27402 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27403 (vectorizable_reduction): Do not re-do reduction analysis
27404 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
27405 * tree-parloops.c (gather_scalar_reductions): Adjust.
27407 2017-05-30 Richard Biener <rguenther@suse.de>
27409 PR middle-end/80901
27410 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
27413 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
27415 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
27416 Introduce unknown_misalignment parameter and remove vf.
27417 (vect_peeling_hash_get_lowest_cost):
27418 Pass unknown_misalignment parameter.
27419 (vect_enhance_data_refs_alignment):
27420 Fix unsupportable data ref treatment.
27422 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27424 * tree-vect-data-refs.c (vect_get_data_access_cost):
27425 Workaround for SLP handling.
27426 (vect_enhance_data_refs_alignment):
27427 Compute costs for doing no peeling at all, compare to the best
27428 peeling costs so far and avoid peeling if cheaper.
27430 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27432 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
27433 Return peeling info and set costs to zero for unlimited cost
27435 (vect_enhance_data_refs_alignment): Also inspect all datarefs
27436 with unknown misalignment. Compute and costs for unknown
27437 misalignment, compare them to the costs for known misalignment
27438 and choose the cheapest for peeling.
27440 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27442 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
27443 (vect_get_peeling_costs_all_drs): Create function.
27444 (vect_peeling_hash_get_lowest_cost):
27445 Use vect_get_peeling_costs_all_drs.
27446 (vect_peeling_supportable): Create function.
27447 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
27449 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
27451 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
27452 DR_HAS_NEGATIVE_STEP.
27453 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
27454 (vect_enhance_data_refs_alignment): Use.
27455 (vect_duplicate_ssa_name_ptr_info): Use.
27456 * tree-vectorizer.h (dr_misalignment): Use.
27457 (known_alignment_for_access_p): Use.
27459 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
27462 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
27464 (has_section_name): New function.
27466 2017-05-30 Martin Liska <mliska@suse.cz>
27469 * auto-profile.c (get_function_decl_from_block): Fix
27472 2017-05-30 Richard Biener <rguenther@suse.de>
27474 PR middle-end/80876
27475 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
27477 2017-05-30 Martin Liska <mliska@suse.cz>
27479 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
27480 * dumpfile.h (struct dump_file_info): Remove ctors.
27482 2017-05-30 Martin Liska <mliska@suse.cz>
27484 * predict.def: Fix GNU coding style.
27486 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27488 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
27489 Mark 'to' argument with ATTRIBUTE_UNUSED.
27491 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
27493 * config/xtensa/xtensa.c (xtensa_emit_call): Use
27494 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
27495 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
27498 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
27500 * doc/install.texi (Options specification): Restore entry of
27501 --enable-sjlj-exceptions.
27503 2017-05-27 Michael Eager <eager@eagercon.com>
27506 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
27508 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
27510 * config/microblaze/microblaze.h
27511 (FIXED_REGISTERS): Update in macro.
27512 (CALL_USED_REGISTERS): Update in macro.
27514 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
27516 * doc/install.texi: Add links to macOS binary distributions.
27518 2017-05-27 Jakub Jelinek <jakub@redhat.com>
27522 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27524 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27526 2017-05-26 Martin Liska <mliska@suse.cz>
27528 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
27530 2017-05-26 Martin Liska <mliska@suse.cz>
27532 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
27533 always leading ';; '.
27534 (dump_bb_info): Likewise.
27535 (brief_dump_cfg): Likewise.
27536 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
27537 * dumpfile.c: Remove usage of TDF_VERBOSE.
27538 * dumpfile.h (enum dump_kind): Likewise.
27539 (dump_gimple_bb_header): Do not use TDF_COMMENT.
27540 * print-tree.c (debug_verbose): Remove.
27541 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
27542 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
27543 * tree-diagnostic.c (default_tree_printer): Replace
27544 TDF_DIAGNOSTIC with TDF_SLIM.
27546 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27548 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
27549 in parameter loop, rather than loop_vinfo.
27550 (create_intersect_range_checks): Ditto.
27551 (vect_create_cond_for_alias_checks): Update call to above functions.
27553 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27555 PR tree-optimization/80815
27556 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
27557 for merging runtime alias checks. Handle negative DR_STEPs.
27559 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27561 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
27563 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
27564 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
27565 out code pruning runtime alias checks.
27566 * tree-data-ref.c (prune_runtime_alias_test_list): New function
27567 factored out from above.
27568 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27570 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
27572 (prune_runtime_alias_test_list): New decalaration.
27574 2017-05-26 Bin Cheng <bin.cheng@arm.com>
27576 * tree-vect-data-refs.c (compare_tree): Rename and move ...
27577 * tree-data-ref.c (data_ref_compare_tree): ... to here.
27578 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
27579 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
27580 (operator==, comp_dr_with_seg_len_pair): Ditto.
27581 (vect_prune_runtime_alias_test_list): Ditto.
27583 2017-05-26 Martin Liska <mliska@suse.cz>
27586 * params.def: Bound partial-inlining-entry-probability param.
27588 2017-05-26 Marek Polacek <polacek@redhat.com>
27591 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
27594 2017-05-26 Richard Biener <rguenther@suse.de>
27596 PR tree-optimization/80842
27597 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
27600 2017-05-26 Richard Biener <rguenther@suse.de>
27602 PR tree-optimization/80844
27603 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
27605 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
27607 * doc/md.texi (Machine Constraints): Update x86 family
27608 machine constraints section to match 'config/i386/constraints.md'.
27610 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
27612 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
27614 2017-05-25 Nathan Sidwell <nathan@acm.org>
27616 * doc/invoke.texi (--enable-languages): Update documentation.
27618 2017-05-25 Martin Liska <mliska@suse.cz>
27620 * dumpfile.c: Add TDF_FOLDING.
27621 * dumpfile.h (enum dump_kind): Likewise.
27622 * genmatch.c (dt_simplify::gen_1): Use it.
27624 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27626 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
27628 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27630 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
27631 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
27633 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
27635 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
27636 * match.pd (X == C): Rewrite it here.
27637 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
27638 with_certain_nonzero_bits2): New predicates.
27639 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
27641 2017-05-24 Nathan Sidwell <nathan@acm.org>
27643 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
27646 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
27649 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27651 * config/powerpcspe: New port. Files are copied from the rs6000
27652 port, with "rs6000" in filenames replaced by "powerpcspe".
27654 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
27656 PR rtl-optimization/80754
27657 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
27659 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
27661 * config/sparc/sparc.md (length): Return the correct value for -mflat
27662 sibcalls to match output_sibcall.
27664 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
27668 * config/rs6000/rs6000.c (struct machine_function): Add new field
27670 (rs6000_get_separate_components): Init that field, use it.
27671 (rs6000_components_for_bb): Use the field.
27673 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
27675 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
27677 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
27679 PR middle-end/80823
27680 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
27682 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27685 * config/s390/s390.c (s390_check_qrst_address): Check incoming
27686 address against address_operand predicate.
27687 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
27689 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
27691 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
27692 parameters passed indirectly.
27694 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
27696 * config/i386/i386.md (*movdi_internal): Remove SSE4
27697 alternative 18 (?r, *v). Update insn attributes.
27698 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
27699 Update insn attributes.
27700 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
27701 Update insn attributes.
27702 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
27703 alternative 1 (r, v). Remove isa attribute.
27704 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
27705 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
27706 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
27708 2017-05-23 Tom de Vries <tom@codesourcery.com>
27710 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
27713 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27715 * cgraphunit.c (symbol_table::process_new_functions): Update.
27716 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
27717 (inline_generate_summary): Rename to ...
27718 (ipa_fn_summary_generate): ... this one.
27719 (inline_read_summary): Rename to ...
27720 (ipa_fn_summary_read): ... this one.
27721 (inline_write_summary): Rename to ...
27722 (ipa_fn_summary_write): ... this one.
27723 (inline_free_summary): Rename to ...
27724 (ipa_free_fn_summary): ... this one.
27725 (pass_data_local_fn_summary, pass_local_fn_summary,
27726 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
27727 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
27728 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
27729 make_pass_ipa_fn_summary): New.
27730 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
27731 inline_write_summary, inline_free_summary): Remove.
27732 (ipa_free_fn_summary) : New.
27733 * ipa-inline.c (ipa_inline): Update.
27734 (pass_ipa_inline): Do not generate summaries.
27735 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
27737 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
27738 and add pass_ipa_fn_summary.
27739 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
27741 (make_pass_inline_parameters): Remove.
27743 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
27745 * omp-low.c (struct omp_context): Remove "default_kind" member.
27748 * omp-offload.c (execute_oacc_device_lower): Remove the
27749 parallelism dimensions function attributes for unparallelized
27750 OpenACC kernels constructs.
27752 2017-05-23 Martin Liska <mliska@suse.cz>
27754 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
27756 (cgraph_edge::make_speculative): Likewise.
27757 (cgraph_edge::resolve_speculation): Likewise.
27758 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
27759 (cgraph_node::dump): Likewise.
27760 * cgraph.h: Likewise.
27761 * cgraphunit.c (analyze_functions): Likewise.
27762 (symbol_table::compile): Likewise.
27763 * ipa-cp.c (print_all_lattices): Likewise.
27764 (determine_versionability): Likewise.
27765 (initialize_node_lattices): Likewise.
27766 (ipcp_verify_propagated_values): Likewise.
27767 (estimate_local_effects): Likewise.
27768 (update_profiling_info): Likewise.
27769 (create_specialized_node): Likewise.
27770 (perhaps_add_new_callers): Likewise.
27771 (decide_about_value): Likewise.
27772 (decide_whether_version_node): Likewise.
27773 (identify_dead_nodes): Likewise.
27774 (ipcp_store_bits_results): Likewise.
27775 * ipa-devirt.c (dump_targets): Likewise.
27776 (ipa_devirt): Likewise.
27777 * ipa-icf.c (sem_item::dump): Likewise.
27778 (sem_function::equals): Likewise.
27779 (sem_variable::equals): Likewise.
27780 (sem_item_optimizer::read_section): Likewise.
27781 (sem_item_optimizer::execute): Likewise.
27782 (congruence_class::dump): Likewise.
27783 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
27784 (dump_inline_summary): Likewise.
27785 (estimate_node_size_and_time): Likewise.
27786 (inline_analyze_function): Likewise.
27787 * ipa-inline-transform.c (inline_call): Likewise.
27788 * ipa-inline.c (report_inline_failed_reason): Likewise.
27789 (want_early_inline_function_p): Likewise.
27790 (edge_badness): Likewise.
27791 (update_edge_key): Likewise.
27792 (inline_small_functions): Likewise.
27793 * ipa-profile.c (ipa_profile): Likewise.
27794 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
27795 (ipa_make_edge_direct_to_target): Likewise.
27796 (remove_described_reference): Likewise.
27797 (ipa_impossible_devirt_target): Likewise.
27798 (propagate_controlled_uses): Likewise.
27799 (ipa_print_node_params): Likewise.
27800 (ipcp_transform_function): Likewise.
27801 * ipa-pure-const.c (pure_const_read_summary): Likewise.
27802 (propagate_pure_const): Likewise.
27803 * ipa-reference.c (generate_summary): Likewise.
27804 (read_write_all_from_decl): Likewise.
27805 (propagate): Likewise.
27806 (ipa_reference_read_optimization_summary): Likewise.
27807 * ipa-utils.c (ipa_merge_profiles): Likewise.
27808 * ipa.c (walk_polymorphic_call_targets): Likewise.
27809 (symbol_table::remove_unreachable_nodes): Likewise.
27810 (ipa_single_use): Likewise.
27811 * passes.c (execute_todo): Likewise.
27812 * predict.c (drop_profile): Likewise.
27813 * symtab.c (symtab_node::get_dump_name): New function.
27814 (symtab_node::dump_name): Likewise.
27815 (symtab_node::dump_asm_name): Likewise.
27816 (symtab_node::dump_references): Likewise.
27817 (symtab_node::dump_referring): Likewise.
27818 (symtab_node::dump_base): Likewise.
27819 (symtab_node::debug_symtab): Likewise.
27820 * tree-sra.c (convert_callers_for_node): Likewise.
27821 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
27822 * value-prof.c (init_node_map): Likewise.
27824 2017-05-23 Martin Liska <mliska@suse.cz>
27826 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
27827 and symtab_node::debug_symtab to symbol_table::debug.
27828 * cgraphunit.c (analyze_functions): Use the renamed function.
27829 (symbol_table::compile): Likewise.
27830 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
27831 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
27832 * passes.c (execute_todo): Likewise.
27833 * symtab.c (symbol_table::dump): New function.
27834 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
27836 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27838 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
27839 that nonconst implies exec.
27841 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
27843 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
27844 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
27845 (inline_edge_summary_vec): Turn into ...
27846 (ipa_call_summaries): ... this one.
27847 (redirect_to_unreachable, edge_set_predicate,
27848 evaluate_properties_for_edge, inline_summary_alloc,
27849 reset_ipa_call_summary, reset_inline_summary,
27850 inline_summary_t::duplicate): Update.
27851 (inline_edge_duplication_hook): Turn to ...
27852 (ipa_call_summary_t::duplicate): ... this one.
27853 (inline_edge_removal_hook): Turn to ...
27854 (ipa_call_summary_t::remove): ... this one.
27855 (dump_inline_edge_summary): Turn to ...
27856 (dump_ipa_call_summary): ... this one.
27857 (estimate_function_body_sizes): Update.
27858 (inline_update_callee_summaries): Update.
27859 (remap_edge_change_prob): Update.
27860 (remap_edge_summaries): Update.
27861 (inline_merge_summary): Update.
27862 (do_estimate_edge_time): Update.
27863 (inline_generate_summary): Update.
27864 (inline_read_section): Update.
27865 (inline_read_summary): Update.
27866 (inline_free_summary): Update.
27867 * ipa-inline.c (can_inline_edge_p): Update.
27868 (compute_inlined_call_time): Update.
27869 (want_inline_small_function_p): Update.
27870 (edge_badness): Update.
27871 (early_inliner): Update.
27872 * ipa-inline.h (inline_edge_summary): Turn to ...
27873 (ipa_call_summary): ... this one.
27874 (ipa_call_summary_t): New class.
27875 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
27876 (ipa_call_summaries): New.
27877 (inline_edge_summary): Remove.
27878 (estimate_edge_growth): Update.
27879 * ipa-profile.c (ipa_propagate_frequency_1): Update.
27880 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
27881 * ipa-split.c (execute_split_functions): Update.
27882 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
27884 2017-05-23 Tom de Vries <tom@codesourcery.com>
27886 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27887 attributes): Document rdrand effective target.
27889 2017-05-23 Tom de Vries <tom@codesourcery.com>
27891 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
27892 attributes): Sort alphabetically.
27894 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
27896 * config/avr/genmultilib.awk: Use gsub instead of gensub.
27898 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
27901 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
27902 V2DF/V2DI splat into two separate patterns, one that handles
27903 registers, and the other that only handles memory. Drop support
27904 for splatting from a GPR on ISA 2.07 and then splitting the
27905 splat into direct move and splat.
27906 (vsx_splat_<mode>_reg): Likewise.
27907 (vsx_splat_<mode>_mem): Likewise.
27909 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
27911 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
27913 2017-05-22 Jakub Jelinek <jakub@redhat.com>
27915 PR middle-end/80809
27916 * omp-low.c (finish_taskreg_remap): New function.
27917 (finish_taskreg_scan): If unit size of ctx->record_type
27918 is non-constant, unshare the size expression and replace
27919 decls in it with possible outer var refs.
27921 PR middle-end/80809
27922 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
27923 GOVD_SHARED rather than GOVD_PRIVATE with it.
27924 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
27925 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
27927 PR middle-end/80853
27928 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
27929 as last argument to build_outer_var_ref for pointer bases of array
27930 section reductions.
27932 2017-05-19 Martin Sebor <msebor@redhat.com>
27934 * print-tree.c (print_node): Print DECL_READ_P flag.
27936 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27938 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
27939 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
27940 * cgraph.c: Likewise.
27941 * cgraphunit.c: Likewise.
27942 * gengtype.c: Likewise.
27943 * ipa-cp.c: Likewise.
27944 * ipa-devirt.c: Likewise.
27945 * ipa-icf.c: Likewise.
27946 * ipa-predicate.c: Likewise.
27947 * ipa-profile.c: Likewise.
27948 * ipa-prop.c: Likewise.
27949 * ipa-split.c: Likewise.
27951 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
27952 edge_predicate_pool, dump_inline_hints,
27953 inline_summary::account_size_time, redirect_to_unreachable,
27954 edge_set_predicate, set_hint_predicate,
27955 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
27956 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
27957 inline_summary_t::remove, remap_hint_predicate_after_duplication,
27958 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
27959 ipa_call_summary_t::remove, initialize_growth_caches,
27960 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
27961 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
27962 mark_modified, unmodified_parm_1, unmodified_parm,
27963 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
27964 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
27965 compute_bb_predicates, will_be_nonconstant_expr_predicate,
27966 will_be_nonconstant_predicate, record_modified_bb_info,
27967 get_minimal_bb, record_modified, param_change_prob,
27968 phi_result_unknown_predicate, predicate_for_phi_result,
27969 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
27970 estimate_function_body_sizes, compute_inline_parameters,
27971 compute_inline_parameters_for_curren, pass_data_inline_parameters,
27972 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
27973 inline_update_callee_summaries, remap_edge_change_prob,
27974 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
27975 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
27976 inline_analyze_function, inline_summary_t::insert,
27977 inline_generate_summary, read_ipa_call_summary, inline_read_section,
27978 inline_read_summary, write_ipa_call_summary, inline_write_summary,
27979 inline_free_summary): Move to ipa-fnsummary.h
27980 (predicate_t): Remove.
27981 * ipa-fnsummary.c: New file.
27982 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
27983 (enum inline_hints_vals, inline_hints, agg_position_info,
27984 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
27985 inline_summaries, ipa_call_summary, ipa_call_summary_t,
27986 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
27987 dump_inline_summary, dump_inline_hints, inline_generate_summary,
27988 inline_read_summary, inline_write_summary, inline_free_summary,
27989 inline_analyze_function, initialize_inline_failed,
27990 inline_merge_summary, inline_update_overall_summary,
27991 compute_inline_parameters): Move to ipa-fnsummary.h
27992 * ipa-fnsummary.h: New file.
27993 * ipa-inline-transform.h: Include ipa-inline.h.
27994 * ipa-inline.c: LIkewise.
27996 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
27998 * ipa-inline.c (edge_badness): Use inlined_time instead of
27999 inline_summaries->get.
28001 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28003 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
28005 2017-05-22 Nathan Sidwell <nathan@acm.org>
28007 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
28008 (fdump-lang): Document 'raw' option.
28009 * dumpfile.h (TDI_tu): Delete.
28010 * dumpfile.c (dump_files): Remove translation-unit.
28011 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
28013 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
28015 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
28016 command option from $(AWK) call.
28017 * config/avr/genmultilib.awk: Simplify and rewrite so that it
28018 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
28019 [FORMAT]: Remove handling of variable.
28020 * config/avr/t-multilib: Regenerate.
28022 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28024 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
28026 (dump_inline_summary): Do not print self_time.
28027 (estimate_function_body_sizes): Do not set self_time.
28028 (compute_inline_parameters): Likewise.
28029 (inline_read_section, inline_write_summary): Do not stream self_time.
28030 * ipa-inline.h (inline_summary): Drop self_time.
28032 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
28034 * ipa-inline-analysis.c (account_size_time): Rename to ...
28035 (inline_summary::account_size_time): ... this one.
28036 (reset_ipa_call_summary): Turn to ...
28037 (ipa_call_summary::reset): ... this one.
28038 (reset_inline_summary): Turn to ...
28039 (inline_summary::reset): ... this one.
28040 (inline_summary_t::remove): Update.
28041 (inline_summary_t::duplicate): Update.
28042 (ipa_call_summary_t::remove): Update.
28043 (dump_inline_summary): Update.
28044 (estimate_function_body_sizes): Update.
28045 (compute_inline_parameters): Update.
28046 (estimate_node_size_and_time): Update.
28047 (inline_merge_summary): Update.
28048 (inline_update_overall_summary): Update.
28049 (inline_read_section): Update.
28050 (inline_write_summary): Update.
28051 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
28052 add account_size_time and reset member functions.
28053 (ipa_call_summary): Add reset function.
28054 * ipa-predicate.h (predicate::operator &): Constify.
28056 2017-05-22 Richard Biener <rguenther@suse.de>
28058 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
28060 2017-05-19 Jason Merrill <jason@redhat.com>
28062 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
28064 2017-05-19 Marek Polacek <polacek@redhat.com>
28067 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
28068 TYPE_OVERFLOW_WRAPS checks.
28070 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
28072 * tree-core.h (enum omp_clause_default_kind): Add
28073 "OMP_CLAUSE_DEFAULT_PRESENT".
28074 * tree-pretty-print.c (dump_omp_clause): Handle it.
28075 * gimplify.c (enum gimplify_omp_var_data): Add
28076 "GOVD_MAP_FORCE_PRESENT".
28077 (gimplify_adjust_omp_clauses_1): Map it to
28078 "GOMP_MAP_FORCE_PRESENT".
28079 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
28081 * gimplify.c (oacc_default_clause): Clarify.
28083 2017-05-19 Nathan Sidwell <nathan@acm.org>
28085 LANG_HOOK_REGISTER_DUMPS
28086 * toplev.c (general_init): Call register dump lang hook.
28087 * doc/invoke.texi: Document -fdump-lang option family.
28088 * dumpfile.c (dump_files): Remove class dump here.
28089 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
28090 * dumpfile.h (tree_dump_index): Remove TDI_class.
28091 * langhooks-def.h (lhd_register_dumps): Declare.
28092 (LANG_HOOKS_REGISTER_DUMPS): Define.
28093 (LANG_HOOKS_INITIALIZER): Add it.
28094 * langhooks.c (lhd_register_dumps): Define.
28095 * langhooks.h (struct lang_hooks): Add register_dumps.
28097 2017-05-19 Nathan Sidwell <nathan@acm.org>
28099 * context.h (context::set_passes): New.
28100 * context.c (context::context): Do not create pass manager.
28101 * toplev.c (general_init): Create pass manager here.
28103 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
28105 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
28106 use this splitter if two add or or instructions would also work for
28107 the constant we want to generate.
28109 2017-05-19 Richard Biener <rguenther@suse.de>
28112 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
28113 predicate evaluation.
28115 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28117 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
28119 * ipa-inline.c (want_inline_small_function_p): Do not cast to
28122 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
28124 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
28125 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
28126 (inline_edge_summary_vec): Turn into ...
28127 (ipa_call_summaries): ... this one.
28128 (redirect_to_unreachable, edge_set_predicate,
28129 evaluate_properties_for_edge, inline_summary_alloc,
28130 reset_ipa_call_summary, reset_inline_summary,
28131 inline_summary_t::duplicate): Update.
28132 (inline_edge_duplication_hook): Turn to ...
28133 (ipa_call_summary_t::duplicate): ... this one.
28134 (inline_edge_removal_hook): Turn to ...
28135 (ipa_call_summary_t::remove): ... this one.
28136 (dump_inline_edge_summary): Turn to ...
28137 (dump_ipa_call_summary): ... this one.
28138 (estimate_function_body_sizes): Update.
28139 (inline_update_callee_summaries): Update.
28140 (remap_edge_change_prob): Update.
28141 (remap_edge_summaries): Update.
28142 (inline_merge_summary): Update.
28143 (do_estimate_edge_time): Update.
28144 (inline_generate_summary): Update.
28145 (inline_read_section): Update.
28146 (inline_read_summary): Update.
28147 (inline_free_summary): Update.
28148 * ipa-inline.c (can_inline_edge_p): Update.
28149 (compute_inlined_call_time): Update.
28150 (want_inline_small_function_p): Update.
28151 (edge_badness): Update.
28152 (early_inliner): Update.
28153 * ipa-inline.h (inline_edge_summary): Turn to ...
28154 (ipa_call_summary): ... this one.
28155 (ipa_call_summary_t): New class.
28156 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
28157 (ipa_call_summaries): New.
28158 (inline_edge_summary): Remove.
28159 (estimate_edge_growth): Update.
28160 * ipa-profile.c (ipa_propagate_frequency_1): Update.
28161 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
28162 * ipa-split.c (execute_split_functions): Update.
28163 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
28165 2017-05-19 Richard Biener <rguenther@suse.de>
28167 PR middle-end/80764
28168 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
28170 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
28172 * config/rs6000/rs6000.c (struct machine_function): Add field
28173 fpr_is_wrapped_separately.
28174 (rs6000_get_separate_components): Use 64 components. Handle the
28175 new FPR components.
28176 (rs6000_components_for_bb): Handle the FPR components.
28177 (rs6000_emit_prologue_components): Handle the FPR components.
28178 (rs6000_emit_epilogue_components): Handle the FPR components.
28179 (rs6000_set_handled_components): Handle the FPR components.
28180 (rs6000_emit_prologue): Don't output prologue code for those FPRs
28181 that are already separately shrink-wrapped.
28182 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
28183 that are already separately shrink-wrapped.
28185 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28188 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
28191 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
28192 (define_peephole2 for Altivec d-form load): Add peepholes to catch
28193 cases where the register allocator uses a move and an offsettable
28194 memory operation to/from a FPR register on ISA 2.06/2.07.
28195 (define_peephole2 for Altivec d-form store): Likewise.
28197 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
28200 * config/i386/mmx.md (*mov<mode>_internal): Enable
28201 alternatives 11, 12, 13 and 14 also for 32bit targets.
28202 Remove alternatives 15, 16, 17 and 18.
28203 * config/i386/sse.md (vec_concatv2di): Change
28204 alternative (!x, *y) to (x, ?!*Yn).
28206 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
28208 * dumpfile.h (enum dump_kind): Remove stray comma.
28210 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28212 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
28213 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
28214 predicate::num_conditions
28215 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
28216 (CHANGED): turn into predicate::changed.
28217 (agg_position_info): Move to ipa-predicate.h
28218 (add_condition, predicate::add_clause, predicate::operator &=,
28219 predicate::or_with, predicate::evaluate, predicate::probability,
28220 dump_condition, dump_clause, predicate::dump,
28221 predicate::remap_after_duplication, predicate::remap_after_inlining,
28222 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
28223 (evaluate_conditions_for_known_args): Update.
28224 (set_cond_stmt_execution_predicate): Update.
28225 * ipa-inline.h: Include ipa-predicate.h
28226 (condition, inline_param_summary, conditions, agg_position_info,
28227 predicate): Move to ipa-predicate.h
28228 * ipa-predicate.c: New file.
28229 * ipa-predicate.h: New file.
28231 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
28233 * final.c (leaf_function_p): Check we are not in a sequence.
28235 2017-05-18 Martin Liska <mliska@suse.cz>
28237 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
28238 * dumpfile.c (dump_register): Use new enum dump_kind.
28239 (get_dump_file_name): Likewise.
28240 (dump_enable_all): Likewise.
28241 (dump_switch_p_1): Likewise.
28242 (enable_rtl_dump_file): Remove usage of TDF_RTL.
28243 * dumpfile.h (enum dump_kind): New enum type.
28244 (struct dump_file_info): Create constructor and
28245 format fields and comments.
28246 * passes.c (pass_manager::register_one_dump_file):
28248 * statistics.c (statistics_early_init): Likewise.
28249 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
28250 TDF_TREE with TDF_SLIM.
28251 (gather_memory_references_ref): Likewise.
28253 2017-05-18 Martin Liska <mliska@suse.cz>
28255 * vec.h (struct vnull): Use it.
28257 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
28259 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
28260 (true_predicate, false_predicate, true_predicate_p,
28261 false_predicate_p): Remove.
28262 (single_cond_predicate, not_inlined_predicate): Turn to member function
28264 (add_condition): Update.
28265 (add_clause): Turn to...
28266 (predicate::add_clause): ... this one; update; allow passing NULL
28268 (and_predicates): Turn to ...
28269 (predicate::operator &=): ... this one.
28270 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
28271 (or_predicates): Turn to ...
28272 (predicate::or_with): ... this one.
28273 (evaluate_predicate): Turn to ...
28274 (predicate::evaluate): ... this one.
28275 (predicate_probability): Turn to ...
28276 (predicate::probability): ... this one.
28277 (dump_condition): Update.
28278 (dump_predicate): Turn to ...
28279 (predicate::dump): ... this one.
28280 (account_size_time): Update.
28281 (edge_set_predicate): Update.
28282 (set_hint_predicate): UPdate.
28283 (evaluate_conditions_for_known_args): Update.
28284 (evaluate_properties_for_edge): Update.
28285 (remap_predicate_after_duplication): Turn to...
28286 (predicate::remap_after_duplication): ... this one.
28287 (remap_hint_predicate_after_duplication): Update.
28288 (inline_summary_t::duplicate): UPdate.
28289 (dump_inline_edge_summary): Update.
28290 (dump_inline_summary): Update.
28291 (set_cond_stmt_execution_predicate): Update.
28292 (set_switch_stmt_execution_predicate): Update.
28293 (compute_bb_predicates): Update.
28294 (will_be_nonconstant_expr_predicate): Update.
28295 (will_be_nonconstant_predicate): Update.
28296 (phi_result_unknown_predicate): Update.
28297 (predicate_for_phi_result): Update.
28298 (array_index_predicate): Update.
28299 (estimate_function_body_sizes): Update.
28300 (estimate_node_size_and_time): Update.
28301 (estimate_ipcp_clone_size_and_time): Update.
28302 (remap_predicate): Rename to ...
28303 (predicate::remap_after_inlining): ... this one.
28304 (remap_hint_predicate): Update.
28305 (inline_merge_summary): Update.
28306 (inline_update_overall_summary): Update.
28307 (estimate_size_after_inlining): Update.
28308 (read_predicate): Rename to ...
28309 (predicate::stream_in): ... this one.
28310 (read_inline_edge_summary): Update.
28311 (write_predicate): Rename to ...
28312 (predicate::stream_out): ... this one.
28313 (write_inline_edge_summary): Update.
28314 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
28315 (clause_t): Turn to uint32_t
28316 (predicate): Turn to class; implement constructor and operators
28318 (size_time_entry): Update.
28319 (inline_summary): Update.
28320 (inline_edge_summary): Update.
28322 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
28324 * fold-const.c (fold_binary_loc): Move transformation...
28325 * match.pd (C - X CMP X): ... here.
28327 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
28329 * config/sparc/sparc.c (sparc_option_override): Set function
28330 alignment for -mcpu=niagara7 to 64 to match the I$ line.
28331 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
28333 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
28335 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
28337 2017-05-18 Marek Polacek <polacek@redhat.com>
28340 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
28341 (pass_ubsan::execute): Call gimple_assign_single_p instead of
28342 gimple_assign_load_p.
28344 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
28346 PR middle-end/80692
28347 * real.c (do_compare): Give decimal_do_compare preference over
28348 comparing just the signs.
28350 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
28352 * doc/md.texi (Canonicalization of Instructions): Describe the
28353 canonical form of instructions that inherently set a condition
28356 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
28358 PR middle-end/80775
28359 * tree-cfg.c: Move deletion of unreachable case statements to after
28360 the merging of consecutive case labels.
28362 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
28364 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
28365 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
28366 restoring of callee-saved registers.
28368 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
28370 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
28371 * config/visium/visium.c (single_set_and_flags): Likewise.
28372 * config/visium/visium.md (Substitutions): Likewise.
28374 2017-05-17 Martin Liska <mliska@suse.cz>
28376 * cfg.c: Introduce dump_flags_t type and
28377 use it instead of int type.
28379 * cfghooks.c: Likewise.
28380 * cfghooks.h (struct cfg_hooks): Likewise.
28381 * cfgrtl.c: Likewise.
28382 * cfgrtl.h: Likewise.
28383 * cgraph.c (cgraph_node::get_body): Likewise.
28384 * coretypes.h: Likewise.
28385 * domwalk.c: Likewise.
28386 * domwalk.h: Likewise.
28387 * dumpfile.c (struct dump_option_value_info): Likewise.
28388 (dump_enable_all): Likewise.
28389 (dump_switch_p_1): Likewise.
28390 (opt_info_switch_p): Likewise.
28391 * dumpfile.h (enum tree_dump_index): Likewise.
28392 (struct dump_file_info): Likewise.
28393 * genemit.c: Likewise.
28394 * generic-match-head.c: Likewise.
28395 * gengtype.c (open_base_files): Likewise.
28396 * gimple-pretty-print.c: Likewise.
28397 * gimple-pretty-print.h: Likewise.
28398 * graph.c (print_graph_cfg): Likewise.
28399 * graphite-scop-detection.c (dot_all_sese): Likewise.
28400 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
28401 * loop-unroll.c (report_unroll): Likewise.
28402 * passes.c (pass_manager::register_one_dump_file): Likewise.
28403 * print-tree.c: Likewise.
28404 * statistics.c: Likewise.
28405 * tree-cfg.c: Likewise.
28406 * tree-cfg.h: Likewise.
28407 * tree-dfa.c: Likewise.
28408 * tree-dfa.h: Likewise.
28409 * tree-dump.c (dump_function): Likewise.
28410 * tree-dump.h (struct dump_info): Likewise.
28411 * tree-pretty-print.c: Likewise.
28412 * tree-pretty-print.h: Likewise.
28413 * tree-ssa-live.c: Likewise.
28414 * tree-ssa-live.h: Likewise.
28415 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
28416 * tree-vect-loop.c: Likewise.
28417 * tree-vect-slp.c: Likewise.
28419 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
28420 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28422 PR tree-optimization/80457
28423 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
28424 of all arguments to a statement as scalar_to_vec operations.
28425 (vectorizable_call): Adjust call to vect_model_simple_cost for
28427 (vectorizable_conversion): Likewise.
28428 (vectorizable_assignment): Likewise.
28429 (vectorizable_shift): Likewise.
28430 (vectorizable_operation): Likewise.
28431 (vectorizable_comparison): Likewise.
28432 (vect_is_simple_cond): Record the def types for operands.
28433 (vectorizable_condition): Likewise, call vect_model_simple_cost.
28434 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
28435 for statement argument count.
28437 2017-05-16 Carl Love <cel@us.ibm.com>
28439 * config/rs6000/rs6000-c: Add support for built-in functions
28440 vector unsigned long long vec_bperm (vector unsigned long long,
28441 vector unsigned char)
28442 vector signed long long vec_mule (vector signed int,
28444 vector unsigned long long vec_mule (vector unsigned int,
28445 vector unsigned int)
28446 vector signed long long vec_mulo (vector signed int,
28448 vector unsigned long long vec_mulo (vector unsigned int,
28449 vector unsigned int)
28450 vector signed char vec_sldw (vector signed char,
28451 vector signed char,
28453 vector unsigned char vec_sldw (vector unsigned char,
28454 vector unsigned char,
28456 vector signed short vec_sldw (vector signed short,
28457 vector signed short,
28459 vector unsigned short vec_sldw (vector unsigned short,
28460 vector unsigned short,
28462 vector signed int vec_sldw (vector signed int,
28465 vector unsigned int vec_sldw (vector unsigned int,
28466 vector unsigned int,
28468 vector signed long long vec_sldw (vector signed long long,
28469 vector signed long long,
28471 vector unsigned long long vec_sldw (vector unsigned long long,
28472 vector unsigned long long,
28474 * config/rs6000/rs6000-c: Add support for built-in functions
28475 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
28476 * config/rs6000/altivec.h: Add defintion for vec_sldw.
28477 * doc/extend.texi: Update the built-in documentation for the
28478 new built-in functions.
28480 2017-05-16 Marek Polacek <polacek@redhat.com>
28484 * tree.c (save_expr): Don't fold the expression.
28486 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
28488 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
28489 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
28490 and (?*y,m). Update insn attributes.
28492 2017-05-16 Martin Liska <mliska@suse.cz>
28494 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
28495 flags argument of print_gimple_stmt, print_gimple_expr,
28496 print_generic_stmt and print_generic_expr.
28497 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
28498 * coretypes.h: Likewise.
28499 * except.c (dump_eh_tree): Likewise.
28500 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
28501 * gimple-pretty-print.h: Likewise.
28502 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
28503 (backprop::push_to_worklist): Likewise.
28504 (backprop::pop_from_worklist): Likewise.
28505 (backprop::process_use): Likewise.
28506 (backprop::intersect_uses): Likewise.
28507 (note_replacement): Likewise.
28508 * gimple-ssa-store-merging.c
28509 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
28510 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
28511 (pass_store_merging::execute): Likewise.
28512 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
28513 (ssa_base_cand_dump_callback): Likewise.
28514 (dump_incr_vec): Likewise.
28515 (replace_refs): Likewise.
28516 (replace_mult_candidate): Likewise.
28517 (create_add_on_incoming_edge): Likewise.
28518 (create_phi_basis): Likewise.
28519 (insert_initializers): Likewise.
28520 (all_phi_incrs_profitable): Likewise.
28521 (introduce_cast_before_cand): Likewise.
28522 (replace_one_candidate): Likewise.
28523 * gimplify.c (gimplify_expr): Likewise.
28524 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
28525 (set_rename): Likewise.
28526 (rename_uses): Likewise.
28527 (copy_loop_phi_nodes): Likewise.
28528 (add_close_phis_to_merge_points): Likewise.
28529 (copy_loop_close_phi_args): Likewise.
28530 (copy_cond_phi_args): Likewise.
28531 (graphite_copy_stmts_from_block): Likewise.
28532 (translate_pending_phi_nodes): Likewise.
28533 * graphite-poly.c (print_pdr): Likewise.
28534 (dump_gbb_cases): Likewise.
28535 (dump_gbb_conditions): Likewise.
28536 (print_scop_params): Likewise.
28537 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
28538 (build_cross_bb_scalars_use): Likewise.
28539 (gather_bbs::before_dom_children): Likewise.
28540 * hsa-dump.c (dump_hsa_immed): Likewise.
28541 * ipa-cp.c (print_ipcp_constant_value): Likewise.
28542 (get_replacement_map): Likewise.
28543 * ipa-inline-analysis.c (dump_condition): Likewise.
28544 (estimate_function_body_sizes): Likewise.
28545 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
28546 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
28547 * ipa-prop.c (ipa_dump_param): Likewise.
28548 (ipa_print_node_jump_functions_for_edge): Likewise.
28549 (ipa_modify_call_arguments): Likewise.
28550 (ipa_modify_expr): Likewise.
28551 (ipa_dump_param_adjustments): Likewise.
28552 (ipa_dump_agg_replacement_values): Likewise.
28553 (ipcp_modif_dom_walker::before_dom_children): Likewise.
28554 * ipa-pure-const.c (check_stmt): Likewise.
28555 (pass_nothrow::execute): Likewise.
28556 * ipa-split.c (execute_split_functions): Likewise.
28557 * omp-offload.c (dump_oacc_loop_part): Likewise.
28558 (dump_oacc_loop): Likewise.
28559 * trans-mem.c (tm_log_emit): Likewise.
28560 (tm_memopt_accumulate_memops): Likewise.
28561 (dump_tm_memopt_set): Likewise.
28562 (dump_tm_memopt_transform): Likewise.
28563 * tree-cfg.c (gimple_verify_flow_info): Likewise.
28564 (print_loop): Likewise.
28565 * tree-chkp-opt.c (chkp_print_addr): Likewise.
28566 (chkp_gather_checks_info): Likewise.
28567 (chkp_get_check_result): Likewise.
28568 (chkp_remove_check_if_pass): Likewise.
28569 (chkp_use_outer_bounds_if_possible): Likewise.
28570 (chkp_reduce_bounds_lifetime): Likewise.
28571 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
28572 (chkp_mark_completed_bounds): Likewise.
28573 (chkp_register_incomplete_bounds): Likewise.
28574 (chkp_mark_invalid_bounds): Likewise.
28575 (chkp_maybe_copy_and_register_bounds): Likewise.
28576 (chkp_build_returned_bound): Likewise.
28577 (chkp_get_bound_for_parm): Likewise.
28578 (chkp_build_bndldx): Likewise.
28579 (chkp_get_bounds_by_definition): Likewise.
28580 (chkp_generate_extern_var_bounds): Likewise.
28581 (chkp_get_bounds_for_decl_addr): Likewise.
28582 * tree-chrec.c (chrec_apply): Likewise.
28583 * tree-data-ref.c (dump_data_reference): Likewise.
28584 (dump_subscript): Likewise.
28585 (dump_data_dependence_relation): Likewise.
28586 (analyze_overlapping_iterations): Likewise.
28587 * tree-inline.c (expand_call_inline): Likewise.
28588 (tree_function_versioning): Likewise.
28589 * tree-into-ssa.c (dump_defs_stack): Likewise.
28590 (dump_currdefs): Likewise.
28591 (dump_names_replaced_by): Likewise.
28592 (dump_update_ssa): Likewise.
28593 (update_ssa): Likewise.
28594 * tree-object-size.c (pass_object_sizes::execute): Likewise.
28595 * tree-parloops.c (build_new_reduction): Likewise.
28596 (try_create_reduction_list): Likewise.
28597 (ref_conflicts_with_region): Likewise.
28598 (oacc_entry_exit_ok_1): Likewise.
28599 (oacc_entry_exit_single_gang): Likewise.
28600 * tree-pretty-print.h: Likewise.
28601 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
28602 (get_scalar_evolution): Likewise.
28603 (add_to_evolution): Likewise.
28604 (get_loop_exit_condition): Likewise.
28605 (analyze_evolution_in_loop): Likewise.
28606 (analyze_initial_condition): Likewise.
28607 (analyze_scalar_evolution): Likewise.
28608 (instantiate_scev): Likewise.
28609 (number_of_latch_executions): Likewise.
28610 (gather_chrec_stats): Likewise.
28611 (final_value_replacement_loop): Likewise.
28612 (scev_const_prop): Likewise.
28613 * tree-sra.c (dump_access): Likewise.
28614 (disqualify_candidate): Likewise.
28615 (create_access): Likewise.
28616 (reject): Likewise.
28617 (maybe_add_sra_candidate): Likewise.
28618 (create_access_replacement): Likewise.
28619 (analyze_access_subtree): Likewise.
28620 (analyze_all_variable_accesses): Likewise.
28621 (sra_modify_assign): Likewise.
28622 (initialize_constant_pool_replacements): Likewise.
28623 (find_param_candidates): Likewise.
28624 (decide_one_param_reduction): Likewise.
28625 (replace_removed_params_ssa_names): Likewise.
28626 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
28627 * tree-ssa-copy.c (dump_copy_of): Likewise.
28628 (copy_prop_visit_cond_stmt): Likewise.
28629 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
28630 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
28631 (record_equivalences_from_stmt): Likewise.
28632 * tree-ssa-dse.c (compute_trims): Likewise.
28633 (delete_dead_call): Likewise.
28634 (delete_dead_assignment): Likewise.
28635 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
28636 (forward_propagate_into_cond): Likewise.
28637 (pass_forwprop::execute): Likewise.
28638 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
28639 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
28641 (move_computations_worker): Likewise.
28642 (execute_sm): Likewise.
28643 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
28644 (remove_exits_and_undefined_stmts): Likewise.
28645 (remove_redundant_iv_tests): Likewise.
28646 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
28647 (adjust_iv_update_pos): Likewise.
28648 * tree-ssa-math-opts.c (bswap_replace): Likewise.
28649 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
28650 (value_replacement): Likewise.
28651 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
28652 * tree-ssa-pre.c (print_pre_expr): Likewise.
28653 (get_representative_for): Likewise.
28654 (create_expression_by_pieces): Likewise.
28655 (insert_into_preds_of_block): Likewise.
28656 (eliminate_insert): Likewise.
28657 (eliminate_dom_walker::before_dom_children): Likewise.
28658 (eliminate): Likewise.
28659 (remove_dead_inserted_code): Likewise.
28660 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
28661 * tree-ssa-reassoc.c (get_rank): Likewise.
28662 (eliminate_duplicate_pair): Likewise.
28663 (eliminate_plus_minus_pair): Likewise.
28664 (eliminate_not_pairs): Likewise.
28665 (undistribute_ops_list): Likewise.
28666 (eliminate_redundant_comparison): Likewise.
28667 (update_range_test): Likewise.
28668 (optimize_range_tests_var_bound): Likewise.
28669 (optimize_vec_cond_expr): Likewise.
28670 (rewrite_expr_tree): Likewise.
28671 (rewrite_expr_tree_parallel): Likewise.
28672 (linearize_expr): Likewise.
28673 (break_up_subtract): Likewise.
28674 (linearize_expr_tree): Likewise.
28675 (attempt_builtin_powi): Likewise.
28676 (attempt_builtin_copysign): Likewise.
28677 (transform_stmt_to_copy): Likewise.
28678 (transform_stmt_to_multiply): Likewise.
28679 (dump_ops_vector): Likewise.
28680 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
28681 (print_scc): Likewise.
28682 (set_ssa_val_to): Likewise.
28683 (visit_reference_op_store): Likewise.
28684 (visit_use): Likewise.
28685 (sccvn_dom_walker::before_dom_children): Likewise.
28686 (run_scc_vn): Likewise.
28687 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
28689 (expr_hash_elt::print): Likewise.
28690 (const_and_copies::pop_to_marker): Likewise.
28691 (const_and_copies::record_const_or_copy_raw): Likewise.
28692 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
28693 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
28694 (dump_predicates): Likewise.
28695 (find_uninit_use): Likewise.
28696 (warn_uninitialized_phi): Likewise.
28697 (pass_late_warn_uninitialized::execute): Likewise.
28698 * tree-ssa.c (verify_vssa): Likewise.
28699 (verify_ssa): Likewise.
28700 (maybe_optimize_var): Likewise.
28701 * tree-vrp.c (dump_value_range): Likewise.
28702 (dump_all_value_ranges): Likewise.
28703 (dump_asserts_for): Likewise.
28704 (register_edge_assert_for_2): Likewise.
28705 (vrp_visit_cond_stmt): Likewise.
28706 (vrp_visit_switch_stmt): Likewise.
28707 (vrp_visit_stmt): Likewise.
28708 (vrp_visit_phi_node): Likewise.
28709 (simplify_cond_using_ranges_1): Likewise.
28710 (fold_predicate_in): Likewise.
28711 (evrp_dom_walker::before_dom_children): Likewise.
28712 (evrp_dom_walker::push_value_range): Likewise.
28713 (evrp_dom_walker::pop_value_range): Likewise.
28714 (execute_early_vrp): Likewise.
28716 2017-05-16 Richard Biener <rguenther@suse.de>
28718 * dwarf2out.c (loc_list_from_tree_1): Do not create
28719 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
28721 2017-05-16 Richard Biener <rguenther@suse.de>
28723 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
28725 (note_variable_value_in_expr): If we resolved the decl ref
28726 do not push to the stack.
28728 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
28730 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
28731 operations in fast-math mode.
28732 (vaddq_f16): Likewise.
28733 (vmul_f16): Likewise.
28734 (vmulq_f16): Likewise.
28735 (vsub_f16): Likewise.
28736 (vsubq_f16): Likewise.
28737 * config/arm/neon.md (add<mode>3): New.
28739 (fma:<VH:mode>3): New. Also remove outdated comment.
28742 2017-05-16 Martin Liska <mliska@suse.cz>
28746 * ipa-devirt.c (warn_types_mismatch): Fix typo.
28747 (odr_types_equivalent_p): Likewise.
28749 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
28751 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
28753 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
28756 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
28757 non-interunit SSE move alternatives with '?'.
28758 (zero-extendsidi peephole2): New peephole to skip intermediate
28759 general register in SSE zero-extend sequence.
28761 2017-05-15 Jeff Law <law@redhat.com>
28763 * reorg.c (relax_delay_slots): Create a new variable to hold
28764 the temporary target rather than clobbering TARGET_LABEL.
28766 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
28767 missing argument to extract_bit_field call.
28768 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
28770 2017-05-15 Martin Liska <mliska@suse.cz>
28773 * gcc.c (process_command): Do not allow empty argument of -o option.
28775 2017-05-15 Renlin Li <renlin.li@arm.com>
28777 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
28778 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
28779 * config/aarch64/constraints.md (Usf): Add long call check.
28780 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
28781 (call_value): Likewise.
28782 (sibcall): Likewise.
28783 (sibcall_value): Likewise.
28785 (call_value_insn): New.
28786 (sibcall_insn): Update rtx pattern.
28787 (sibcall_value_insn): Likewise.
28788 (call_internal): Remove.
28789 (call_value_internal): Likewise.
28790 (sibcall_internal): Likewise.
28791 (sibcall_value_internal): Likewise.
28792 (call_reg): Likewise.
28793 (call_symbol): Likewise.
28794 (call_value_reg): Likewise.
28795 (call_value_symbol): Likewise.
28797 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
28800 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
28802 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
28804 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
28805 compatible with CCGOCmode and with CCZmode.
28807 2017-05-14 Martin Sebor <msebor@redhat.com>
28809 PR middle-end/77671
28810 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
28811 (gimple_fold_builtin_snprintf): Same.
28812 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
28813 (gimple_fold_builtin_snprintf): Same.
28814 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
28815 of character types.
28816 (is_call_safe): New function.
28817 (try_substitute_return_value): Call it.
28818 (try_simplify_call): New function.
28819 (pass_sprintf_length::handle_gimple_call): Call it.
28821 2017-05-14 Martin Sebor <msebor@redhat.com>
28823 PR middle-end/80669
28824 * builtins.c (expand_builtin_stpncpy): Simplify.
28826 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
28828 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
28829 * config/i386/i386.h
28830 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
28831 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
28832 (struct machine_function): Add new members call_ms2sysv,
28833 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
28834 (struct machine_frame_state): New fields sp_realigned and
28835 sp_realigned_offset.
28836 * config/i386/i386.c
28837 (enum xlogue_stub): New enum.
28838 (enum xlogue_stub_sets): New enum.
28839 (class xlogue_layout): New class.
28840 (struct ix86_frame): New fields stack_realign_allocate_offset,
28841 stack_realign_offset and outlined_save_offset. Modify comments to
28842 detail stack layout when using out-of-line stubs.
28843 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
28844 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
28845 -mcall-ms2sysv-xlogues.
28846 (stub_managed_regs): New static variable.
28847 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
28848 registers managed by out-of-line stub.
28849 (disable_call_ms2sysv_xlogues): New function.
28850 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
28851 m->call_ms2sysv when appropriate and compute frame layout for
28853 (sp_valid_at, fp_valid_at): New inline functions.
28854 (choose_basereg): New function.
28855 (choose_baseaddr): Add align parameter, use choose_basereg and modify
28857 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
28858 Use align parameter of choose_baseaddr to generated aligned SSE movs
28860 (pro_epilogue_adjust_stack): Modify to track
28861 machine_frame_state::sp_realigned.
28862 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
28863 (ix86_nsaved_sseregs): Likewise.
28864 (ix86_emit_save_regs): Likewise.
28865 (ix86_emit_save_regs_using_mov): Likewise.
28866 (ix86_emit_save_sse_regs_using_mov): Likewise.
28867 (get_scratch_register_on_entry): Likewise.
28868 (gen_frame_set): New function.
28869 (gen_frame_load): Likewise.
28870 (gen_frame_store): Likewise.
28871 (emit_outlined_ms2sysv_save): Likewise.
28872 (emit_outlined_ms2sysv_restore): Likewise.
28873 (ix86_expand_prologue): Modify stack re-alignment code and call
28874 emit_outlined_ms2sysv_save when appropriate.
28875 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
28876 parameter rtx_insn *insn, which allows the function to be used to only
28877 generate the notes.
28878 (ix86_expand_epilogue): Modify validity checks of frame and stack
28879 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
28880 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
28881 * config/i386/predicates.md
28882 (save_multiple): New predicate.
28883 (restore_multiple): Likewise.
28884 * config/i386/sse.md
28885 (save_multiple<mode>): New pattern.
28886 (save_multiple_realign<mode>): Likewise.
28887 (restore_multiple<mode>): Likewise.
28888 (restore_multiple_and_return<mode>): Likewise.
28889 (restore_multiple_leave_return<mode>): Likewise.
28890 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
28892 2017-05-14 Julia Koval <julia.koval@intel.com>
28894 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
28895 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
28896 (__builtin_ia32_xsetbv): New builtins.
28897 * config/i386/i386.c (ix86_expand_special_args_builtin):
28899 (ix86_expand_builtin): Special expand for new intrinsics.
28900 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
28901 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
28902 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
28904 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28906 * cfganal.c (inverted_post_order_compute): Change argument type
28908 * cfganal.h (inverted_post_order_compute): Adjust prototype.
28909 * df-core.c (rest_of_handle_df_initialize): Adjust.
28910 (rest_of_handle_df_finish): Likewise.
28911 (df_analyze_1): Likewise.
28912 (df_analyze): Likewise.
28913 (loop_inverted_post_order_compute): Change argument to be a vec *.
28914 (df_analyze_loop): Adjust.
28915 (df_get_n_blocks): Likewise.
28916 (df_get_postorder): Likewise.
28917 * df.h (struct df_d): Change field to be a vec.
28918 * lcm.c (compute_laterin): Adjust.
28919 (compute_available): Likewise.
28920 * lra-lives.c (lra_create_live_ranges_1): Likewise.
28921 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
28922 * tree-ssa-pre.c (compute_antic): Likewise.
28924 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28926 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
28927 (depth_first_search::depth_first_search): Change structure init
28928 function to this constructor.
28929 (depth_first_search::add_bb): Rename function to this member.
28930 (depth_first_search::execute): Likewise.
28931 (flow_dfs_compute_reverse_finish): Adjust.
28933 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28935 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
28936 (longest_simple_path): Likewise.
28937 * shrink-wrap.c (spread_components): Likewise.
28938 (disqualify_problematic_components): Likewise.
28939 (emit_common_heads_for_components): Likewise.
28940 (emit_common_tails_for_components): Likewise.
28941 (insert_prologue_epilogue_for_components): Likewise.
28943 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28945 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
28948 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28950 * df-core.c (df_set_blocks): Start using auto_bitmap.
28951 (df_compact_blocks): Likewise.
28952 * df-problems.c (df_rd_confluence_n): Likewise.
28953 * df-scan.c (df_insn_rescan_all): Likewise.
28954 (df_process_deferred_rescans): Likewise.
28955 (df_update_entry_block_defs): Likewise.
28956 (df_update_exit_block_uses): Likewise.
28957 (df_entry_block_bitmap_verify): Likewise.
28958 (df_exit_block_bitmap_verify): Likewise.
28959 (df_scan_verify): Likewise.
28960 * lra-constraints.c (lra_constraints): Likewise.
28961 (undo_optional_reloads): Likewise.
28962 (lra_undo_inheritance): Likewise.
28963 * lra-remat.c (calculate_gen_cands): Likewise.
28964 (do_remat): Likewise.
28965 * lra-spills.c (assign_spill_hard_regs): Likewise.
28966 (spill_pseudos): Likewise.
28967 * tree-ssa-pre.c (bitmap_set_and): Likewise.
28968 (bitmap_set_subtract_values): Likewise.
28970 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28972 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
28973 management with auto_bitmap.
28974 (fix_inter_tick): Likewise.
28975 (fix_recovery_deps): Likewise.
28976 * ira.c (add_store_equivs): Likewise.
28977 (find_moveable_pseudos): Likewise.
28978 (split_live_ranges_for_shrink_wrap): Likewise.
28979 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
28980 (rtx_reuse_manager::seen_def_p): Likewise.
28981 (rtx_reuse_manager::set_seen_def): Likewise.
28982 * print-rtl.h (class rtx_reuse_manager): Likewise.
28984 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
28986 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
28988 (migrate_btr_def): Likewise.
28989 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
28990 * df-core.c (loop_post_order_compute): Likewise.
28991 (loop_inverted_post_order_compute): Likewise.
28992 * hsa-common.h: Likewise.
28993 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
28994 * init-regs.c (initialize_uninitialized_regs): Likewise.
28995 * ipa-inline.c (resolve_noninline_speculation): Likewise.
28996 (inline_small_functions): Likewise.
28997 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
28998 * ira.c (combine_and_move_insns): Likewise.
28999 (build_insn_chain): Likewise.
29000 * loop-invariant.c (find_invariants): Likewise.
29001 * lower-subreg.c (propagate_pseudo_copies): Likewise.
29002 * predict.c (tree_predict_by_opcode): Likewise.
29003 (predict_paths_leading_to): Likewise.
29004 (predict_paths_leading_to_edge): Likewise.
29005 (estimate_loops_at_level): Likewise.
29006 (estimate_loops): Likewise.
29007 * shrink-wrap.c (try_shrink_wrapping): Likewise.
29008 (spread_components): Likewise.
29009 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
29010 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
29011 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
29012 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
29013 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
29014 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
29015 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
29016 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
29017 (mark_threaded_blocks): Likewise.
29018 (thread_through_all_blocks): Likewise.
29019 * tree-ssa.c (verify_ssa): Likewise.
29020 (execute_update_addresses_taken): Likewise.
29021 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
29023 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29025 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
29027 (post_order_compute): Likewise.
29028 (inverted_post_order_compute): Likewise.
29029 (pre_and_rev_post_order_compute_fn): Likewise.
29031 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29033 * genrecog.c (int_set::int_set): Explicitly construct our
29034 auto_vec base class.
29035 * vec.h (auto_vec::auto_vec): New constructor.
29037 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29039 * bitmap.h (class auto_bitmap): New constructor taking
29040 bitmap_obstack * argument.
29042 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
29044 * bitmap.h (class auto_bitmap): Change type of m_bits to
29045 bitmap_head, and adjust ctor / dtor and member operators.
29047 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
29049 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
29050 when returned register mode doesn't match original mode.
29052 2017-05-12 Jeff Law <law@redhat.com>
29053 Jakub Jelinek <jakub@redhat.com>
29055 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
29056 we look for cc setter after the compare-elim changes.
29057 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
29058 within the vector to match what compare-elim now expects.
29059 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
29060 (xorsi3_flags, one_cmplsi2_flags): Likewise.
29062 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
29063 after the compare-elim changes.
29064 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
29065 the vector to match what compare-elim now expects.
29066 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
29067 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
29068 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
29069 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
29070 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
29072 * config/visium/visium.c (single_set_and_flags): Fix where
29073 we look for cc setter after the compare-elim changes.
29074 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
29075 with the vector to match what compare-elim now expects.
29076 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
29077 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
29078 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
29079 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
29080 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
29081 (neg<mode>2_insn_set_overflow): Likewise.
29083 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
29085 PR middle-end/79794
29086 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
29087 maybe_expand_insn call, set ops[0].target. If still set after call,
29088 set alt_rtl. Add extra arg to recursive calls.
29089 (extract_bit_field): Add alt_rtl argument. Pass to
29091 * expmed.h (extract_bit_field): Fix prototype.
29092 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
29093 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
29094 to extract_bit_field_calls.
29095 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
29096 Pass alt_rtl to extract_bit_field calls.
29097 * calls.c (store_unaligned_arguments_into_psuedos)
29098 load_register_parameters): Pass extra NULL to extract_bit_field calls.
29099 * optabs.c (maybe_legitimize_operand): Clear op->target when call
29101 * optabs.h (struct expand_operand): Add target bitfield.
29103 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29105 * compare-elim.c (try_eliminate_compare): Canonicalize
29106 operation with embedded compare to
29107 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
29108 (set (reg) (operation)].
29110 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
29112 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
29115 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
29116 cost of adding a carry flag for ADC instruction.
29117 [case MINUS]: Ignore the cost of subtracting a carry flag
29118 for SBB instruction.
29120 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
29122 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
29124 * config/rs6000/bmiintrin.h: New file.
29125 * config/rs6000/bmi2intrin.h: New file.
29126 * config/rs6000/x86intrin.h: New file.
29128 2017-05-12 Jeff Law <law@redhat.com>
29130 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
29133 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
29135 PR middle-end/80707
29136 * tree-cfg.c: Remove cfg edges of unreachable case statements.
29138 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29140 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29141 early expansion of vector divide builtins.
29142 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
29143 builtins identified as having unsigned arguments.
29145 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29147 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
29148 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
29149 expansion of vector logical operations (and, andc, or, xor,
29152 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29154 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
29155 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
29157 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
29159 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
29160 early GIMPLE expansion of vector multiplies.
29162 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29164 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
29165 TARGET_HAVE_MOVT conditional.
29166 (movt splitter): Likewise.
29168 2017-05-12 Richard Biener <rguenther@suse.de>
29170 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
29171 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29172 Fold all stmts not inplace.
29174 2017-05-12 Richard Biener <rguenther@suse.de>
29176 PR tree-optimization/80713
29177 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
29178 inserted_exprs bit for not removed stmts.
29180 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
29182 PR middle-end/69921
29183 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
29184 parallelized" attribute for parallelized OpenACC kernels.
29185 * omp-offload.c (execute_oacc_device_lower): Use it.
29187 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
29188 Set "oacc kernels" attribute.
29189 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
29190 parameter. Adjust all users.
29191 (oacc_fn_attrib_kernels_p): Remove function.
29192 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
29193 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
29194 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
29195 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
29196 assert "oacc kernels" attribute is set.
29198 2017-05-11 Carl Love <cel@us.ibm.com>
29200 * config/rs6000/rs6000-c: Add support for built-in functions
29201 vector unsigned char vec_popcnt (vector signed char)
29202 vector unsigned char vec_popcnt (vector unsigned char)
29203 vector unsigned short vec_popcnt (vector signed short)
29204 vector unsigned short vec_popcnt (vector unsigned short)
29205 vector unsigned int vec_popcnt (vector signed int)
29206 vector unsigned int vec_popcnt (vector unsigned int)
29207 vector unsigned long long vec_popcnt (vector signed long long)
29208 vector unsigned long long vec_popcnt (vector unsigned long long)
29209 vector signed long long vec_slo (vector signed long long,
29210 vector signed char)
29211 vector signed long long vec_slo (vector signed long long,
29212 vector unsigned char)
29213 vector unsigned long long vec_slo (vector unsigned long long,
29214 vector signed char)
29215 vector unsigned long long vec_slo (vector unsigned long long,
29216 vector unsigned char)
29217 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
29218 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
29219 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
29220 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
29221 * doc/extend.texi: Update the built-in documentation file for the
29222 new built-in functions.
29224 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
29226 * attribs.h (sorted_attr_string): Move machine independent
29227 functions for target clone support from the i386 port to common
29228 code. Rename ix86_function_versions to common_function_versions.
29229 Rename make_name to make_unique_name.
29230 (common_function_versions): Likewise.
29231 (make_unique_name): Likewise.
29232 (make_dispatcher_decl): Likewise.
29233 (is_function_default_version): Likewise.
29234 * attribs.c (attr_strcmp): Likewise.
29235 (sorted_attr_string): Likewise.
29236 (common_function_versions): Likewise.
29237 (make_unique_name): Likewise.
29238 (make_dispatcher_decl): Likewise.
29239 (is_function_default_version): Likewise.
29240 * config/i386/i386.c (attr_strcmp): Likewise.
29241 (sorted_attr_string): Likewise.
29242 (ix86_function_versions): Likewise.
29243 (make_name): Likewise.
29244 (make_dispatcher_decl): Likewise.
29245 (is_function_default_version): Likewise.
29246 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
29248 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29251 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
29252 Account for direct move costs for vec_construct of integer
29255 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
29258 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
29259 (UNSPEC_STX_ATOMIC): Ditto.
29260 (loaddi_via_sse): New insn.
29261 (storedi_via_sse): Ditto.
29262 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
29263 Update corresponding peephole2 patterns.
29264 (atomic_storedi_fpu): Ditto.
29266 2017-05-11 Julia Koval <julia.koval@intel.com>
29268 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
29269 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
29271 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
29272 (__builtin_ia32_rsqrt14ss_mask): New builtins.
29273 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
29275 2017-05-11 Nathan Sidwell <nathan@acm.org>
29277 * graphite-poly.c: Include dumpfile.h.
29279 * dumpfle.h (dump_function): Declare here ...
29280 * tree-dump.h (dump_function): ... not here.
29281 * dumpfile.c: #include tree-cfg.h.
29282 (dump_function): Move here from ...
29283 * tree-dump.c (dump_function): ... here.
29284 * gimplify.c: #include splay-tree.h, not tree-dump.h.
29285 * graphite-poly.c: Don't include tree-dump.h.
29286 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
29287 * print-tree.c: Likewise.
29288 * stor-layout.c: Likewise.
29289 * tree-nested.c: Likewise.
29291 * dumpfile.c (dump_start): Use TDF_FLAGS.
29292 (dump_enable_all): Fix TDF_KIND check thinko.
29294 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
29296 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
29297 array entries to represent two legal parameterizations of the
29298 overloaded __builtin_cmpb function, as represented by the
29299 P6_OV_BUILTIN_CMPB constant.
29300 (altivec_resolve_overloaded_builtin): Add special case handling
29301 for the __builtin_cmpb function, as represented by the
29302 P6_OV_BUILTIN_CMPB constant.
29303 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
29304 (BU_P6_64BIT_2): New macro.
29305 (BU_P6_OVERLOAD_2): New macro
29306 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
29307 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
29308 (CMPB): Add overload support to represent both 32-bit and 64-bit
29309 compare-bytes function.
29310 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
29311 support for TARGET_CMPB.
29312 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
29313 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
29314 documentation of the __builtin_cmpb overloaded built-in function.
29316 2017-05-11 Richard Biener <rguenther@suse.de>
29318 PR tree-optimization/80705
29319 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
29320 bases are not vectorizable.
29322 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29324 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
29325 when counting register pressure.
29327 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29329 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
29330 (iv_ca_more_deps): Renamed to ...
29331 (iv_ca_compare_deps): ... this.
29332 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
29334 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29336 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
29338 (determine_group_iv_costs): ... here.
29339 (find_inv_vars_cb): Record inv var if it's not recorded before.
29341 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29343 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
29344 (get_shiftadd_cost): Ditto.
29346 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29348 * tree-ssa-address.c: Include header file.
29349 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
29351 (add_to_parts): Refactor.
29352 (addr_to_parts): New parameter. Update use of move_hint_to_base.
29353 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
29356 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29358 PR tree-optimization/53090
29359 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
29361 (extract_cond_operands): Detect condition with IV on both sides
29362 and return COMP_IV_EXPR_2.
29363 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
29364 (rewrite_use_compare): Simplify by removing call to function
29365 extract_cond_operands.
29367 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29369 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
29370 (extract_cond_operands): Detect condition comparing against non-
29371 invariant bound and return appropriate enum value.
29372 (find_interesting_uses_cond): Update use of extract_cond_operands.
29373 Handle its return value accordingly.
29374 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
29376 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29378 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
29379 nonlinear iv_use computation in loop invariant sensitive way.
29381 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29383 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
29384 (find_iv_candidates): Call relate_compare_use_with_all_cands.
29386 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29388 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
29389 (dump_cand): Support iv_cand.inv_exprs.
29390 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
29392 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
29395 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29397 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
29399 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
29400 as local function. Include necessary header files.
29401 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
29403 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29405 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
29407 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29409 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
29410 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
29411 RSHIFT_EXPR and BIT_NOT_EXPR.
29413 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29415 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
29416 (adjust_setup_cost): New parameter supporting round up adjustment.
29417 (struct address_cost_data): Delete.
29418 (force_expr_to_var_cost): Don't bound cost with spill_cost.
29419 (split_address_cost, ptr_difference_cost): Delete.
29420 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
29421 (struct ainc_cost_data): New struct.
29422 (get_address_cost_ainc): New function.
29423 (get_address_cost, get_computation_cost): Reimplement.
29424 (determine_group_iv_cost_address): Record inv_expr for all uses of
29426 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
29427 (iv_ca_has_deps): Reimplemented to ...
29428 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
29430 (iv_ca_extend): Call iv_ca_more_deps.
29432 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29434 * tree-ssa-address.c (struct mem_address): Move to header file.
29435 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
29436 * tree-ssa-address.h (struct mem_address): Move from C file.
29437 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
29439 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29441 * tree-affine.h (aff_combination_type): New interface.
29442 (aff_combination_zero_p): Remove static.
29443 (aff_combination_const_p): New interface.
29444 (aff_combination_singleton_var_p): New interfaces.
29446 2017-05-11 Richard Biener <rguenther@suse.de>
29448 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
29449 Skip unreachable blocks and destinations.
29450 (eliminate): Move stmt removal and fixup ...
29451 (fini_eliminate): ... here. Skip inserted exprs.
29452 (pass_pre::execute): Move fini_pre after fini_eliminate.
29453 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
29454 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
29455 PRE to get rid of dead code that has invalid SSA form and
29456 split critical edges again.
29458 2017-05-11 Bin Cheng <bin.cheng@arm.com>
29460 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
29462 2017-05-11 Richard Biener <rguenther@suse.de>
29464 * passes.c (execute_function_todo): Verify loops if they are
29465 said to be up-to-date.
29466 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
29467 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
29469 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
29472 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
29473 handle calling assemble_external ourself.
29476 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
29477 modes with zero size. Enhance comment.
29479 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29481 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
29482 built-ins for vec_xl and vec_xst with short and char pointer
29485 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29487 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
29488 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
29489 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
29490 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
29491 (_mm_maskz_min_round_ss): New intrinsics.
29492 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
29493 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
29494 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
29495 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
29496 (__builtin_ia32_minss_mask_round): New builtins.
29497 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29498 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29499 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
29501 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
29502 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
29504 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
29507 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
29509 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
29510 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
29511 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
29512 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
29513 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
29514 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
29515 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
29516 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29517 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
29518 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
29519 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
29520 (__builtin_ia32_mulss_mask_round): New builtins.
29521 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
29522 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
29523 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
29525 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
29526 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
29528 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
29531 2017-05-10 Julia Koval <julia.koval@intel.com>
29533 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
29534 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
29535 (_mm256_setr_m128i): New intrinsics.
29537 2017-05-10 Julia Koval <julia.koval@intel.com>
29539 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
29540 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
29541 (_mm_maskz_rcp14_ss): New intrinsics.
29542 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
29543 (__builtin_ia32_rcp14ss_mask): New builtins.
29544 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
29546 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
29548 PR tree-optimization/51513
29549 * tree-cfg.c (gimple_seq_unreachable_p): New function.
29550 (assert_unreachable_fallthru_edge_p): Use it.
29551 (group_case_labels_stmt): Likewise.
29552 * tree-cfg.h: Prototype it.
29553 * stmt.c: Include cfghooks.h and tree-cfg.h.
29554 (emit_case_dispatch_table) <gap_label>: New local variable.
29555 Use it to fill dispatch table gaps.
29556 Test for default_label before updating probabilities.
29557 (expand_case) <default_label>: Remove unneeded initialization.
29558 Test for unreachable default case statement and remove its edge.
29559 Set default_label accordingly.
29560 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
29562 2017-05-10 Carl Love <cel@us.ibm.com>
29564 * config/rs6000/rs6000-c: Add support for built-in functions
29565 vector signed char vec_neg (vector signed char)
29566 vector signed short int vec_neg (vector short int)
29567 vector signed int vec_neg (vector signed int)
29568 vector signed long long vec_neg (vector signed long long)
29569 vector float vec_neg (vector float)
29570 vector double vec_neg (vector double)
29571 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
29573 * config/rs6000/altivec.h: Add define for vec_neg
29574 * doc/extend.texi: Update the built-in documentation for the
29575 new built-in functions.
29577 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29579 PR tree-optimization/77644
29580 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
29582 2017-05-10 Nathan Sidwell <nathan@acm.org>
29584 * dumpfile.h (TDI_lang_all): New.
29585 (TDF_KIND): New. Renumber others
29586 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
29588 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
29590 (get_dump_file_name): Adjust suffix generation.
29591 (dump_enable_all): Use TDF_KIND.
29592 * doc/invoke.texi (-fdump-lang-all): Document.
29594 * dumpfile.h: Tabify.
29596 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
29599 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
29600 Move member access before delete.
29602 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
29604 * tree-inline.c (expand_call_inline): Split block at stmt
29607 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
29610 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
29611 are now unused after splitting mov{sf,sd}_hardfloat.
29612 (f32_lr2): Likewise.
29613 (f32_lm): Likewise.
29614 (f32_lm2): Likewise.
29615 (f32_li): Likewise.
29616 (f32_li2): Likewise.
29617 (f32_lv): Likewise.
29618 (f32_sr): Likewise.
29619 (f32_sr2): Likewise.
29620 (f32_sm): Likewise.
29621 (f32_sm2): Likewise.
29622 (f32_si): Likewise.
29623 (f32_si2): Likewise.
29624 (f32_sv): Likewise.
29625 (f32_dm): Likewise.
29626 (f32_vsx): Likewise.
29627 (f32_av): Likewise.
29628 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
29629 For movsf, order stores so the VSX stores occur before the GPR
29630 store which encourages the register allocator to use a traditional
29631 FPR instead of a GPR. For movsd, order the stores so that the GPR
29632 store comes before the VSX stores to allow the power6 to work.
29633 This is due to the power6 not having a 32-bit integer store
29634 instruction from a FPR.
29635 (movsf_hardfloat): Likewise.
29636 (movsd_hardfloat): Likewise.
29638 2017-05-09 Martin Sebor <msebor@redhat.com>
29640 PR translation/80280
29641 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
29644 PR translation/80280
29645 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
29646 data member added in r247778.
29647 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
29649 2017-05-09 Nathan Sidwell <nathan@acm.org>
29651 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
29653 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
29656 2017-05-09 Marek Polacek <polacek@redhat.com>
29658 * doc/invoke.texi: Fix typo.
29660 2017-05-09 Richard Biener <rguenther@suse.de>
29662 * tree-vrp.c (vrp_val_is_max): Adjust comment.
29663 (vrp_val_is_min): Likewise.
29664 (set_value_range_to_value): Likewise.
29665 (set_value_range_to_nonnegative): Likewise.
29666 (gimple_assign_nonzero_p): Likewise.
29667 (gimple_stmt_nonzero_p): Likewise.
29668 (vrp_int_const_binop): Likewise. Remove unreachable case.
29669 (adjust_range_with_scev): Adjust comments.
29670 (compare_range_with_value): Likewise.
29671 (extract_range_from_phi_node): Likewise.
29672 (test_for_singularity): Likewise.
29674 2017-05-09 Richard Biener <rguenther@suse.de>
29676 * tree-vrp.c (get_single_symbol): Add assert that we don't
29677 get overflowed constants as invariant part.
29678 (compare_values_warnv): Add comment before the TREE_NO_WARNING
29679 checks. Use wi::cmp instead of recursing for integer constants.
29680 (compare_values): Just ignore whether we assumed undefined
29681 overflow instead of failing the compare.
29682 (extract_range_for_var_from_comparison_expr): Add comment before the
29683 TREE_NO_WARNING sets.
29684 (test_for_singularity): Likewise.
29685 (extract_range_from_comparison): Do not disable optimization
29686 when we assumed undefined overflow.
29687 (extract_range_basic): Remove init of unused var.
29689 2017-05-09 Richard Biener <rguenther@suse.de>
29691 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
29692 (extract_range_from_multiplicative_op_1): Adjust.
29693 (extract_range_from_binary_expr_1): Use int_const_binop.
29695 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
29698 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
29699 rs6000_store_data_bypass_p in seven define_bypass directives and
29700 in several comments.
29701 * config/rs6000/rs6000-protos.h: Add prototype for
29702 rs6000_store_data_bypass_p function.
29703 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
29704 function implements slightly different (rs6000-specific) semantics
29705 than store_data_bypass_p, returning false rather than aborting
29706 with assertion error when arguments do not satisfy the
29707 requirements of store data bypass.
29708 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
29709 rs6000_store_data_bypass_p.
29711 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
29713 * config/xtensa/xtensa-protos.h
29714 (xtensa_initial_elimination_offset): New declaration.
29715 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
29716 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
29717 macro definition, add case for FRAME_POINTER_REGNUM when
29718 FRAME_GROWS_DOWNWARD.
29719 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
29720 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
29721 xtensa_initial_elimination_offset.
29723 2017-05-08 Nathan Sidwell <nathan@acm.org>
29725 * doc/invoke.texi: Alphabetize -fdump options.
29727 2017-05-08 Martin Sebor <msebor@redhat.com>
29729 PR translation/80280
29730 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
29732 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
29734 * target.def (compute_frame_layout): New optional target hook.
29735 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
29736 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
29737 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
29739 * reload1.c (verify_initial_elim_offsets): Likewise.
29740 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
29741 (use_simple_return_p): Call arm_compute_frame_layout if needed.
29742 (arm_get_frame_offsets): Split up into this ...
29743 (arm_compute_frame_layout): ... and this function.
29745 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
29747 * config/aarch64/constraints.md (Usa): New constraint.
29748 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
29750 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
29752 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
29753 with_multilib_list after it has been checked.
29755 2017-05-08 Richard Biener <rguenther@suse.de>
29757 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
29758 (bitmap_set_subtract_values): Likewise.
29760 2017-05-08 Richard Biener <rguenther@suse.de>
29762 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
29763 (gimple_assign_nonzero): ... this and remove strict_overflow_p
29765 (gimple_stmt_nonzero_warnv_p): Rename to ...
29766 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
29768 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
29769 (extract_range_basic): Adjust, do not disable propagation on
29770 strict overflow sensitive simplification.
29771 (vrp_visit_cond_stmt): Likewise.
29773 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
29775 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
29776 body size unconditionally.
29778 2017-05-07 Jeff Law <law@redhat.com>
29781 2017-05-06 Jeff Law <law@redhat.com>
29782 PR tree-optimization/78496
29783 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29786 PR tree-optimization/78496
29787 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29788 (simplify_stmt_using_ranges): Call it.
29789 (vrp_dom_walker::before_dom_children): Extract equivalences
29790 from an ASSERT_EXPR with an equality comparison against a
29793 2017-05-06 Jeff Law <law@redhat.com>
29795 PR tree-optimization/78496
29796 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
29799 PR tree-optimization/78496
29800 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
29801 (simplify_stmt_using_ranges): Call it.
29802 (vrp_dom_walker::before_dom_children): Extract equivalences
29803 from an ASSERT_EXPR with an equality comparison against a
29806 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29808 * lra-constraints.c (lra_copy_reg_equiv): New function.
29809 (split_reg): Use it to copy equivalence information from the
29810 original register to the spill register.
29812 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
29814 PR rtl-optimization/75964
29815 * simplify-rtx.c (simplify_const_relational_operation): Remove
29816 invalid handling of comparisons of integer ABS.
29818 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
29820 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
29821 initialize to zero.
29822 (init_regs): Remove declaration.
29823 (function_arg_advance_32): Initialize error_p as boolean variable.
29825 2017-05-05 Nathan Sidwell <nathan@acm.org>
29827 * store-motion.c (remove_reachable_equiv_notes): Reformat long
29828 lines. Use for (;;).
29830 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29832 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
29833 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
29834 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
29835 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
29836 VF=2 that require versioning.
29838 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29840 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
29843 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29845 * diagnostic.h (diagnostic_override_option_index): Convert from
29846 macro to inline function.
29848 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29850 * diagnostic.c (last_module_changed_p): New function.
29851 (set_last_module): New function.
29852 (diagnostic_report_current_module): Convert macro usage to
29853 the above functions.
29854 * diagnostic.h (diagnostic_context::last_module): Strengthen
29855 from const line_map * to const line_map_ordinary *.
29856 (diagnostic_last_module_changed): Delete macro.
29857 (diagnostic_set_last_module): Delete macro.
29859 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29861 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
29862 with diagnostic_report_diagnostic.
29863 (diagnostic_n_impl_richloc): Likewise.
29864 * diagnostic.h (report_diagnostic): Delete macro.
29865 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
29866 with diagnostic_report_diagnostic.
29867 * substring-locations.c (format_warning_va): Likewise.
29869 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29871 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
29872 save/restor of format_spec. Move option-printing code to...
29873 (print_option_information): ...this new function, and
29874 reimplement by simply printing to the pretty_printer,
29875 rather than appending to the format string.
29877 2017-05-05 David Malcolm <dmalcolm@redhat.com>
29879 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
29880 handling logic into...
29881 (update_effective_level_from_pragmas): ...this new function.
29883 2017-05-04 Andrew Waterman <andrew@sifive.com>
29885 * config/riscv/riscv.opt (mstrict-align): New option.
29886 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
29887 (SLOW_UNALIGNED_ACCESS): Define.
29888 (riscv_slow_unaligned_access): Declare.
29889 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
29891 (riscv_slow_unaligned_access): New variable.
29892 (rocket_tune_info): Set slow_unaligned_access to true.
29893 (optimize_size_tune_info): Set slow_unaligned_access to false.
29894 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
29895 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
29896 (riscv_option_override): Set riscv_slow_unaligned_access.
29897 * doc/invoke.texi: Add -mstrict-align to RISC-V.
29899 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
29901 * config/riscv/riscv.md: Unify indentation.
29903 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29908 * config/rs6000/rs6000.md (u code attribute): Add FIX and
29910 (extendsi<mode>2): Add support for doing sign extension via
29911 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
29912 don't have ISA 3.0 instructions.
29913 (extendsi<mode>2 splitter): Likewise.
29914 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
29915 generate the normal insns since SImode can now go in vector
29916 registers. Disallow the special UNSPECs needed for previous
29917 machines to hide SImode being used. Add new insns
29918 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
29919 (fix_trunc<mode>si2_stfiwx): Likewise.
29920 (fix_trunc<mode>si2_internal): Likewise.
29921 (fixuns_trunc<mode>si2): Likewise.
29922 (fixuns_trunc<mode>si2_stfiwx): Likewise.
29923 (fctiw<u>z_<mode>_smallint): Likewise.
29924 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
29925 of floating point to 32-bit integer from doing a direct move to
29926 the GPR registers to do a store.
29927 (fctiwz_<mode>): Break long line.
29929 2017-05-05 Bin Cheng <bin.cheng@arm.com>
29931 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
29932 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
29933 (addr_list, addr_offset_valid_p): New.
29934 (split_address_groups): Check offset validity with above function.
29935 (gt-tree-ssa-loop-ivopts.h): Include header file.
29937 2017-05-05 Nathan Sidwell <nathan@acm.org>
29939 * config.gcc (arm*-*-*): Add missing 'fi'.
29941 2017-05-05 Steve Ellcey <sellcey@cavium.com>
29943 * doc/invoke.texi (-fopt-info): Explicitly say order of options
29944 included in -fopt-info does not matter.
29945 * doc/optinfo.texi (-fopt-info): Fix description of default
29946 behavour. Explicitly say order of options included in -fopt-info
29949 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29951 * config.gcc: Allow combinations of aprofile and rmprofile values for
29952 --with-multilib-list.
29953 * config/arm/t-multilib: New file.
29954 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
29955 variables. Remove setting of ISA and floating-point ABI in
29956 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
29957 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
29958 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
29959 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
29961 * config/arm/t-rmprofile: Likewise except for the matches changes.
29962 * doc/install.texi (--with-multilib-list): Document the combination of
29963 aprofile and rmprofile values and warn about pitfalls in doing that.
29965 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
29967 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
29968 (movdi_aarch64): Likewise.
29970 2017-05-05 Jakub Jelinek <jakub@redhat.com>
29972 PR tree-optimization/80632
29973 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
29975 (build_arrays): Initialize it for virtual phis.
29976 (fix_phi_nodes): Use it for virtual phis.
29978 PR tree-optimization/80558
29979 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
29980 [x, y] op z into [x op, y op z] for op & or | if conditions
29983 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
29984 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
29987 * config/arm/arm.md (use_literal_pool): Remove.
29988 (64-bit immediate split): No longer takes cost into consideration
29989 if arm_disable_literal_pool is enabled.
29990 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
29991 used when arm_disable_literal_pool is enabled.
29992 (arm_max_const_double_inline_cost): Remove use of
29993 arm_disable_literal_pool.
29994 (push_minipool_fix): Add assert.
29995 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
29996 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
29997 (no_literal_pool_sf_immediate): New.
29999 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30001 PR tree-optimization/80613
30002 * tree-ssa-dce.c (propagate_necessity): Remove cases for
30003 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
30005 2017-05-05 Richard Biener <rguenther@suse.de>
30007 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
30009 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
30011 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
30012 of this flag from insn conditions due to removal from r247495.
30014 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
30016 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
30018 (arm_early_store_addr_dep_ptr): Likewise.
30019 * config/arm/aarch-common-protos.h
30020 (arm_early_load_addr_dep_ptr): Add prototype.
30021 (arm_early_store_addr_dep_ptr): Likewise.
30022 * config/arm/cortex-a53.md: Add new bypasses.
30024 2017-05-05 Jakub Jelinek <jakub@redhat.com>
30026 * tree.c (next_type_uid): Change type to unsigned.
30027 (type_hash_canon): Decrement back next_type_uid if
30028 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
30029 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
30032 2017-05-04 Martin Sebor <msebor@redhat.com>
30034 * builtins.c: Fix a trivial typo in a comment.
30036 PR middle-end/79234
30037 * builtins.c (check_sizes): Adjust to handle reading past the end.
30038 Avoid printing excessive upper bound of ranges. Use %E to print
30039 tree nodes instead of converting them to %wu.
30040 (expand_builtin_memchr): New function.
30041 (compute_dest_size): Rename...
30042 (compute_objsize): ...to this.
30043 (expand_builtin_memcpy): Adjust.
30044 (expand_builtin_mempcpy): Adjust.
30045 (expand_builtin_strcat): Adjust.
30046 (expand_builtin_strcpy): Adjust.
30047 (check_strncat_sizes): Adjust.
30048 (expand_builtin_strncat): Adjust.
30049 (expand_builtin_strncpy): Adjust and simplify.
30050 (expand_builtin_memset): Adjust.
30051 (expand_builtin_bzero): Adjust.
30052 (expand_builtin_memcmp): Adjust.
30053 (expand_builtin): Handle memcmp.
30054 (maybe_emit_chk_warning): Check strncat just once.
30056 2017-05-04 Martin Sebor <msebor@redhat.com>
30058 PR preprocessor/79214
30059 PR middle-end/79222
30060 PR middle-end/79223
30061 * builtins.c (check_sizes): Add inlining context and issue
30062 warnings even when -Wno-system-headers is set.
30063 (check_strncat_sizes): Same.
30064 (expand_builtin_strncat): Same.
30065 (expand_builtin_memmove): New function.
30066 (expand_builtin_stpncpy): Same.
30067 (expand_builtin): Handle memmove and stpncpy.
30069 2017-05-04 Bin Cheng <bin.cheng@arm.com>
30071 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
30072 which is not used any more.
30074 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30076 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
30078 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30080 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
30081 (cortexa53_tunings): Likewise.
30082 (cortexa57_tunings): Likewise.
30083 (cortexa72_tunings): Likewise.
30084 (cortexa73_tunings): Likewise.
30086 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30088 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
30089 Set loop alignment to 8.
30091 2017-05-04 Martin Sebor <msebor@redhat.com>
30093 PR translation/80280
30094 * builtins.c (expand_builtin_object_size): Add missing quoting to
30095 %D and like directives.
30096 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
30097 (hsa_type_for_tree_type): Same.
30098 (verify_function_arguments): Same.
30099 * symtab.c (symbol_table::change_decl_assembler_name): Same.
30100 * varasm.c (get_section): Same.
30103 2017-05-04 Martin Sebor <msebor@redhat.com>
30105 PR translation/80280
30106 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
30108 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
30110 * config/aarch64/aarch64.c (generic_addrcost_table):
30111 Change HI/TI mode setting.
30113 2017-05-04 Martin Jambor <mjambor@suse.cz>
30115 PR tree-optimization/80622
30116 * tree-sra.c (comes_initialized_p): New function.
30117 (build_accesses_from_assign): Only set write lazily when
30118 comes_initialized_p is false.
30119 (analyze_access_subtree): Use comes_initialized_p.
30120 (propagate_subaccesses_across_link): Assert !comes_initialized_p
30121 instead of testing for PARM_DECL.
30123 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30125 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
30126 constraint on operand 0 to allow more general addressing modes.
30127 Adjust output template.
30128 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
30130 * config/aarch64/aarch64-protos.h
30131 (aarch64_address_valid_for_prefetch_p): Declare prototype.
30132 * config/aarch64/constraints.md (Dp): New address constraint.
30133 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
30136 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
30138 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
30139 update use of estimate_ipcp_clone_size_and_time.
30140 (estimate_local_effects): Update use of
30141 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
30142 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
30143 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
30144 Return nonspecialized time.
30146 2017-05-04 Richard Biener <rguenther@suse.de>
30148 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
30149 for the last VUSE which def dominates the PHI. Directly call
30151 (get_continuation_for_phi_1): Remove.
30153 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
30155 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
30156 to explain the use of truncating division. Cap the number of
30157 iterations to the maximum given by nb_iterations_upper_bound,
30160 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
30162 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
30163 * configure: Regenerate.
30164 * config.in: Regenerate.
30165 * config/i386/driver-mingw32.c: new file.
30166 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
30167 * config.host: Link driver-mingw32.o on MinGW host.
30168 * doc/install.texi: Document new --enable-mingw-wildcard configure
30171 2017-05-04 Marek Polacek <polacek@redhat.com>
30173 PR tree-optimization/80612
30174 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
30176 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30177 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
30179 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
30180 (movt splitter): Likewise.
30181 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
30182 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
30183 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
30184 block for Thumb-1 with MOVT.
30185 (thumb2_legitimate_address_p): Move code block ...
30186 (can_avoid_literal_pool_for_label_p): ... into this new function.
30187 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
30189 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
30190 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
30191 "M-profile targets with the MOVT instruction".
30193 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
30195 * config/arm/arm-builtins.c (arm_init_builtins): Rename
30196 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
30197 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
30199 2017-05-04 Martin Liska <mliska@suse.cz>
30201 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
30202 variable cond_code.
30204 2017-05-04 Richard Biener <rguenther@suse.de>
30206 * tree.c (array_at_struct_end_p): Handle arrays at struct
30207 end with flexarrays more conservatively. Refactor and treat
30208 arrays of arrays or aggregates more strict. Fix
30209 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
30210 * tree.c (array_at_struct_end_p): Adjust prototype.
30211 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
30212 * gimple-fold.c (get_range_strlen): Likewise.
30213 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
30215 2017-05-04 Richard Biener <rguenther@suse.de>
30217 PR tree-optimization/31130
30218 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
30220 (supports_overflow_infinity): Likewise.
30221 (is_negative_overflow_infinity): Likewise.
30222 (is_positive_overflow_infinity): Likewise.
30223 (is_overflow_infinity): Likewise.
30224 (stmt_overflow_infinity): Likewise.
30225 (overflow_infinity_range_p): Likewise.
30226 (usable_range_p): Remove as always returning true.
30227 (make_overflow_infinity): Remove.
30228 (negative_overflow_infinity): Likewise.
30229 (positive_overflow_infinity): Likewise.
30230 (avoid_overflow_infinity): Likewise.
30231 (set_value_range): Adjust accordingly.
30232 (set_value_range_to_nonnegative): Likewise, remove now unused
30233 overflow_infinity arg.
30234 (vrp_operand_equal_p): Adjust.
30235 (update_value_range): Likewise.
30236 (range_int_cst_singleton_p): Likewise.
30237 (operand_less_p): Likewise.
30238 (compare_values_warnv): Likewise.
30239 (extract_range_for_var_from_comparison_expr): Likewise.
30240 (vrp_int_const_binop): Likewise.
30241 (zero_nonzero_bits_from_vr): Likewise.
30242 (extract_range_from_multiplicative_op_1): Likewise.
30243 (extract_range_from_binary_expr_1): Likewise.
30244 (extract_range_from_unary_expr): Likewise.
30245 (extract_range_from_comparison): Likewise.
30246 (extract_range_basic): Likewise.
30247 (adjust_range_with_scev): Likewise.
30248 (compare_ranges): Likewise.
30249 (compare_range_with_value): Likewise.
30250 (dump_value_range): Likewise.
30251 (test_for_singularity): Likewise, remove strict_overflow_p parameter
30253 (simplify_cond_using_ranges): Adjust.
30255 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30257 * brig-builtins.def: Added a builtin for class_f64.
30258 * builtin-types.def: Added a builtin type needed by class_f64.
30260 2017-05-03 Jason Merrill <jason@redhat.com>
30262 * timevar.def: Add TV_CONSTEXPR.
30264 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30266 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
30268 2017-05-03 Martin Jambor <mjambor@suse.cz>
30270 * ipa-prop.c (ipa_update_after_lto_read): Removed.
30271 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
30272 * ipa-cp.c (ipcp_propagate_stage): Do not call
30273 ipa_update_after_lto_read.
30274 * ipa-inline.c (ipa_inline): Likewise.
30276 2017-05-03 Martin Jambor <mjambor@suse.cz>
30278 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
30279 tag. Added a default constructor and a destructor.
30280 (ipa_edge_args_sum_t): New class;
30281 (ipa_edge_args_sum): Declare.
30282 (ipa_edge_args_vector): Remove declaration.
30283 (IPA_EDGE_REF): Use ipa_edge_args_sum.
30284 (ipa_free_edge_args_substructures): Remove declaration.
30285 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
30286 (ipa_edge_args_info_available_for_edge_p): Likewise.
30287 * ipa-prop.c (ipa_edge_args_vector): Removed.
30288 (edge_removal_hook_holder): Likewise.
30289 (edge_duplication_hook_holder): Likewise.
30290 (ipa_edge_args_sum): New variable.
30291 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
30292 ipa_edge_args_vector.
30293 (ipa_free_edge_args_substructures): Likewise.
30294 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
30295 ipa_edge_args_vector.
30296 (ipa_edge_removal_hook): Turned into method
30297 ipa_edge_args_sum_t::remove.
30298 (ipa_edge_duplication_hook): Turned into method
30299 ipa_edge_args_sum_t::duplicate.
30300 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
30301 registering edge hooks.
30302 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
30303 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
30304 ipa_edge_args_sum instead of ipa_edge_args_vector.
30305 * ipa-profile.c (ipa_profile): Likewise.
30307 2017-05-03 Martin Jambor <mjambor@suse.cz>
30309 * symbol-summary.h (function_summary): New method exists.
30310 (function_summary::symtab_removal): Deallocate through release.
30311 (call_summary): New class.
30312 (gt_ggc_mx): New overload.
30313 (gt_pch_nx): Likewise.
30314 (gt_pch_nx): Likewise.
30316 2017-05-03 Jeff Law <law@redhat.com>
30318 PR tree-optimization/78496
30319 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
30320 from simplify_cond_using_ranges. Split off code to walk
30321 backwards through casts into ...
30322 (simplify_cond_using_ranges_2): New function.
30323 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
30324 (execute_vrp): After identifying jump threads, call
30325 simplify_cond_using_ranges_2.
30327 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
30330 * ipa-inline.h (inline_summary): Add ctor.
30331 (create_ggc): Do not use ggc_cleared_alloc.
30333 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
30334 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
30336 * gcc.c (handle_braces): Support escaping in switch matching
30338 * doc/invoke.texi (Spec Files): Document it.
30339 Remove superfluous @code markup in items.
30341 2017-05-03 David Malcolm <dmalcolm@redhat.com>
30343 * diagnostic-show-locus.c (struct column_range): New struct.
30344 (get_affected_columns): New function.
30345 (get_printed_columns): New function.
30346 (struct correction): New struct.
30347 (correction::ensure_capacity): New function.
30348 (correction::ensure_terminated): New function.
30349 (struct line_corrections): New struct.
30350 (line_corrections::~line_corrections): New dtor.
30351 (line_corrections::add_hint): New function.
30352 (layout::print_trailing_fixits): Reimplement in terms of the new
30354 (selftest::test_overlapped_fixit_printing): New function.
30355 (selftest::diagnostic_show_locus_c_tests): Call it.
30357 2017-05-03 Nathan Sidwell <nathan@acm.org>
30359 Canonicalize canonical type hashing
30360 * tree.h (type_hash_canon_hash): Declare.
30361 * tree.c (type_hash_list, attribute_hash_list): Move into
30362 type_hash_canon_hash.
30363 (build_type_attribute_qual_variant): Break out hash code calc into
30364 type_hash_canon_hash.
30365 (type_hash_canon_hash): New. Generic type hash computation.
30366 (build_range_type_1, build_array_type_1, build_function_type,
30367 build_method_type_directly, build_offset_type, build_complex_type,
30368 make_vector_type): Call it.
30370 2017-05-03 Richard Biener <rguenther@suse.de>
30372 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
30373 When all DRs have unknown misaligned do not always peel
30374 when there is a store but apply the same costing model as if
30375 there were only loads.
30377 2017-05-03 Richard Biener <rguenther@suse.de>
30380 PR tree-optimization/80492
30381 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
30382 compare_base_decls returning dont-know properly.
30384 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
30386 * config/arm/iterators.md (CCSI): New mode iterator.
30387 (arch): New mode attribute.
30388 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
30389 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
30390 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
30391 code iterator for success result mode.
30392 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
30393 the corresponding new insn generators.
30395 2017-05-03 Bin Cheng <bin.cheng@arm.com>
30398 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30399 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30401 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
30403 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
30404 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
30405 (DDR_A): Wrap DDR argument in brackets.
30406 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
30407 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
30408 (DDR_REVERSED_P): Likewise.
30410 2017-05-03 Jakub Jelinek <jakub@redhat.com>
30412 PR tree-optimization/79472
30413 * tree-switch-conversion.c (struct switch_conv_info): Add
30414 contiguous_range and default_case_nonstandard fields.
30415 (collect_switch_conv_info): Compute contiguous_range and
30416 default_case_nonstandard fields, don't clear final_bb if
30417 contiguous_range and only the default case doesn't have the required
30419 (check_all_empty_except_final): Set default_case_nonstandard instead
30420 of failing if contiguous_range and the default case doesn't have empty
30422 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
30423 and only the default case doesn't have the required constants. Skip
30425 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
30426 if default_case_nonstandard.
30427 (build_constructors): Build constant 1 just once. Assert that default
30428 values aren't inserted in between cases if contiguous_range. Skip
30430 (build_arrays): Skip virtual phis.
30431 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
30432 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
30433 Handle virtual phis.
30434 (gen_inbound_check): Handle default_case_nonstandard case.
30435 (process_switch): Adjust check_final_bb caller. Call
30436 gather_default_values with the first non-default case instead of
30437 default case if default_case_nonstandard.
30439 2017-05-02 Nathan Sidwell <nathan@acm.org>
30441 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
30442 check. Fix formatting.
30444 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
30446 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
30447 errors when comparing specialized and unspecialized times.
30449 2017-05-02 David Malcolm <dmalcolm@redhat.com>
30451 * diagnostic-show-locus.c
30452 (layout::should_print_annotation_line_p): Make private.
30453 (layout::print_annotation_line): Make private.
30454 (layout::annotation_line_showed_range_p): Make private.
30455 (layout::show_ruler): Make private.
30456 (layout::print_source_line): Make private. Pass in line and
30457 line_width, rather than calling location_get_source_line. Drop
30459 (layout::print_leading_fixits): New method.
30460 (layout::print_any_fixits): Rename to...
30461 (layout::print_trailing_fixits): ...this, and make private.
30462 Don't print newline fixits.
30463 (diagnostic_show_locus): Move logic for printing one row into...
30464 (layout::print_line): ...this new function. Move the
30465 location_get_source_line call and error-handling from
30466 print_source_line to here. Call print_leading_fixits, and rename
30467 print_any_fixits to print_trailing_fixits.
30468 (selftest::test_fixit_insert_containing_newline): Update now that
30469 newlines are partially supported.
30470 (selftest::test_fixit_insert_containing_newline_2): New test.
30471 (selftest::test_fixit_replace_containing_newline): Update comments.
30472 (selftest::diagnostic_show_locus_c_tests): Call the new test.
30473 * edit-context.c (class added_line): New class.
30474 (class edited_line): Describe newline handling in comment.
30475 (edited_line::actually_edited_p): New method.
30476 (edited_line::print_content): Delete redundant decl.
30477 (edited_line::m_predecessors): New field.
30478 (edited_file::print_content): Call edited_line::print_content.
30479 (edited_file::print_diff): Update to support newlines.
30480 (edited_file::print_diff_hunk): Likewise.
30481 (edited_file::print_run_of_changed_lines): New function.
30482 (edited_file::print_diff_line): Convert to...
30483 (print_diff_line): ...this.
30484 (edited_file::get_effective_line_count): New function.
30485 (edited_line::edited_line): Initialize new field m_predecessors.
30486 (edited_line::~edited_line): Clean up m_predecessors.
30487 (edited_line::apply_fixit): Handle newlines.
30488 (edited_line::get_effective_line_count): New function.
30489 (edited_line::print_content): New function.
30490 (edited_line::print_diff_lines): New function.
30491 (selftest::test_applying_fixits_insert_containing_newline): New
30493 (selftest::test_applying_fixits_replace_containing_newline): New
30495 (selftest::insert_line): New function.
30496 (selftest::test_applying_fixits_multiple_lines): Add example of
30498 (selftest::edit_context_c_tests): Call the new tests.
30500 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30502 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
30503 parameter cand. Update dump information.
30504 (get_computation_cost): Update uses.
30506 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30508 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
30509 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
30510 (get_computation_at, rewrite_use_address): Update use of
30511 get_computation_aff.
30513 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30515 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
30516 (get_computation): Delete.
30517 (get_computation_cost): Implement like get_computation_cost_at.
30518 Use get_computation_at.
30519 (get_computation_cost_at): Delete.
30520 (rewrite_use_nonlinear_expr): Use get_computation_at.
30521 (rewrite_use_compare, remove_unused_ivs): Ditto.
30523 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30525 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
30527 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30529 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
30530 (ivopts_global_cost_for_size): Rename parameter and update uses.
30531 (iv_ca_recount_cost): Update uses.
30532 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
30533 candidates seperately in n_invs and n_cands.
30534 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
30536 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30538 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
30539 (find_inv_vars_cb): New.
30540 (find_depends): Renamed to ...
30541 (find_inv_vars): ... this.
30542 (add_candidate_1, force_var_cost): Call find_inv_vars.
30543 (split_address_cost, determine_group_iv_cost_cond): Ditto.
30545 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30547 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
30548 inv_vars. Add inv_exprs.
30549 (struct iv_cand): Rename depends_on to inv_vars.
30550 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
30551 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
30552 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
30553 (dump_cand): Dump inv_vars.
30554 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
30555 (record_invariant, find_depends, add_candidate_1): Ditto.
30556 (set_group_iv_cost, force_var_cost): Ditto.
30557 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
30558 (get_computation_cost_at, get_computation_cost): Ditto.
30559 (determine_group_iv_cost_generic): Ditto.
30560 (determine_group_iv_cost_address): Ditto.
30561 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
30562 (determine_group_iv_costs): Ditto.
30563 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
30564 (iv_ca_set_remove_invariants): Renamed to ...
30565 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
30566 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
30567 (iv_ca_set_add_invariants): Renamed to ...
30568 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
30569 (iv_ca_set_cp): Use iv_ca_set_add_invs.
30570 (iv_ca_has_deps): Support inv_vars and inv_exprs.
30571 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
30572 (create_new_ivs): Remove useless dump.
30574 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30576 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
30578 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
30579 (iv_ca_set_no_cp, create_new_iv): Ditto.
30581 2017-05-02 Bin Cheng <bin.cheng@arm.com>
30583 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
30585 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30587 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
30588 function tree_check2.
30590 2017-05-02 Martin Liska <mliska@suse.cz>
30592 * doc/gcov.texi: Add missing preposition.
30593 * gcov.c (function_info::function_info): Properly fill up
30594 all member variables.
30596 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30598 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
30600 2017-05-02 Tamar Christina <tamar.christina@arm.com>
30602 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
30604 2017-05-02 Martin Liska <mliska@suse.cz>
30607 * lto-streamer-in.c (lto_read_tree_1): Remove
30608 LTO_STREAMER_DEBUG.
30609 * lto-streamer.c (struct tree_hash_entry): Likewise.
30610 (struct tree_entry_hasher): Likewise.
30611 (tree_entry_hasher::hash): Likewise.
30612 (tree_entry_hasher::equal): Likewise.
30613 (lto_streamer_init): Likewise.
30614 (lto_orig_address_map): Likewise.
30615 (lto_orig_address_get): Likewise.
30616 (lto_orig_address_remove): Likewise.
30617 * lto-streamer.h: Likewise.
30618 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
30619 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
30621 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
30623 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
30624 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
30625 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
30626 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
30627 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
30628 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
30629 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
30630 (mm_maskz_sub_ss): New intrinsics.
30631 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30632 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
30633 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
30634 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
30635 (__builtin_ia32_subss_mask_round): New builtins.
30636 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
30637 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
30638 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
30640 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
30641 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
30643 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
30646 2017-05-02 Martin Jambor <mjambor@suse.cz>
30648 PR tree-optimization/78687
30649 * tree-sra.c (access): New field parent.
30650 (process_subtree_disqualification): New function.
30651 (disqualify_candidate): Call it.
30652 (build_accesses_from_assign): Reset write flag if creating an
30654 (build_access_subtree): Fill in parent field and also prpagate
30655 down grp_write flag.
30656 (create_artificial_child_access): New parameter set_grp_write, set
30657 grp_write to its value.
30658 (propagate_subaccesses_across_link): Also propagate grp_write flag
30660 (propagate_all_subaccesses): Push the closest parent back to work
30661 queue if add_access_to_work_queue returned true.
30663 2017-05-02 Richard Biener <rguenther@suse.de>
30665 * common.opt (fstrict-overflow): Alias negative to fwrapv.
30666 * doc/invoke.texi (fstrict-overflow): Remove all traces of
30667 -fstrict-overflow documentation.
30668 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
30669 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
30670 flag_strict_overflow.
30671 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
30672 * lto-opts.c (lto_write_options): Do not stream it.
30673 * lto-wrapper.c (merge_and_complain): Do not handle it.
30674 * opts.c (default_options_table): Do not set -fstrict-overflow.
30675 (finish_options): Likewise do not clear it when sanitizing.
30676 * simplify-rtx.c (simplify_const_relational_operation): Do not
30677 test flag_strict_overflow.
30679 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30681 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
30682 using enabled attribute.
30683 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
30684 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
30685 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
30686 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
30687 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
30688 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
30689 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
30690 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
30691 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
30692 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
30694 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
30696 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
30698 2017-05-02 Richard Biener <rguenther@suse.de>
30700 PR tree-optimization/80591
30702 2017-04-10 Richard Biener <rguenther@suse.de>
30704 * tree-ssa-structalias.c (find_func_aliases): Properly handle
30707 2017-05-02 Richard Biener <rguenther@suse.de>
30709 PR tree-optimization/80549
30710 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
30711 (cleanup_tree_cfg_noloop): Create forwarders to known loop
30712 headers if they do not have a preheader.
30714 2017-05-02 Martin Liska <mliska@suse.cz>
30717 * common.opt: Fix typo.
30718 * doc/invoke.texi: Likewise.
30720 2017-05-01 Jan Beulich <jbeulich@suse.com>
30722 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
30723 swapping, add (x,x,m,x,n) alternative.
30725 2017-05-01 Nathan Sidwell <nathan@acm.org>
30727 * calls.c (combine_pending_stack_adjustment_and_call): Remove
30728 unnecessary unadjusted_alignment check.
30730 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
30733 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
30735 * gimplify.c (gimplify_cilk_detach): New function.
30736 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
30737 * tree-core.h: Document EXPR_CILK_SPAWN.
30738 * tree.h (EXPR_CILK_SPAWN): Define.
30740 2017-05-01 David Malcolm <dmalcolm@redhat.com>
30742 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
30743 to use new fixit_hint representation, using the "replace" logic.
30744 (get_line_span_for_fixit_hint): Likewise.
30745 (layout::print_any_fixits): Likewise.
30746 (selftest::test_one_liner_many_fixits): Rename to...
30747 (selftest::test_one_liner_many_fixits_1): ...this, and update
30748 comment and expected output to reflect that the multiple fix-it
30749 hints are now consolidated into one insertion.
30750 (selftest::test_one_liner_many_fixits_2): New test.
30751 (selftest::test_diagnostic_show_locus_one_liner): Update for
30753 (selftest::test_fixit_consolidation): Update for fix-it API
30755 * diagnostic.c (print_parseable_fixits): Likewise.
30756 * edit-context.c (edited_line::m_line_events): Convert from
30757 auto_vec <line_event *> to auto_vec <line_event>.
30758 (class line_event): Convert from abstract base class to a concrete
30759 class, taking over the role of replace_event.
30760 (class insert_event): Delete.
30761 (class replace_event): Rename to class line_event. Convert to
30763 (edit_context::add_fixits): Reimplement.
30764 (edit_context::apply_insert): Delete.
30765 (edit_context::apply_replace): Rename to...
30766 (edit_context::apply_fixit): ...this. Convert to half-open range.
30767 (edited_file::apply_insert): Delete.
30768 (edited_file::apply_replace): Rename to...
30769 (edited_file::apply_fixit): ...this.
30770 (edited_line::~edited_line): Drop deletion of events.
30771 (edited_line::apply_insert): Delete.
30772 (edited_line::apply_replace): Rename to...
30773 (edited_line::apply_fixit): ...this. Convert to half-open range.
30774 Update for change to type of m_line_events.
30775 * edit-context.h (edit_context::apply_insert): Delete.
30776 (edit_context::apply_replace): Rename to...
30777 (edit_context::apply_fixit): ...this.
30779 2017-05-01 Martin Sebor <msebor@redhat.com>
30781 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
30784 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
30787 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
30788 __get_cpuid_max returns 0.
30789 (__get_cpuid_count): Ditto.
30791 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
30793 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
30794 replacement expression is another instance of one of its arguments.
30796 2017-05-01 Jakub Jelinek <jakub@redhat.com>
30799 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
30800 check for stack push/pop autoinc.
30801 * config/i386/i386.c (ix86_agi_dependent): Return false
30802 if the only reason why modified_in_p returned true is that
30803 addr is SP based and set_insn is a push or pop.
30805 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30807 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
30810 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
30813 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
30814 (account_size_time): Use two predicates - exec_pred and
30816 (evaluate_conditions_for_known_args): Compute both clause and
30818 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
30819 (inline_summary_t::duplicate): Update.
30820 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
30822 (compute_inline_parameters): Likewise.
30823 (estimate_edge_size_and_time): Update caluclation of time.
30824 (estimate_node_size_and_time): Compute both time and nonspecialized
30826 (estimate_ipcp_clone_size_and_time): Update.
30827 (inline_merge_summary): Update.
30828 (do_estimate_edge_time): Update.
30829 (do_estimate_edge_size): Update.
30830 (do_estimate_edge_hints): Update.
30831 (inline_read_section, inline_write_summary): Stream both new predicates.
30832 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
30834 (compute_inlined_call_time): Cleanup.
30835 (big_speedup_p): Update.
30836 (edge_badness): Update.
30837 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
30838 (size_time_entry): Replace predicate by exec_predicate and
30839 nonconst_predicate.
30840 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
30841 (estimate_edge_time): Return also nonspec_time.
30842 (reset_edge_growth_cache): Update.
30844 2017-04-29 Jakub Jelinek <jakub@redhat.com>
30846 PR rtl-optimization/80491
30847 * ifcvt.c (noce_process_if_block): When looking for x setter
30848 with missing else_bb, don't check only the insn right before
30849 cond_earliest, but look for the last insn that x is modified in
30850 within the same bb.
30852 PR rtl-optimization/80491
30853 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
30855 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
30857 PR tree-optimization/80487
30858 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
30860 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30862 PR tree-optimization/79697
30863 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
30864 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
30865 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
30867 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
30868 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
30870 2017-04-28 Martin Sebor <msebor@redhat.com>
30872 PR tree-optimization/80523
30873 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
30874 (init_target_to_host_charmap, target_to_host, target_strtol10): New
30876 (maybe_warn, format_directive, parse_directive): Use new functions.
30877 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
30879 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
30881 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
30883 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
30885 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
30886 target_header_dir): Set correctly.
30887 * configure: Regenerated.
30888 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30889 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30890 instead of SYSTEM_HEADER_DIR.
30892 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
30894 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
30895 (estimate_local_effects): Likewise.
30896 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
30897 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
30898 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
30899 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
30900 do_estimate_edge_time, estimate_edge_time): Likewise.
30901 * ipa-inline-analysis.c (estimate_node_size_and_time,
30902 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
30903 (estimate_time_after_inlining): Remove.
30905 2017-04-28 Martin Liska <mliska@suse.cz>
30907 * doc/gcov.texi: Enhance documentation of gcov.
30909 2017-04-28 Martin Liska <mliska@suse.cz>
30911 * doc/gcov.texi: Sort options in alphabetic order.
30912 * doc/gcov-dump.texi: Likewise.
30913 * doc/gcov-tool.texi: Likewise.
30914 * gcov.c (print_usage): Likewise.
30915 * gcov-dump.c (print_usage): Likewise.
30916 * gcov-tool.c (print_merge_usage_message): Likewise.
30917 (print_rewrite_usage_message): Likewise.
30918 (print_overlap_usage_message): Likewise.
30920 2017-04-28 Martin Liska <mliska@suse.cz>
30922 PR gcov-profile/53915
30923 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
30925 2017-04-28 Martin Liska <mliska@suse.cz>
30927 PR gcov-profile/79891
30928 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
30929 is marked by compiler as living on a line.
30930 (get_cycles_count): Remove usage of the union.
30931 (output_intermediate_file): Likewise.
30932 (find_source): Fix GNU coding style.
30933 (accumulate_line_counts): Remove old non-all block mode.
30934 (output_lines): Remove usage of the union.
30935 * profile.c (output_location): Include all BBs, even if
30936 belonging to a same line (and file) as a previous BB.
30938 2017-04-28 Martin Liska <mliska@suse.cz>
30940 * gcov.c (process_args): Handle new argument 'w'.
30941 (read_graph_file): Assign ID to BBs.
30942 (output_branch_count): Display BB # if verbose flag is set.
30943 (output_lines): Likewise for arcs.
30944 (print_usage): Add '--verbose' option help.
30945 * doc/gcov.texi: Document --verbose (-w) option.
30947 2017-04-28 Martin Liska <mliska@suse.cz>
30949 * gcov.c (struct block_location_info): New struct.
30950 (process_file): Fill up the new structure.
30951 (read_graph_file): Replace usage of encoding by the newly added
30953 (add_line_counts): Likewise.
30954 (accumulate_line_counts): Remove usage of the union.
30955 (function_info::function_info): New function.
30956 (function_info::~function_info): Likewise.
30957 (process_file): Call delete instead of release_function.
30958 (release_function): Release the function.
30959 (release_structures): Call delete instead of release_function.
30960 (solve_flow_graph): Replace usage of num_blocks.
30961 (find_exception_blocks): Likewise.
30962 (output_lines): Fix GNU coding style.
30964 2017-04-28 Martin Liska <mliska@suse.cz>
30967 * coverage.c (coverage_remove_note_file): New function.
30968 * coverage.h: Declare the function.
30969 * toplev.c (finalize): Clean if an error has been seen.
30971 2017-04-28 Martin Liska <mliska@suse.cz>
30973 PR gcov-profile/80031
30974 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
30975 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
30976 * gcov.c (read_graph_file): Read just number of blocks.
30977 * profile.c (branch_prob): Do not stream 0 flags per a basic
30980 2017-04-28 Martin Liska <mliska@suse.cz>
30982 * gcov-dump.c (tag_*): Add new argument to declarations.
30983 (dump_gcov_file): Likewise.
30984 (tag_blocks): Add and use new argument depth.
30985 (tag_arcs): Likewise.
30986 (tag_lines): Likewise.
30987 (tag_counters): Likewise.
30988 (tag_summary): Likewise.
30989 (dump_working_sets): Use depth to do a proper indentation.
30991 2017-04-28 Jakub Jelinek <jakub@redhat.com>
30994 * cgraph.h (symtab_node::debug_symtab): No longer inline.
30995 * symtab.c (symtab_node::debug_symtab): Move definition here.
30997 2017-04-28 Richard Biener <rguenther@suse.de>
30999 * lto-streamer.h (LTO_major_version): Bump to 7.
31001 2017-04-28 Richard Biener <rguenther@suse.de>
31003 * tree-vrp.c (assert_info): New struct.
31004 (add_assert_info): New helper.
31005 (register_edge_assert_for_2): Refactor to add asserts to a vector
31007 (register_edge_assert_for_1): Likewise.
31008 (register_edge_assert_for): Likewise.
31009 (finish_register_edge_assert_for): New helper actually registering
31010 asserts where live on edge.
31011 (find_conditional_asserts): Adjust.
31012 (find_switch_asserts): Likewise.
31013 (evrp_dom_walker::try_find_new_range): Generalize.
31014 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
31016 2017-04-27 Marek Polacek <polacek@redhat.com>
31019 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
31020 arg10 and arg11 to itype.
31022 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
31024 * doc/extend.texi (Object Size Checking): Improve grammar.
31026 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
31029 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
31030 that the logic for permitting reciprocal estimates matches that
31033 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31036 * tree.c (type_cache_hasher::equal): Only compare
31037 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
31038 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
31039 non-aggregate element types.
31040 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
31041 about the flag on ARRAY_TYPEs in the comment, formatting fix.
31043 2017-04-27 Richard Biener <rguenther@suse.de>
31045 PR middle-end/80533
31046 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
31047 stripping ARRAY_REFs from MEM_EXPR make sure we're not
31048 keeping a reference to a trailing array.
31050 2017-04-27 Richard Biener <rguenther@suse.de>
31052 PR middle-end/80539
31053 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
31054 being in loop-closed SSA form conservatively.
31055 (chrec_fold_multiply_poly_poly): Likewise.
31057 2017-04-27 Tamar Christina <tamar.christina@arm.com>
31059 PR middle-end/79665
31060 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
31061 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
31063 2017-04-27 Jakub Jelinek <jakub@redhat.com>
31066 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
31067 (aarch64_function_arg_alignment): Return unsigned int again, but still
31068 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
31069 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31070 Don't emit -Wpsabi note.
31071 (aarch64_function_arg_boundary): Likewise.
31072 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31075 2017-04-26 Nathan Sidwell <nathan@acm.org>
31077 * tree.h (crc32_unsigned_n): Declare.
31078 (crc32_unsigned, crc32_unsigned): Make inline.
31079 * tree.c (crc32_unsigned_bits): Replace with ...
31080 (crc32_unsigned_n): ... this.
31081 (crc32_unsigned, crc32_byte): Remove.
31082 (crc32_string): Remove unnecessary braces.
31084 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31086 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
31087 * ipa-inline-analysis.c (MAX_TIME): Remove.
31088 (account_size_time): Use sreal for time.
31089 (dump_inline_summary): Update.
31090 (estimate_function_body_sizes): Update.
31091 (estimate_edge_size_and_time): Update.
31092 (estimate_calls_size_and_time): Update.
31093 (estimate_node_size_and_time): Update.
31094 (inline_merge_summary): Update.
31095 (inline_update_overall_summary): Update.
31096 (estimate_time_after_inlining): Update.
31097 (inline_read_section): Update.
31098 (inline_write_summary): Update.
31099 * ipa-inline.c (compute_uninlined_call_time): Update.
31100 (compute_inlined_call_time): Update.
31101 (recursive_inlining): Update.
31102 (inline_small_functions): Update.
31103 (dump_overall_stats): Update.
31104 * ipa-inline.h: Include sreal.h.
31105 (size_time_entry): Turn time to sreal.
31106 (inline_summary): Turn self_time nad time to sreal.
31108 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
31110 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
31112 (sreal::stream_out, sreal::stream_in): New.
31113 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
31115 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31117 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
31120 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
31123 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
31125 (dimode_scalar_chain::compute_convert_gain): Ditto.
31126 (dimode_scalar_chain::make_vector_copies): Ditto.
31127 (dimode_scalar_chain::convert_reg): Ditto.
31128 (dimode_scalar_chain::convert_insn): Ditto.
31129 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
31130 (VI248_AVX512BW_1): New mode iterator.
31131 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
31132 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
31135 2017-04-25 Martin Sebor <msebor@redhat.com>
31137 PR tree-optimization/80497
31138 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
31139 constants are representable in HOST_WIDE_INT.
31140 (parse_directive): Ditto.
31142 2017-04-25 Martin Sebor <msebor@redhat.com>
31145 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
31146 (new_zero_array): Adjust signature.
31147 (dom_info::dom_init): Used unsigned rather that size_t.
31148 (dom_info::dom_info): Same.
31150 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31151 Jakub Jelinek <jakub@redhat.com>
31154 * config/arm/arm.c: Include gimple.h.
31155 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31156 returns negative, increment ncrn only if it returned positive.
31157 (arm_needs_doubleword_align): Return int instead of bool,
31158 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
31159 members, but if there is any such non-FIELD_DECL
31160 > PARM_BOUNDARY aligned decl, return -1 instead of false.
31161 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
31162 returns negative, increment nregs only if it returned positive.
31163 (arm_setup_incoming_varargs): Likewise.
31164 (arm_function_arg_boundary): Emit -Wpsabi note if
31165 arm_needs_doubleword_align returns negative, return
31166 DOUBLEWORD_ALIGNMENT only if it returned positive.
31168 2017-04-25 Marek Polacek <polacek@redhat.com>
31171 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31172 first argument to type.
31174 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
31177 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
31178 type checks to test for compatibility instead of equality.
31180 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
31181 Jakub Jelinek <jakub@redhat.com>
31184 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
31186 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
31187 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
31188 the alignment computation, but return their maximum in warn_alignment.
31189 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
31190 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
31192 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
31193 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
31196 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31198 * config/arc/simdext.md (dmpyh): Fix typo.
31200 2017-04-25 Richard Biener <rguenther@suse.de>
31202 PR tree-optimization/80492
31203 * alias.c (compare_base_decls): Handle registers with asm
31204 specification conservatively.
31205 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
31206 compare_base_decls returning dont-know properly.
31208 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31210 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
31211 (legitimate_offset_address_p): New function.
31212 (arc_legitimate_address_p): Use above function.
31214 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31216 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
31218 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31220 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
31221 ACCH registers whenever they are available.
31223 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31225 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
31226 double regs fix when not used.
31228 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31230 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
31232 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
31233 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
31235 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31237 * config/arc/arc.c (arc_output_addsi): Check for h-register class
31238 when emitting short ADD instructions.
31240 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
31242 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
31244 (cmpsi_cc_c_insn): Likewise.
31245 (cbranchsi4_scratch): Compute proper instruction length using
31246 compact_hreg_operand.
31247 * config/arc/predicates.md (compact_hreg_operand): New predicate.
31249 2017-04-25 Richard Biener <rguenther@suse.de>
31251 PR middle-end/80509
31252 * passes.c (pass_manager::pass_manager): Initialize
31253 m_name_to_pass_map.
31255 2017-04-25 Richard Biener <rguenther@suse.de>
31257 PR tree-optimization/79201
31258 * tree-ssa-sink.c (statement_sink_location): Handle calls.
31260 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31263 * config/s390/vector.md: Split MEM->GPR vector moves for
31264 non-s_operand addresses.
31266 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31269 * config/s390/predicates.md (reload_const_wide_int_operand): New
31271 * config/s390/s390.md ("movti"): Remove d/P alternative.
31272 ("movti_bigconst"): New pattern definition.
31274 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31277 * s390-protos.h (s390_expand_cs_hqi): Removed.
31278 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
31279 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
31280 modes as well as CCZ1mode and CCZmode.
31281 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
31282 signature of s390_emit_compare_and_swap.
31283 (s390_expand_cs_hqi): Likewise, make static.
31284 (s390_expand_cs_tdsi): Generate an explicit compare before trying
31285 compare-and-swap, in some cases.
31286 (s390_expand_cs): Wrapper function.
31287 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
31289 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
31290 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
31291 patterns for small and large integers. Forbid symref memory operands.
31292 Move expander to s390.c. Require cc register.
31293 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
31294 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
31295 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
31296 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
31297 symref memory operands. Remove CC mode and call s390_match_ccmode
31299 ("atomic_exchange<mode>"): Allow and implement all integer modes.
31301 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31303 * config/s390/s390.md (define_peephole2): New peephole to help
31304 combining the load-and-test pattern with volatile memory.
31306 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
31308 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
31309 with CCZmode for TARGET_Z196.
31311 2017-04-25 Jakub Jelinek <jakub@redhat.com>
31313 PR rtl-optimization/80501
31314 * combine.c (make_compound_operation_int): Set subreg_code to SET
31315 even for AND with mask of the sign bit of mode.
31317 PR rtl-optimization/80500
31318 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
31319 sum's initial value.
31321 2017-04-25 Julian Brown <julian@codesourcery.com>
31322 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31324 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
31326 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
31328 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
31330 2017-04-25 Julian Brown <julian@codesourcery.com>
31331 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31333 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
31334 (thunderx2t99_sha): New Reservation.
31336 2017-04-25 Julian Brown <julian@codesourcery.com>
31337 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31339 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
31340 type for 1-element load.
31342 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
31344 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
31346 2017-04-24 Martin Jambor <mjambor@suse.cz>
31348 PR tree-optimization/80293
31349 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
31350 char arrays not totally scalarizable if it is false.
31351 (analyze_all_variable_accesses): Pass correct value in the new
31352 parameter. Add a statistics counter.
31354 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
31356 PR middle-end/79931
31357 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
31359 2017-04-24 Richard Biener <rguenther@suse.de>
31361 PR tree-optimization/80494
31362 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
31363 out for complex types.
31365 2017-04-24 Richard Biener <rguenther@suse.de>
31367 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
31368 * tree-ssa-sccvn.c (print_scc): Print SCC size.
31369 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
31370 (DFS): Adjust and never fail.
31371 (sccvn_dom_walker::fail): Remove.
31372 (sccvn_dom_walker::before_dom_children): Adjust.
31373 (run_scc_vn): Likewise and never fail.
31374 * tree-ssa-pre.c (pass_pre::execute): Adjust.
31375 (pass_fre::execute): Likewise.
31377 2017-04-24 Richard Biener <rguenther@suse.de>
31379 PR tree-optimization/79725
31380 * tree-ssa-sink.c (statement_sink_location): Return whether
31381 failure reason was zero uses. Move that check later.
31382 (sink_code_in_bb): Deal with zero uses by removing the stmt
31385 2017-04-24 Richard Biener <rguenther@suse.de>
31388 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
31389 pointer-based references.
31391 2017-04-24 Richard Biener <rguenther@suse.de>
31394 * pass_manager.h (pass_manager::operator new): Remove.
31395 (pass_manager::operator delete): Likewise.
31396 * passes.c (pass_manager::operator new): Remove.
31397 (pass_manager::operator delete): Likewise.
31398 (pass_manager::pass_manager): Zero individual pass members.
31400 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
31403 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
31404 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
31405 Check "XEXP (src, 1)" operand here.
31406 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
31407 Check "XEXP (src, 1)" operand here.
31408 (dimode_scalar_chain::make_vector_copies): Detect count register
31409 of a shift instruction. Zero extend count register from QImode
31410 to DImode to satisfy vector shift pattern count operand predicate.
31411 Substitute vector shift count operand with a DImode copy.
31412 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
31415 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
31417 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
31418 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31419 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
31420 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
31421 (UNSPEC_NOREX_MEM): Remove definition.
31423 2017-04-21 Richard Biener <rguenther@suse.de>
31425 PR tree-optimization/79547
31426 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
31427 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
31428 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
31429 without any constraints.
31431 2017-04-21 Richard Biener <rguenther@suse.de>
31433 PR tree-optimization/78847
31434 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
31436 2017-04-21 Richard Biener <rguenther@suse.de>
31438 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
31439 (build_distinct_type_copy): Likewise.
31440 (build_variant_type_copy): Likewise.
31441 * tree.c (build_qualified_type): Pass down mem-stat info.
31442 (build_distinct_type_copy): Likewise.
31443 (build_variant_type_copy): Likewise.
31445 2017-04-21 Richard Biener <rguenther@suse.de>
31447 PR tree-optimization/80237
31448 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
31450 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
31451 for a simplified result.
31453 2016-04-21 Richard Biener <rguenther@suse.de>
31455 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
31456 sth as strict as a simple_iv but a chrec without symbols and an
31457 operand defined in the loop we are peeling (and not some subloop).
31458 (propagate_constants_for_unrolling): Propagate all constants.
31460 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31463 * config/i386/i386.c (print_reg): Remove assert for disalowed
31464 regno values, call output_operand_lossage instead.
31466 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
31469 * config/i386/constraints.md (Yc): New register constraint.
31470 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
31471 Use Yc constraint for alternative 2 of operand 0. Remove
31472 preferred_for_speed attribute.
31474 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31476 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
31477 lastprivate clauses in SIMT case.
31479 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
31481 * doc/invoke.texi (-Wextra-semi): Document new warning option.
31483 2017-04-20 Richard Biener <rguenther@suse.de>
31485 PR tree-optimization/57796
31486 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
31487 as N scalar stores.
31488 (vect_model_load_cost): Cost gathers as N scalar loads.
31490 2017-04-20 Richard Biener <rguenther@suse.de>
31492 * ggc-page.c (ggc_allocated_p): Rename to ...
31493 (safe_lookup_page_table_entry): ... this and return the lookup
31495 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
31497 2017-04-20 Richard Biener <rguenther@suse.de>
31499 PR tree-optimization/80453
31500 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
31501 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
31502 from the conditions.
31503 (vn_phi_eq): Pass them down.
31504 (vn_phi_lookup): Record them.
31505 (vn_phi_insert): Likewise.
31507 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
31509 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
31510 uninitialized variable warning to avoid buffer overrun.
31512 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
31515 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
31516 is suppressed for '{ 0 }' in C.
31518 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31520 * BASE-VER: Set to 8.0.0.
31522 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
31524 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
31525 priority .init_array and .fini_array section with SECTION_NOTYPE
31528 2017-04-20 Jakub Jelinek <jakub@redhat.com>
31530 PR middle-end/80423
31531 * tree.h (build_array_type): Add typeless_storage default argument.
31532 * tree.c (type_cache_hasher::equal): Also compare
31533 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
31534 (build_array_type): Add typeless_storage argument, set
31535 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
31537 (build_nonshared_array_type): Adjust build_array_type_1 caller.
31538 (build_array_type): Likewise. Add typeless_storage argument.
31540 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31541 Jakub Jelinek <jakub@redhat.com>
31543 PR tree-optimization/80426
31544 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
31545 operation on symbolic operands, also compute the overflow for the
31546 invariant part when the operation degenerates into a negation.
31548 2017-04-19 Jakub Jelinek <jakub@redhat.com>
31551 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
31552 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
31555 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
31557 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
31560 * config/avr/avr.c (tree.h): Include it.
31561 (cgraph.h): Include it.
31562 (avr_encode_section_info): Don't warn for uninitialized progmem
31563 variable if it's just an alias.
31565 2017-04-19 Richard Biener <rguenther@suse.de>
31568 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
31569 when needed by AutoPGO.
31571 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
31574 * doc/lto.texi: Remove an extra 'that'.
31576 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
31578 PR rtl-optimization/80429
31579 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
31580 are only used in debug insns.
31582 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
31583 Vladimir Makarov <vmakarov@redhat.com>
31585 * config/sparc/predicates.md (input_operand): Add comment. Return
31586 true for any memory operand when LRA is in progress.
31587 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
31589 2017-04-18 Jeff Law <law@redhat.com>
31592 * mips.md ({return,simple_return}_internal): Do not overwrite
31595 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31597 PR tree-optimization/80443
31598 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
31599 instead of adding 1, subtract -1 and similarly instead of subtracting
31602 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
31604 PR rtl-optimization/80357
31605 * haifa-sched.c (tmp_bitmap): New variable.
31606 (model_recompute): Handle duplicate use records.
31607 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
31608 (free_global_sched_pressure_data): Free it.
31610 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
31613 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
31614 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
31615 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
31616 instead of SYSTEM_HEADER_DIR.
31618 2017-04-18 Jeff Law <law@redhat.com>
31620 PR middle-end/80422
31621 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
31622 predecessors after walking up the insn chain.
31624 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31627 * dwarf2out.c (modified_type_die): Try harder not to emit internal
31628 sizetype type into debug info.
31630 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31633 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
31634 unneeded test for TARGET_UPPER_REGS_SF.
31635 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
31637 2017-04-18 Jakub Jelinek <jakub@redhat.com>
31640 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
31641 instead of gsi_after_labels.
31643 2017-04-18 Jeff Law <law@redhat.com>
31645 * regcprop.c (maybe_mode_change): Avoid creating copies of the
31649 2017-04-13 Jeff Law <law@redhat.com>
31650 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31651 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31653 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
31656 * config/avr/avr.c (intl.h): Include it.
31657 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
31659 2017-04-18 Martin Liska <mliska@suse.cz>
31661 PR gcov-profile/78783
31662 * gcov-tool.c (gcov_output_files): Validate that destination
31663 file is either removed by the tool or by a user.
31665 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31666 Guy Benyei <guybe@mellanox.com>
31668 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
31669 block, and do not negate it, the stored id is already negative.
31671 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31673 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
31675 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
31678 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
31679 masks of options that should be turned off if the VSX vector
31680 options are turned off.
31681 (OTHER_P8_VECTOR_MASKS): Likewise.
31682 (OTHER_VSX_VECTOR_MASKS): Likewise.
31683 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
31684 rs6000_disable_incompatible_switches to validate no type switches
31686 (rs6000_incompatible_switch): New function to disallow turning on
31687 other vector options if -mno-vsx, -mno-power8-vector, or
31688 -mno-power9-vector are specified.
31690 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31692 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
31694 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31696 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
31697 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
31698 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
31699 (ARG_POINTER_CFA_OFFSET): Likewise.
31701 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
31703 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
31704 conditions to take advantage of various optimizations.
31706 2017-04-13 Jeff Law <law@redhat.com>
31708 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
31709 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
31710 (zero_extendsidi2_dext): Likewise.
31712 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31715 * fold-const.c (fold_ternary_loc): Revert
31716 use op0 instead of fold_convert_loc (loc, type, arg0) part of
31719 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
31721 PR rtl-optimization/80343
31722 * lra-remat.c (update_scratch_ops): Assign original hard reg to
31723 new scratch pseudo.
31725 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
31728 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
31729 to ubsan_encode_value.
31731 2017-04-13 Jeff Law <law@redhat.com>
31733 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
31734 appearing in DEBUG_INSNs.
31736 2017-04-13 Martin Liska <mliska@suse.cz>
31738 PR gcov-profile/80413
31739 * gcov-io.c (gcov_write_string): Copy to buffer just when
31740 allocated size is greater than zero.
31742 2017-04-13 Jakub Jelinek <jakub@redhat.com>
31745 * dwarf2out.c (decls_for_scope): Ignore declarations of
31746 current_function_decl in BLOCK_NONLOCALIZED_VARS.
31748 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
31751 * ipa-visibility.c (non_local_p): Fix typos.
31752 (localize_node): When localizing symbol in same comdat group,
31753 dissolve the group only when we know external symbols are going
31755 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
31757 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31759 PR tree-optimization/79390
31760 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
31761 order does not result in usable sequence, retry with reversed operand
31767 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
31768 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
31769 op0 instead of fold_convert_loc (loc, type, arg0).
31771 2017-04-12 Jeff Law <law@redhat.com>
31773 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
31774 has a delay slot in the generated code.
31776 * config/cris/cris.md (cris_preferred_reload_class): Return
31777 GENNONACR_REGS rather than GENERAL_REGS.
31779 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31782 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
31783 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
31784 signedness of the result type.
31786 2017-04-12 Richard Biener <rguenther@suse.de>
31787 Jeff Law <law@redhat.com>
31789 PR tree-optimization/80359
31790 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
31791 trim stores to TARGET_MEM_REFs.
31793 2017-04-12 Richard Biener <rguenther@suse.de>
31795 PR tree-optimization/79390
31796 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
31797 threading case even more.
31799 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
31802 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
31803 for quad_address_p for TImode, instead of just not indexed_address.
31805 2017-04-12 Richard Biener <rguenther@suse.de>
31806 Bernd Edlinger <bernd.edlinger@hotmail.de>
31808 PR middle-end/79671
31809 * alias.c (component_uses_parent_alias_set_from): Handle
31810 TYPE_TYPELESS_STORAGE.
31811 (get_alias_set): Likewise.
31812 * tree-core.h (tree_type_common): Add typeless_storage flag.
31813 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
31814 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
31815 for types containing members with TYPE_TYPELESS_STORAGE.
31816 (place_field): Likewise.
31817 (layout_type): Likewise for ARRAY_TYPE.
31818 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
31819 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
31820 TYPE_TYPELESS_STORAGE.
31821 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
31823 2017-04-12 Jakub Jelinek <jakub@redhat.com>
31826 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
31827 first argument to type.
31829 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31833 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
31834 CONST0_RTX (mode) rather than const0_rtx where appropriate.
31835 (rs6000_expand_binop_builtin): Likewise.
31836 (rs6000_expand_ternop_builtin): Likewise; also add missing
31837 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
31838 vshasigma built-ins.
31839 * doc/extend.texi: Document that vec_xxpermdi's third argument
31840 must be a constant.
31842 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
31844 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
31845 Use shift_const cost parameter when calculating gain of STV shifts.
31847 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
31849 PR rtl-optimization/70478
31850 * lra-constraints.c (process_alt_operands): Check memory for
31851 disfavoring memory insn operand.
31853 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31855 PR middle-end/80100
31856 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
31857 left shift in unsigned HOST_WIDE_INT type.
31859 PR rtl-optimization/80385
31860 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
31861 (not (neg X)) into (plus X -1) for complex or non-integral modes.
31864 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
31865 if they have any depend clauses.
31867 2017-04-11 Martin Liska <mliska@suse.cz>
31870 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
31871 * ipa-split.c (split_function): Create a local comdat symbol
31872 if caller is in a comdat group.
31874 2017-04-11 Martin Liska <mliska@suse.cz>
31877 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
31880 2017-04-11 Martin Sebor <msebor@redhat.com>
31882 PR middle-end/80364
31883 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
31884 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
31886 (directive::set_width, directive::set_precision, format_character):
31888 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
31891 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
31894 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
31895 conflict, set target->arch_name instead of target->cpu_name.
31897 2017-04-11 Richard Biener <rguenther@suse.de>
31899 PR tree-optimization/80374
31900 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
31901 build_zero_cst, remove fold_convertible_p check again.
31903 2017-04-11 Martin Liska <mliska@suse.cz>
31906 * ubsan.c (instrument_object_size): Do not instrument register
31909 2017-04-11 Jakub Jelinek <jakub@redhat.com>
31912 * config/i386/i386-builtin-types.def
31913 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
31914 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
31915 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
31916 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
31917 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
31918 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
31919 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
31920 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
31921 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
31922 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
31923 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
31924 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
31925 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
31926 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
31927 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
31928 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
31929 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
31930 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
31931 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
31932 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
31933 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
31934 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
31935 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
31936 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
31937 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
31938 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
31939 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
31940 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
31941 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
31942 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
31943 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
31944 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
31945 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
31946 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
31947 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
31948 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
31949 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
31950 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
31951 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
31952 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
31953 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
31954 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
31955 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
31956 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
31957 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
31958 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
31960 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
31961 flag to second_arg_count, handle 4 argument function type _COUNT
31962 aliases, handle second_arg_count on second argument rather than last.
31964 2017-04-10 Jeff Law <law@redhat.com>
31966 PR tree-optimization/80374
31967 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
31968 record anything if we can not convert integer_zero_node to the
31971 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
31974 * config/rs6000/rs6000.c (rs6000_option_override_internal):
31975 Enhance special handling given to the TARGET_P9_MINMAX option in
31976 relation to certain other options.
31978 2017-04-10 Bin Cheng <bin.cheng@arm.com>
31980 PR tree-optimization/80153
31981 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
31982 remove POINTER_PLUS_EXPR's base part directly, rather than through
31985 2017-04-10 Richard Biener <rguenther@suse.de>
31986 Bin Cheng <bin.cheng@arm.com>
31988 PR tree-optimization/80153
31989 * tree-affine.c (aff_combination_to_tree): Get base pointer from
31990 the first element of pointer type aff_tree. Build result expr in
31992 (add_elt_to_tree): Convert to type unconditionally. Remove other
31993 fold_convert calls.
31994 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
31995 (rewrite_use_nonlinear_expr): Check invariant using iv information.
31997 2017-04-10 Richard Biener <rguenther@suse.de>
31999 * tree-ssa-structalias.c (find_func_aliases): Properly handle
32002 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
32004 PR rtl-optimization/70478
32005 * lra-constraints.c (curr_small_class_check): New.
32006 (update_and_check_small_class_inputs): New.
32007 (process_alt_operands): Update curr_small_class_check. Disfavor
32008 alternative insn memory operands. Check available regs for small
32011 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32014 * config/mips/mips.opt (-mvirt): Update description.
32015 * doc/invoke.texi (-mvirt): Likewise.
32017 2017-04-10 Richard Biener <rguenther@suse.de>
32019 PR middle-end/80362
32020 * fold-const.c (fold_binary_loc): Look at unstripped ops when
32021 looking for NEGATE_EXPR in -A / -B to A / B folding.
32023 2017-04-10 Martin Liska <mliska@suse.cz>
32025 PR gcov-profile/80224
32026 * gcov.c (print_usage): Fix usage string.
32027 (get_gcov_intermediate_filename): Remove.
32028 (output_gcov_file): Use both for normal and intermediate format.
32029 (generate_results): Do not initialize special file for
32030 intermediate format.
32032 2017-04-10 Richard Biener <rguenther@suse.de>
32034 PR tree-optimization/80304
32035 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
32038 2017-04-10 Nathan Sidwell <nathan@acm.org>
32041 * config/rs6000/rs6000.c (rs6000_vector_type): New.
32042 (rs6000_init_builtins): Use it.
32044 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32046 * config/arm/arm.md (<mrc>): Add mode to SET source.
32047 (<mrrc>): Likewise.
32049 2017-04-10 Richard Biener <rguenther@suse.de>
32051 PR middle-end/80344
32052 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
32054 2017-04-10 Jakub Jelinek <jakub@redhat.com>
32057 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
32058 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
32059 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
32060 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
32061 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
32062 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
32063 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
32064 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
32065 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
32066 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
32067 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
32068 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
32069 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
32070 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
32071 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
32072 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
32073 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
32074 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
32075 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
32076 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
32077 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
32078 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
32079 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
32081 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
32083 PR rtl-optimization/70478
32084 * lra-constraints.c: Reverse the last patch.
32086 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
32088 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
32089 Add comment for WCHAR_T.
32091 2017-04-08 Martin Liska <mliska@suse.cz>
32094 2017-04-07 Martin Liska <mliska@suse.cz>
32097 * ipa-split.c (split_function): Add function part to a same comdat
32100 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32103 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
32105 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
32107 * rs6000/rs6000.c (vec_load_pendulum): Rename...
32108 (vec_pairing): ...to this.
32109 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
32110 (rs6000_sched_init): Adjust for name change.
32111 (struct rs6000_sched_context): Likewise.
32112 (rs6000_init_sched_context): Likewise.
32113 (rs6000_set_sched_context): Likewise.
32115 2017-04-07 Jakub Jelinek <jakub@redhat.com>
32121 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
32123 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
32124 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
32125 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
32127 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
32129 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
32131 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32133 PR rtl-optimization/70703
32134 * ira-color.c (update_conflict_hard_regno_costs): Use
32135 int64_t instead of HOST_WIDE_INT.
32137 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
32139 PR rtl-optimization/70478
32140 * lra-constraints.c (process_alt_operands): Disfavor alternative
32141 insn memory operands.
32143 2017-04-07 Jeff Law <law@redhat.com>
32145 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
32146 CALL and NOTE_INSN_CALL_ARG_LOCATION.
32148 2017-04-07 Martin Liska <mliska@suse.cz>
32151 * config/aarch64/aarch64.c (aarch64_process_target_attr):
32152 Show error message instead of an ICE.
32154 2017-04-07 Martin Liska <mliska@suse.cz>
32157 * ipa-split.c (split_function): Add function part to a same comdat
32160 2017-04-07 Richard Biener <rguenther@suse.de>
32162 PR middle-end/80341
32163 * tree.c (get_unwidened): Also handle ! for_type case for
32165 * convert.c (do_narrow): Split out from ...
32166 (convert_to_integer_1): ... here. Do not pass final truncation
32167 type to get_unwidened for TRUNC_DIV_EXPR.
32169 2017-04-07 Richard Biener <rguenther@suse.de>
32171 * tree-affine.c (wide_int_ext_for_comb): Take type rather
32173 (aff_combination_const): Adjust.
32174 (aff_combination_scale): Likewise.
32175 (aff_combination_add_elt): Likewise.
32176 (aff_combination_add_cst): Likewise.
32177 (aff_combination_convert): Likewise.
32178 (add_elt_to_tree): Likewise. Remove unused argument.
32179 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
32181 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
32183 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
32185 * config/arm/arm.c (arm_default_short_enums): Use
32186 ARM_DEFAULT_SHORT_ENUMS.
32187 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
32189 2017-04-06 Jakub Jelinek <jakub@redhat.com>
32192 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
32193 members with redundant out-of-class redeclaration.
32195 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32198 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
32199 * config/i386/i386.md (*zero_extendsidi2):
32200 Add (?*x,*x) and (?*v,*v) alternatives.
32202 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
32205 * config/i386/i386.c (ix86_expand_builtin)
32206 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
32207 mode from insn data. Convert operands to insn operand mode.
32208 Copy operands that don't satisfy insn predicate to a register.
32210 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
32212 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
32215 2017-04-06 Richard Biener <rguenther@suse.de>
32217 PR tree-optimization/80334
32218 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
32219 preserve alignment of accesses.
32221 2017-04-06 Richard Biener <rguenther@suse.de>
32223 PR tree-optimization/80262
32224 * tree-sra.c (build_ref_for_offset): Preserve address-space
32226 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
32227 Drop useless address-space information on MEM_REF offsets.
32229 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
32231 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
32233 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32235 PR rtl-optimization/70703
32236 * ira-color.c (update_conflict_hard_regno_costs): Use
32237 HOST_WIDE_INT instead of long.
32239 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
32242 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
32243 not defined for x86_64 target. Add -mmmx target option when __SSE2__
32245 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
32246 for x86_64 target. Handle -m3dnowa option.
32248 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
32250 PR rtl-optimization/70703
32251 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
32252 (update_conflict_hard_regno_costs): Use long instead of unsigned
32253 arithmetic for cost calculation.
32255 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32256 Bernd Edlinger <bernd.edlinger@hotmail.de>
32259 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
32262 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
32265 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
32266 ptr_mode with Pmode throughout.
32267 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
32268 into probe_stack_range and use DImode.
32270 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
32273 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
32274 call_eh_return is true.
32276 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32278 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
32279 Initialize last_match_fntype_index.
32281 2017-04-05 Jakub Jelinek <jakub@redhat.com>
32284 * tree-nvr.c: Include internal-fn.h.
32285 (pass_return_slot::execute): Ignore internal calls without
32288 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32289 Richard Biener <rguenther@suse.de>
32292 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
32293 captures used multiple times, except for the last use.
32294 * generic-match-head.c: Include gimplify.h.
32296 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32298 PR tree-optimization/79390
32299 * target.h (struct noce_if_info): Declare.
32300 * targhooks.h (default_noce_conversion_profitable_p): Declare.
32301 * target.def (noce_conversion_profitable_p): New target hook.
32302 * ifcvt.h (struct noce_if_info): New type, moved from ...
32303 * ifcvt.c (struct noce_if_info): ... here.
32304 (noce_conversion_profitable_p): Renamed to ...
32305 (default_noce_conversion_profitable_p): ... this. No longer
32307 (noce_try_store_flag_constants, noce_try_addcc,
32308 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
32309 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
32310 instead of noce_conversion_profitable_p.
32311 * config/i386/i386.c: Include ifcvt.h.
32312 (ix86_option_override_internal): Don't override
32313 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
32314 (ix86_noce_conversion_profitable_p): New function.
32315 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
32316 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
32317 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
32318 * doc/tm.texi: Regenerated.
32320 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32322 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
32325 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32328 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
32329 instructions for small multiply cores.
32331 2017-04-04 Jeff Law <law@redhat.com>
32333 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
32335 (mips_expand_vec_perm_const): Initialize elements in orig_perm
32336 that are not set by the loop over the elements.
32338 2017-04-04 Jakub Jelinek <jakub@redhat.com>
32341 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
32342 int mode, convert_modes it to mode as unsigned, otherwise use
32343 lowpart_subreg to mode rather than SImode.
32344 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
32345 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
32346 Use DImode instead of SImode for the shift count operand.
32347 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
32350 2017-04-04 Richard Biener <rguenther@suse.de>
32352 PR middle-end/80281
32353 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
32354 arithmetic done for the negate or the plus. Simplify.
32355 (A - (-B) -> A + B): Likewise.
32356 * fold-const.c (split_tree): Make sure to not negate pointers.
32358 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
32360 PR rtl-optimization/60818
32361 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
32362 a compare of comparisons with the thing compared if this results
32363 in a different machine mode.
32365 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
32367 * alias.c (base_alias_check): Fix typo in comment.
32368 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
32369 * cgraphunit.c (symbol_table::compile): Likewise.
32370 * collect2.c (maybe_run_lto_and_relink): Likewise.
32371 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
32372 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
32373 * config/avr/avr.c (avr_map_op_t): Likewise.
32374 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
32375 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
32376 * config/epiphany/epiphany.md (movcc): Likewise.
32377 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
32378 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
32380 * config/mips/mips.c (mips_save_restore_reg): Likewise.
32381 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
32382 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
32383 * config/sh/sh.c (sh_rtx_costs): Likewise.
32384 * fold-const.c (fold_truth_andor): Likewise.
32385 * genautomata.c (collapse_flag): Likewise.
32386 * gengtype.h (struct type::u::s): Likewise.
32387 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
32388 * input.c (FORMAT_AMOUNT): Likewise.
32389 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
32390 (known_aggs_to_agg_replacement_list): Likewise.
32391 * ipa-inline-analysis.c: Likewise.
32392 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
32393 * ipa-polymorphic-call.c
32394 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
32395 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
32396 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
32398 * modulo-sched.c (apply_reg_moves): Likewise.
32399 * omp-expand.c (build_omp_regions_1): Likewise.
32400 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
32401 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
32402 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
32403 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32404 * value-prof.c: Likewise.
32405 * var-tracking.c (val_reset): Likewise.
32407 2017-04-03 Richard Biener <rguenther@suse.de>
32409 PR tree-optimization/80275
32410 * fold-const.c (split_address_to_core_and_offset): Handle
32413 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
32415 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
32416 descriptors is at least equal to that of functions.
32418 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32420 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
32422 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
32425 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
32426 (mov<IMOD4:mode>): New expander.
32427 (*mov<IMOD4:mode>_internal): New insn and split pattern.
32429 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
32431 PR rtl-optimization/79405
32432 * fwprop.c (propagations_left): New variable.
32433 (forward_propagate_into): Decrement it.
32434 (fwprop_init): Initialize it.
32435 (fw_prop): If the variable has reached zero, stop propagating.
32436 (fwprop_addr): Ditto.
32438 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32441 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
32442 a FUNCTION_DECL, pass it as decl instead of origin to
32445 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
32447 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
32450 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
32453 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
32454 TARGET_VSX_SMALL_INTEGER.
32456 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32458 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32459 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
32461 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
32463 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
32464 extraction from odd-numbered MSA register.
32466 2017-03-31 Jakub Jelinek <jakub@redhat.com>
32468 PR middle-end/80173
32469 * expmed.c (store_bit_field_1): Don't attempt to create
32470 a word subreg out of hard registers wider than word if they
32471 have HARD_REGNO_NREGS of 1 for their mode.
32473 PR middle-end/80163
32474 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
32475 conversions to integer types wider than word and pointer.
32478 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
32479 (rtx_equal_for_cselib_p): Pass 0 to it.
32480 * cselib.c (cselib_hasher::equal): Likewise.
32481 (rtx_equal_for_cselib_1): Add depth argument. If depth
32482 is 128, don't look up VALUE locs and punt. Increment
32483 depth in recursive calls when walking VALUE locs.
32485 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
32487 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
32488 (make_gcov_file_name): Use the canonical path name for generating
32490 (read_line): Fix handling of files with ascii null bytes.
32492 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
32494 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
32495 to initialise a vector register instead
32496 of using a const_int.
32498 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32500 PR translation/80189
32501 * gimplify.c (omp_default_clause): Use %qs instead of %s in
32502 diagnostic messages.
32504 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
32507 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
32508 (dfp_diex_<mode>): Update mode of operand 1.
32509 * doc/extend.texi (dxex, dxexq): Document change to return type.
32510 (diex, diexq): Document change to argument type.
32512 2017-03-30 Martin Jambor <mjambor@suse.cz>
32515 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
32516 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
32517 it reflects the signature changes performed at the callee side.
32518 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
32519 to cgraph_build_function_type_skip_args.
32520 (build_function_decl_skip_args): Adjust call to the above function.
32522 2017-03-30 Jakub Jelinek <jakub@redhat.com>
32525 * config/i386/sse.md
32526 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
32527 register as dest whenever it is a MEM not rtx_equal_p to the
32528 corresponding dup operand, and when forcing into reg move the
32529 reg into the memory afterwards.
32530 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
32531 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
32532 for the force_reg mode.
32533 (avx512vl_vextractf128<mode>): Use register as dest either
32534 always when a MEM, or when it is a MEM not rtx_equal_p to the
32535 corresponding dup operand, or even not when it is a CONST_VECTOR
32536 depending on the mode and lo vs. hi.
32537 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
32539 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
32540 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
32541 Likewise. Require that operands[2] is even.
32542 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
32543 Remove extraneous parens. Require that operands[2] is a multiple
32545 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
32546 operands[0] is a MEM if <mask_applied>, the predicates/constraints
32547 disallow memory then.
32549 2017-03-30 Richard Biener <rguenther@suse.de>
32551 PR tree-optimization/77498
32552 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
32553 to non-constants over backedges.
32555 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
32557 PR rtl-optimization/80233
32558 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
32559 as last_combined_insn. Do not test for BARRIER_P separately.
32561 2017-03-29 Andreas Schwab <schwab@suse.de>
32564 * calls.c (prepare_call_address): Convert funexp to Pmode before
32565 copying to temp reg.
32567 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32569 PR tree-optimization/80158
32570 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
32571 Handle possible future case of more than one alternate
32573 (replace_rhs_if_not_dup): Likewise.
32574 (replace_one_candidate): Likewise.
32576 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
32578 PR rtl-optimization/80193
32579 * ira.c (ira): Do not check allocation for LRA.
32581 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
32583 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
32584 (nvptx_output_simt_exit): Declare.
32585 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
32586 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
32587 (init_softstack_frame): Move initialization of crtl->is_leaf to...
32588 (nvptx_declare_function_name): ...here. Emit declaration of local
32589 memory space buffer for omp_simt_enter insn.
32590 (nvptx_output_unisimt_switch): New.
32591 (nvptx_output_softstack_switch): New.
32592 (nvptx_output_simt_enter): New.
32593 (nvptx_output_simt_exit): New.
32594 * config/nvptx/nvptx.h (struct machine_function): New fields
32595 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
32596 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
32597 (UNSPECV_SIMT_EXIT): Ditto.
32598 (omp_simt_enter_insn): New insn.
32599 (omp_simt_enter): New expansion.
32600 (omp_simt_exit): New insn.
32601 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
32603 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
32604 (expand_GOMP_SIMT_ENTER_ALLOC): New.
32605 (expand_GOMP_SIMT_EXIT): New.
32606 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
32607 (GOMP_SIMT_ENTER_ALLOC): Ditto.
32608 (GOMP_SIMT_EXIT): Ditto.
32609 * target-insns.def (omp_simt_enter): New insn.
32610 (omp_simt_exit): Ditto.
32611 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
32613 (lower_rec_simd_input_clauses): Implement SIMT privatization.
32614 (lower_rec_input_clauses): Likewise.
32615 (lower_lastprivate_clauses): Handle SIMT privatization.
32617 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
32618 (ompdevlow_adjust_simt_enter): New.
32619 (find_simtpriv_var_op): New.
32620 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
32621 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
32623 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
32624 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
32625 (copy_decl_for_dup_finish): Ditto.
32627 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
32629 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
32632 * config/i386/i386.c (ix86_option_override_internal): Always
32633 allow -mpreferred-stack-boundary=3 for 64-bit targets.
32635 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32637 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
32639 2017-03-28 Bin Cheng <bin.cheng@arm.com>
32641 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
32642 mark new edge's irreducible flag accordign to it.
32643 (vect_do_peeling): Check loop preheader edge's irreducible flag
32644 and pass it to function slpeel_add_loop_guard.
32646 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
32648 PR tree-optimization/80218
32649 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32650 Update block frequencies and counts.
32652 2017-03-28 Richard Biener <rguenther@suse.de>
32654 PR tree-optimization/78644
32655 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
32656 of a simplification result we may not use it at all.
32658 2017-03-28 Richard Biener <rguenther@suse.de>
32661 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
32662 without arguments, generate default definition of a SSA name.
32664 2017-03-28 Richard Biener <rguenther@suse.de>
32666 PR middle-end/80222
32667 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
32668 TYPE_REF_CAN_ALIAS_ALL references.
32669 * fold-const.c (fold_indirect_ref_1): Likewise.
32671 2017-03-28 Martin Liska <mliska@suse.cz>
32674 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
32675 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
32677 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32678 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
32680 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
32681 (EXTRA_SPECS): Define.
32682 (SUBTARGET_EXTRA_SPECS): Likewise.
32683 (SUBTARGET_CPP_SPEC): Likewise.
32684 * config/arc/elf.h (EXTRA_SPECS): Renamed to
32685 SUBTARGET_EXTRA_SPECS.
32686 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
32688 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
32690 * config/arc/simdext.md (vst64_insn): Update pattern.
32691 (vld32wh_insn): Likewise.
32692 (vld32wl_insn): Likewise.
32693 (vld64_insn): Likewise.
32694 (vld32_insn): Likewise.
32696 2017-03-28 Marek Polacek <polacek@redhat.com>
32699 * fold-const.c (fold_comparison): Use protected_set_expr_location
32700 instead of SET_EXPR_LOCATION.
32702 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
32704 * tree.c (add_expr): Avoid name lookup warning.
32706 2017-03-27 Jeff Law <law@redhat.com>
32708 PR tree-optimization/80216
32709 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
32710 function name. Limit recursion depth.
32711 (record_temporary_equivalences): Corresponding changes.
32713 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
32715 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
32718 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32721 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
32723 * cfgcleanup.c (reg_note_cfa_p): New array.
32724 (insns_have_identical_cfa_notes): New function.
32725 (old_insns_match_p): Don't cross-jump in between /f
32726 and non-/f instructions. If both i1 and i2 are frame related,
32727 verify all CFA notes, their order and content.
32729 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
32732 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
32733 HImode and SImode with zero extend to DImode to one insn.
32734 (bswap<mode>2_extenddi): Likewise.
32735 (bswapsi2_extenddi): Likewise.
32736 (bswaphi2_extendsi): Likewise.
32737 (bswaphi2): Combine bswap HImode and SImode into one insn.
32738 Separate memory insns from swapping register.
32739 (bswapsi2): Likewise.
32740 (bswap<mode>2): Likewise.
32741 (bswaphi2_internal): Delete, no longer used.
32742 (bswapsi2_internal): Likewise.
32743 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
32744 store, and gpr<-gpr swap insns.
32745 (bswap<mode>2_store): Likewise.
32746 (bswaphi2_reg): Register only splitter, combine with the splitter.
32747 (bswaphi2 splitter): Likewise.
32748 (bswapsi2_reg): Likewise.
32749 (bswapsi2 splitter): Likewise.
32750 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
32751 the insns into load, store, and register/register insns.
32752 (bswapdi2_ldbrx): Likewise.
32753 (bswapdi2_load): Likewise.
32754 (bswapdi2_store): Likewise.
32755 (bswapdi2_reg): Likewise.
32757 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
32759 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
32760 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
32762 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
32765 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
32767 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32768 special handling for target option conflicts between dform
32769 options (-mpower9-dform, -mpower9-dform-vector,
32770 -mpower9-dform-scalar) and -mno-direct-move.
32772 2017-03-27 Richard Biener <rguenther@suse.de>
32774 PR tree-optimization/80181
32775 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
32777 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32779 * config/arc/predicates.md (move_double_src_operand): Replace the
32780 call to move_double_src_operand with a call to address_operand.
32782 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32784 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
32785 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
32786 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
32788 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32790 * config/arc/predicates.md (long_immediate_loadstore_operand):
32791 Consider scaled addresses cases.
32793 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
32795 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
32796 restored when in interrupt.
32797 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
32798 doesn't have delay slot.
32800 2017-03-27 Richard Biener <rguenther@suse.de>
32803 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
32804 inlined thunk clones.
32806 2017-03-27 Jakub Jelinek <jakub@redhat.com>
32809 * asan.c (instrument_derefs): Copy over last operand from
32810 original COMPONENT_REF to the new COMPONENT_REF with
32811 DECL_BIT_FIELD_REPRESENTATIVE.
32812 * ubsan.c (instrument_object_size): Likewise.
32814 2017-03-27 Richard Biener <rguenther@suse.de>
32816 PR tree-optimization/80170
32817 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
32818 sure DR/SCEV didnt fold in constants we do not see when looking
32819 at the reference base alignment.
32821 2017-03-27 Richard Biener <rguenther@suse.de>
32823 PR middle-end/80171
32824 * gimple-fold.c (fold_ctor_reference): Properly guard against
32825 NULL return value from canonicalize_constructor_val.
32827 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
32830 * config/i386/i386.c (ix86_expand_builtin)
32831 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
32832 flags reg setting and flags reg using instructions.
32833 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
32834 clobbering instructions to zero extend op2.
32836 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
32838 * doc/install.texi (Configuration) <--with-aix-soname>:
32839 Update link to AIX ld.
32841 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
32843 PR rtl-optimization/80160
32844 PR rtl-optimization/80159
32845 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
32846 reg_alternate_class into account.
32848 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
32851 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
32852 to consider in curr_insn_transform.
32854 2017-03-24 Jakub Jelinek <jakub@redhat.com>
32856 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
32857 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
32858 and emit_mode_inner.
32860 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32862 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
32863 argument to the overloaded builtin variants. Use the new flag to
32864 deprecate certain builtin variants.
32865 * config/s390/s390-builtin-types.def: Add new builtin types.
32866 * config/s390/s390-builtins.h: Support new flags field for
32867 overloaded builtins.
32868 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
32869 (s390_macro_to_expand): Enable vector float data type.
32870 (s390_cpu_cpp_builtins_internal): Indicate support of the new
32871 builtins by incrementing the __VEC__ version number.
32872 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
32874 (s390_resolve_overloaded_builtin): Emit error messages depending
32875 on the builtin flags.
32876 * config/s390/s390.c (s390_expand_builtin): Support additional
32877 flags argument. Change error message to match the messages
32878 emitted in s390-c.c.
32879 * config/s390/s390.md: New UNSPEC_* constants.
32880 (op_type): Add new instruction types.
32881 * config/s390/vecintrin.h: Add new builtins and test data class
32883 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
32884 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
32885 (VEC_INEXACT, VEC_NOINEXACT): New constants.
32886 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
32887 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
32888 ("vec_mergel<mode>"): V_HW -> VEC_HW.
32890 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
32891 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
32892 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
32893 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
32895 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
32896 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
32897 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
32898 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
32900 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
32901 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
32902 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
32903 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
32904 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
32905 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
32906 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
32908 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
32909 ("vec_scatter_element<V_HW_4:mode>_DI")
32910 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
32911 ("vec_fpint<mode>", "vflls")
32912 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
32913 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
32914 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
32915 ("*vec_cmphe<mode>_cc"): ... these.
32917 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
32918 mode constant instead of magic value.
32920 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32922 * config/s390/s390.c (s390_expand_vec_compare): Support other
32923 vector floating point modes than just V2DF.
32924 (s390_expand_vcond): Likewise.
32925 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
32926 (s390_cannot_change_mode_class): Prevent mode changes between TF
32927 and V1TF in vector registers.
32928 * config/s390/s390.md (DF, SF): New mode attributes.
32929 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
32930 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
32931 SFmode support for VRs.
32932 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
32934 (VFT, VF_HW): New mode iterators.
32935 (vw, sdx): New mode attributes.
32936 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
32937 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
32938 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
32939 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
32940 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
32941 also the new vector floating point modes. Renaming to ...
32943 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
32944 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
32945 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
32946 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
32947 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
32948 ("vec_unordered<mode>"): ... these.
32950 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
32951 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
32952 ("*vec_extendv2df"): New insn definitions.
32954 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32956 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
32957 ("mulditi3_2", "*muldi3_sign"): New patterns.
32958 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
32959 rename the pattern definition.
32961 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32963 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
32965 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
32967 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32969 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
32970 instruction if possible.
32971 * config/s390/vector.md (vec_halfnumelts): New mode
32973 ("*vec_vllezlf<mode>"): New pattern.
32975 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32977 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
32978 ("popcountv4si2", "popcountv2di2"): Rename to ...
32979 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
32980 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
32982 ("popcount<mode>2_vxe"): New pattern.
32984 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32986 * common/config/s390/s390-common.c (processor_flags_table): Add
32988 * config.gcc: Add arch12.
32989 * config/s390/driver-native.c (s390_host_detect_local_cpu):
32990 Default to arch12 for unknown CPU model numbers.
32991 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
32992 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
32993 PROCESSOR_max sanity check.
32994 * config/s390/s390-opts.h (enum processor_type): Add
32996 * config/s390/s390.c (processor_table): Add arch12.
32997 (s390_expand_builtin): Add check for B_VXE flag.
32998 (s390_issue_rate): Add PROCESSOR_ARCH12.
32999 (s390_get_sched_attrmask): Likewise.
33000 (s390_get_unit_mask): Likewise.
33001 (s390_sched_score): Enable z13 scheduling for arch12.
33002 (s390_sched_reorder): Likewise.
33003 (s390_sched_variable_issue): Likewise.
33004 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
33006 (s390_tune_attr): Use z13 scheduling also for arch12.
33007 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
33008 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
33009 (TARGET_VXE_P): New macros.
33010 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
33011 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
33012 * config/s390/s390.opt: Add arch12 as processor_type.
33014 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33016 * config/s390/s390.md
33017 ("fixuns_truncdddi2", "fixuns_trunctddi2")
33018 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
33019 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
33021 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
33022 Rename expanders to ...
33024 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
33025 ("fixuns_truncdddi2_emu"): ... these.
33027 ("fixuns_trunc<mode>si2_emu"): New expander.
33029 ("*fixuns_truncdfdi2_z13"): Rename to ...
33030 ("*fixuns_truncdfdi2_vx"): ... this.
33032 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33034 * config/s390/2964.md: Remove the single element vector compare
33035 instructions which are no longer used.
33036 * config/s390/s390.c (s390_select_ccmode): Remove handling of
33038 (s390_canonicalize_comparison): Remove handling of DFmode
33040 (s390_expand_vec_compare_scalar): Remove function.
33041 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
33042 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
33044 ("*cmp<mode>_ccs"): Add wfcdb instruction.
33046 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33048 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
33050 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
33051 will anyway by matched by mov<mode>_64dfp.
33053 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33055 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
33056 vlef/vstef. Add missing operand to vleif.
33058 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33060 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
33061 pair for all vector types with 64 bit elements.
33062 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
33063 * config/s390/vector.md (V_HW_64): ... here.
33064 (V_128_NOSINGLE): New mode iterator.
33065 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
33066 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
33067 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
33068 ("*vec_load_pairv2di"): Change to ...
33069 ("*vec_load_pair<mode>"): ... this one.
33071 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33073 * config/s390/constraints.md: Add comments.
33074 (jKK): Reject element sizes > 8 bytes.
33075 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
33077 * config/s390/s390.md: Add the s_operand checks formerly in
33078 s390_split_ok_p to various splitters where they are still
33080 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
33081 for 128 bit vectors. Plus two splitters.
33083 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33085 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
33088 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33091 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
33092 error if the boundary argument is not constant.
33094 2017-03-24 Jakub Jelinek <jakub@redhat.com>
33096 PR rtl-optimization/80112
33097 * loop-doloop.c (doloop_condition_get): Don't check condition
33098 if cmp isn't SET with IF_THEN_ELSE src.
33100 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33102 PR tree-optimization/80158
33103 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
33104 replacing a candidate statement, also replace it for the
33105 candidate's alternate interpretation.
33106 (replace_rhs_if_not_dup): Likewise.
33107 (replace_one_candidate): Likewise.
33109 2017-03-24 Richard Biener <rguenther@suse.de>
33111 PR tree-optimization/80167
33112 * graphite-isl-ast-to-gimple.c
33113 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
33115 (translate_isl_ast_to_gimple::get_rename): Likewise.
33117 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
33119 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
33120 handling of certain combinations of target options, including the
33121 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
33122 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
33124 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33127 * config/arm/arm.md (*load_multiple): Add reload_completed to
33128 matching condition.
33130 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33131 Richard Biener <rguenth@suse.de>
33133 PR tree-optimization/79908
33134 PR tree-optimization/80136
33135 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33136 been cast away, gimplify_and_add suffices.
33138 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
33140 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
33142 2017-03-23 Richard Biener <rguenther@suse.de>
33144 PR tree-optimization/80032
33145 * gimplify.c (gimple_push_cleanup): Forced unconditional
33146 cleanups still have to go to the conditional_cleanups
33149 2017-03-22 Jakub Jelinek <jakub@redhat.com>
33151 PR tree-optimization/80072
33152 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
33154 (next_operand_entry_id): Change type to unsigned int.
33155 (sort_by_operand_rank): Make sure to return the right return value
33156 even if unsigned fields are bigger than INT_MAX.
33157 (struct oecount): Change cnt and id type to unsigned int.
33158 (oecount_hasher::equal): Formatting fix.
33159 (oecount_cmp): Make sure to return the right return value
33160 even if unsigned fields are bigger than INT_MAX.
33161 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
33164 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
33165 TREE_READONLY on result if writing it more than once.
33168 * doc/invoke.texi (-fsanitize=thread): Document that with
33169 -fnon-call-exceptions atomics are not able to throw
33173 * tsan.c: Include tree-eh.h.
33174 (instrument_builtin_call): Call maybe_clean_eh_stmt or
33175 maybe_clean_or_replace_eh_stmt where needed.
33176 (instrument_memory_accesses): Add cfg_changed argument.
33177 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
33178 if it returned true.
33179 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
33181 PR rtl-optimization/63191
33182 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
33183 wrapper function, moved the whole old content into ...
33184 (ix86_delegitimize_address_1): ... this. New inline function.
33185 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
33186 true as last argument instead of ix86_delegitimize_address.
33188 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33190 * config/aarch64/aarch64.c (generic_branch_cost): Copy
33191 cortexa57_branch_cost.
33193 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
33195 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
33197 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33200 * doc/md.texi (Constraints): Document wA constraint.
33201 * config/rs6000/constraints.md (wA): New.
33202 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
33203 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
33204 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
33205 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
33207 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
33210 * gimplify.c (is_oacc_declared): New function.
33211 (oacc_default_clause): Use it to set default flags for acc declared
33212 variables inside parallel regions.
33213 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
33214 declared variables.
33215 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
33216 declare attribute to any decl as necessary.
33218 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
33221 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
33222 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
33223 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
33224 (arm_arch_lpae): This.
33225 * config/arm/arm.c (arm_arch7ve): Rename into ...
33226 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
33227 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
33230 2017-03-22 Martin Liska <mliska@suse.cz>
33233 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
33234 error message instead of an ICE.
33236 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33238 * doc/extend.texi (6.11 Additional Floating Types): Revise.
33240 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
33242 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
33244 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
33247 2017-03-21 Martin Sebor <msebor@redhat.com>
33249 * doc/extend.texi: Use "cannot" instead of "can't."
33250 * doc/hostconfig.texi: Same.
33251 * doc/install.texi: Same.
33252 * doc/invoke.texi: Same.
33253 * doc/loop.texi: Same.
33254 * doc/md.texi: Same.
33255 * doc/objc.texi: Same.
33256 * doc/rtl.texi: Same.
33257 * doc/tm.texi: Same.
33258 * doc/tm.texi.in: Same.
33259 * doc/trouble.texi: Same.
33261 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
33264 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
33265 (collect_checksum_attributes): Set it.
33266 (die_checksum_ordered): Use it.
33268 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33270 PR tree-optimization/79908
33271 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
33272 change: For a VA_ARG whose LHS has been cast away, use
33273 force_gimple_operand to construct the side effects.
33275 2017-03-21 David Malcolm <dmalcolm@redhat.com>
33277 PR translation/80001
33278 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
33279 more amenable to translation.
33280 (oacc_loop_auto_partitions): Likewise.
33282 2017-03-21 Marek Polacek <polacek@redhat.com>
33283 Martin Sebor <msebor@redhat.com>
33285 PR tree-optimization/80109
33286 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
33287 on INTEGRAL_TYPE_P.
33289 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33290 Segher Boessenkool <segher@kernel.crashing.org>
33293 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
33294 check reg_used_between_p between insn and one of succ or succ2
33295 depending on if succ is artificial insn not inserted into insn
33298 2017-03-21 Martin Liska <mliska@suse.cz>
33300 PR gcov-profile/80081
33301 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
33302 * doc/gcc.texi: Include gcov-dump stuff.
33303 * doc/gcov-dump.texi: New file.
33305 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
33307 PR rtl-optimization/79150
33308 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
33309 conditional jump, if the jump is the last insn of the loop.
33311 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33312 Richard Biener <rguenth@suse.de>
33314 PR tree-optimization/79908
33315 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
33316 been cast away, use force_gimple_operand to construct the side
33319 2017-03-21 Martin Liska <mliska@suse.cz>
33321 PR libfortran/79956
33322 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
33325 2017-03-21 Brad Spengler <spender@grsecurity.net>
33328 * plugin.c (htab_hash_plugin): New function.
33329 (add_new_plugin): Use it and adjust.
33330 (parse_plugin_arg_opt): Adjust.
33331 (init_one_plugin): Likewise.
33333 2017-03-21 Richard Biener <rguenther@suse.de>
33335 PR tree-optimization/80032
33336 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
33337 if set force the cleanup to happen unconditionally.
33338 (gimplify_target_expr): Push inserted clobbers with force_uncond
33339 to avoid them being removed by control-dependent DCE.
33341 2017-03-21 Richard Biener <rguenther@suse.de>
33343 PR tree-optimization/80122
33344 * tree-inline.c (copy_bb): Do not expans va-arg packs or
33345 va_arg_pack_len when the inlined call stmt requires pack
33347 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
33349 2017-03-21 Jakub Jelinek <jakub@redhat.com>
33352 * tsan.c (instrument_builtin_call): If the memory model argument
33353 is not a constant, assume it is valid.
33356 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
33359 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33361 PR rtl-optimization/79910
33362 * combine.c (can_combine_p): Do not allow combining an I0 or I1
33363 if its dest is used by an insn before I2 (other than the combined
33364 insns themselves, which are properly handled already).
33366 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
33369 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33371 * combine.c (record_used_regs): New static function.
33372 (try_combine): Handle situations where there is an additional
33373 instruction between I2 and I3 which needs to have a LOG_LINK
33377 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33379 * combine.c (try_combine): Delete redundant i1 test. Call
33380 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33382 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33385 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
33386 alternatives 13/14.
33388 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33390 PR tree-optimization/80054
33391 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
33392 the optimization if a PHI or any of its arguments is not dominated
33393 by the candidate's basis. Use gphi* rather than gimple* as
33395 (replace_profitable_candidates): Clean up a gimple* variable that
33396 should be a gphi* variable.
33398 2017-03-20 Martin Sebor <msebor@redhat.com>
33401 * doc/extend.texi (attribute constructor): Document present limitation.
33403 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
33406 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
33407 __POWER9_VECTOR__ #ifdef control, change template definition to
33408 use Power9-specific built-in function.
33409 (vec_any_eq): Likewise.
33410 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
33411 to control outcomes from this test.
33412 (vector_ae_<mode>p): For VEC_F modes, likewise.
33414 2017-03-20 Ian Lance Taylor <iant@google.com>
33416 * config/i386/i386.c (ix86_function_regparm): Save an extra
33417 register for -fsplit-stack with DECL_STATIC_CHAIN.
33419 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33422 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
33423 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
33425 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
33427 * config/riscv/riscv.c (riscv_print_operand): Use "fence
33429 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
33432 2017-03-20 Marek Polacek <polacek@redhat.com>
33435 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
33437 2017-03-20 Richard Biener <rguenther@suse.de>
33439 PR tree-optimization/80113
33440 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
33441 allocate extra SSA name for PHI def.
33442 (add_close_phis_to_outer_loops): Likewise.
33443 (add_close_phis_to_merge_points): Likewise.
33444 (copy_loop_close_phi_args): Likewise.
33445 (copy_cond_phi_nodes): Likewise.
33447 2017-03-20 Martin Liska <mliska@suse.cz>
33449 PR middle-end/79753
33450 * tree-chkp.c (chkp_build_returned_bound): Do not build
33451 returned bounds for a LHS that's not a BOUNDED_P type.
33453 2017-03-20 Martin Liska <mliska@suse.cz>
33457 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
33458 COMPLEX_CST and VECTOR_CST.
33460 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33463 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
33464 target operand. A new splitter adds the clobber statement in case
33465 the target operand is dead anyway.
33467 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
33469 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
33470 to age-old versions of binutils and glibc.
33472 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
33474 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
33476 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33478 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
33480 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
33482 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
33483 requirement for binutils 2.13.
33485 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
33487 * combine.c (try_combine): Delete redundant i1 test. Call
33488 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
33490 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
33492 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
33493 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
33495 <riscv64-*-elf>: Re-arrange section
33496 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
33497 <riscv32-*-linux>: Likewise.
33498 <riscv64-*-elf>: Likewise
33499 <riscv64-*-linux>: Likewise.
33501 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
33504 * aarch64.opt(verbose-cost-dump): Fix typo.
33506 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
33509 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
33510 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
33512 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
33514 * reload.c (find_reloads): When reloading a nonoffsettable address,
33515 use RELOAD_OTHER for it and its address reloads.
33517 PR rtl-optimization/79910
33518 * combine.c (record_used_regs): New static function.
33519 (try_combine): Handle situations where there is an additional
33520 instruction between I2 and I3 which needs to have a LOG_LINK
33523 2017-03-17 Jeff Law <law@redhat.com>
33525 PR tree-optimization/71437
33526 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
33527 conditional in the hash table first.
33528 (vrp_dom_walker::before_dom_children): Extract condition from
33529 ASSERT_EXPR. Record condition, its inverion and any implied
33530 conditions as well.
33532 2017-03-17 Marek Polacek <polacek@redhat.com>
33533 Markus Trippelsdorf <markus@trippelsdorf.de>
33535 PR tree-optimization/80079
33536 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
33539 2017-03-17 Richard Biener <rguenther@suse.de>
33541 PR middle-end/80075
33542 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
33543 Properly verify the LHS before the RHS possibly claims to be
33545 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
33548 2017-03-17 Martin Jambor <mjambor@suse.cz>
33550 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
33551 (List of -O2 options): Likewise.
33552 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
33555 2017-03-17 Tom de Vries <tom@codesourcery.com>
33557 * gcov-dump.c (print_usage): Print bug_report_url.
33559 2017-03-17 Richard Biener <rguenther@suse.de>
33561 PR middle-end/80050
33562 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
33563 (parser::peek): Likewise.
33565 2017-03-17 Richard Biener <rguenther@suse.de>
33567 PR tree-optimization/80048
33568 * sese.c (free_sese_info): Properly release rename_map and
33569 copied_bb_map elements.
33571 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
33573 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
33574 Add linked-list forward and backlinks. Insert on
33575 construction, remove on destruction.
33576 (class pass_store_merging): Add m_stores_head field.
33577 (pass_store_merging::terminate_and_process_all_chains):
33578 Iterate over m_stores_head list.
33579 (pass_store_merging::terminate_all_aliasing_chains):
33581 (pass_store_merging::execute): Check for debug stmts first.
33582 Push new chains onto the m_stores_head stack.
33584 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
33587 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
33588 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
33589 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
33591 2017-03-16 Jeff Law <law@redhat.com>
33593 PR tree-optimization/71437
33594 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
33595 member function. Implementation moved into after_dom_children
33596 member function and into the threader's thread_outgoing_edges
33598 (dom_opt_dom_walker::after_dom_children): Simplify by moving
33599 some code into new thread_outgoing_edges.
33600 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
33601 definition. Simplify marker handling (do it here). Assume we always
33602 have the available expression and the const/copies tables.
33603 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
33605 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
33606 * tree-vrp.c (equiv_stack): No longer file scoped.
33607 (vrp_dom_walker): New class.
33608 (vrp_dom_walker::before_dom_children): New member function.
33609 (vrp_dom_walker::after_dom_children): Likewise.
33610 (identify_jump_threads): Setup domwalker. Use it rather than
33611 walking edges in a random order by hand. Simplify setup/finalization.
33612 (finalize_jump_threads): Remove.
33613 (vrp_finalize): Do not call identify_jump_threads here.
33614 (execute_vrp): Do it here instead and call thread_through_all_blocks
33617 PR tree-optimization/71437
33618 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
33620 (simplify_stmt_for_jump_threading): Add basic_block argument. All
33622 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
33623 (dom_opt_dom_walker::thread_across_edge): Remove
33624 handle_dominating_asserts argument. All callers changed.
33625 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
33626 changes. Remove calls to lhs_of_dominating_assert. Other
33627 uses of handle_dominating_asserts turn into unconditional code
33628 (simplify_control_stmt_condition_1): Likewise.
33629 (simplify_control_stmt_condition): Likewise.
33630 (thread_through_normal_block, thread_across_edge): Likewise.
33631 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
33632 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
33633 object if it is not an SSA_NAME.
33634 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
33635 before calling into the VRP specific simplifiers.
33636 (identify_jump_threads): Remove handle_dominating_asserts
33639 2017-03-16 Jakub Jelinek <jakub@redhat.com>
33642 * tree-diagnostic.c (default_tree_printer): No longer static.
33643 * tree-diagnostic.h (default_tree_printer): New prototype.
33645 2017-03-16 Tamar Christina <tamar.christina@arm.com>
33647 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
33648 Change ins into fmov.
33650 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33652 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
33653 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
33654 Use h_con constraint for operand 1.
33655 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
33656 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
33658 2017-03-15 Jeff Law <law@redhat.com>
33660 PR tree-optimization/71437
33661 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
33662 (record_temporary_equivalences): Use it.
33664 PR tree-optimization/71437
33665 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
33666 tree-ssa-scopedtables.
33667 (lookup_avail_expr, build_and_record_new_cond): Likewise.
33668 (record_conditions, record_cond, vuse_eq): Likewise.
33669 (record_edge_info): Adjust to API tweak of record_conditions.
33670 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
33671 (record_temporary_equivalences, optimize_stmt): Likewise.
33672 (eliminate_redundant_computations): Likewise.
33673 (record_equivalences_from_stmt): Likewise.
33674 * tree-ssa-scopedtables.c: Include options.h and params.h.
33675 (vuse_eq): New function, moved from tree-ssa-dom.c
33676 (build_and_record_new_cond): Likewise.
33677 (record_conditions): Likewise. Accept vector of conditions rather
33678 than edge_equivalence structure for first argument.
33679 for the first argument.
33680 (avail_exprs_stack::lookup_avail_expr): New member function, moved
33681 from tree-ssa-dom.c.
33682 (avail_exprs_stack::record_cond): Likewise.
33683 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
33684 from tree-ssa-dom.c.
33685 (avail_exprs_stack): Add new member functions lookup_avail_expr
33687 (record_conditions): Declare.
33689 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
33692 * lra-constraints.c (process_alt_operands): Increase reject for
33693 reloading an input/output operand.
33695 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
33698 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
33699 insns to convert from signed/unsigned char/short to IEEE 128-bit
33701 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
33703 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
33706 * config/i386/i386.c (ix86_vector_duplicate_value): Create
33707 subreg of inner mode for values already in registers.
33709 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
33711 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
33712 iteration reg is used after the loop.
33714 2017-03-14 Martin Sebor <msebor@redhat.com>
33716 PR tree-optimization/79800
33717 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
33718 precision in negative-positive range.
33719 (format_floating): Call non-const overload with adjusted precision.
33721 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
33724 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
33727 2017-03-14 Martin Sebor <msebor@redhat.com>
33729 PR middle-end/80020
33730 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
33731 * builtins.def (aligned_alloc): Use it.
33734 * Makefile.in (GTFILES): Add calls.c.
33735 * calls.c: Include "gt-calls.h".
33737 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
33739 PR rtl-optimization/79728
33740 * regs.h (struct target_regs): New field
33741 x_contains_allocatable_regs_of_mode.
33742 (contains_allocatable_regs_of_mode): New macro.
33743 * reginfo.c (init_reg_sets_1): Initialize it, and change
33744 contains_reg_of_mode so it includes global regs as well.
33745 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
33746 rather than contains_regs_of_mode.
33748 2017-03-14 Martin Liska <mliska@suse.cz>
33750 * doc/invoke.texi: Document options that can't be combined with
33751 -fcheck-pointer-bounds.
33753 2017-03-14 Martin Liska <mliska@suse.cz>
33755 PR middle-end/79831
33756 * doc/invoke.texi (-Wchkp): Document the option.
33758 2017-03-14 Martin Liska <mliska@suse.cz>
33760 * Makefile.in: Install gcov-dump.
33762 2017-03-14 Martin Liska <mliska@suse.cz>
33764 * multiple_target.c (expand_target_clones): Bail out for
33765 an invalid attribute.
33767 2017-03-14 Richard Biener <rguenther@suse.de>
33769 * alias.c (struct alias_set_entry): Pack properly.
33770 * cfgloop.h (struct loop): Likewise.
33771 * cse.c (struct set): Likewise.
33772 * ipa-utils.c (struct searchc_env): Likewise.
33773 * loop-invariant.c (struct invariant): Likewise.
33774 * lra-remat.c (struct cand): Likewise.
33775 * recog.c (struct change_t): Likewise.
33776 * rtl.h (struct address_info): Likewise.
33777 * symbol-summary.h (function_summary): Likewise.
33778 * tree-loop-distribution.c (struct partition): Likewise.
33779 * tree-object-size.c (struct object_size_info): Likewise.
33780 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
33781 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
33782 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
33783 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
33784 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
33785 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
33786 (struct _stmt_vec_info): Likewise.
33788 2017-03-14 Martin Liska <mliska@suse.cz>
33791 * multiple_target.c (create_dispatcher_calls): Check that
33792 a target can create a function dispatcher.
33794 2017-03-14 Martin Liska <mliska@suse.cz>
33797 * multiple_target.c (expand_target_clones): Drop local.local
33798 flag for default implementation.
33800 2017-03-14 Richard Biener <rguenther@suse.de>
33802 PR tree-optimization/80030
33803 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
33805 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
33807 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
33808 gcc_fallthrough() instead of __attribute__((fallthrough));
33810 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33812 * doc/gcc.texi: Remove "up" link to (DIR).
33813 * doc/gccint.texi: Ditto.
33815 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
33817 * doc/install.texi (Specific) <avr>: Remove reference to
33820 2017-03-13 Jeff Law <law@redhat.com>
33822 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
33823 attribute rather than comments.
33825 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
33826 match_scratch operand is highest.
33828 2017-03-13 Martin Liska <mliska@suse.cz>
33830 PR middle-end/78339
33831 * ipa-pure-const.c (warn_function_noreturn): If the declarations
33832 is a CHKP clone, use original declaration.
33834 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33836 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
33837 (arc_conditional_register_usage): Use a different allocation order
33838 when optimizing for size.
33839 * common/config/arc/arc-common.c (arc_option_optimization_table):
33840 Section anchors default on when optimizing for size.
33842 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33844 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
33846 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33848 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
33849 * config/arc/arc.md (cpu_facility): Add cd variant.
33850 (*movqi_insn): Add code density variant.
33851 (*movhi_insn): Likewise.
33852 (*movqi_insn): Likewise.
33853 (*addsi3_mixed): Likewise.
33854 (subsi3_insn): Likewise.
33856 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33858 * config/arc/arc.md (movsi_cond_exec): Update constraint.
33860 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
33862 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
33863 expressions with MINUS and UNARY ops.
33865 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33868 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
33870 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
33871 between vec_select and vector argument.
33872 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
33873 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
33874 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
33875 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
33876 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
33877 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
33879 2017-03-13 Richard Biener <rguenther@suse.de>
33882 * params.def (vect-max-peeling-for-alignment): Fix typo.
33884 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33886 * doc/install.texi (Specific) <mips-*-*>: Remove description of
33887 issue that only occurred with binutils below 2.18.
33889 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33891 * doc/install.texi (Specific) <cris-axis-elf>: No longer
33892 refer to binutils 2.11/2.12 minimum.
33894 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
33896 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
33897 ftp.kernel.org and simplify binutils requirement.
33899 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
33901 * doc/invoke.texi (Warning Options): Fix spelling of link-time
33903 (Optimize Options): Ditto. Also remove redundancy.
33905 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33907 PR translation/79848
33908 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
33910 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
33911 to G_ to avoid double translation.
33913 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33915 PR translation/79923
33916 * auto-profile.c (get_combined_location): Convert leading
33917 character of diagnostics to lower case and remove trailing period.
33918 (read_profile): Likewise for various diagnostics.
33919 * config/arm/arm.c (arm_option_override): Remove trailing period
33920 from various diagnostics.
33921 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
33922 (msp430_expand_delay_cycles): Likewise.
33924 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33927 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
33928 full command-line argument, rather than just "str".
33929 (aarch64_validate_march): Likewise.
33930 (aarch64_validate_mtune): Likewise.
33932 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
33934 PR rtl-optimization/78911
33935 * lra-assigns.c (must_not_spill_p): New function.
33936 (spill_for): Use it.
33938 2017-03-10 Jakub Jelinek <jakub@redhat.com>
33940 PR tree-optimization/79981
33941 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
33942 ATOMIC_COMPARE_EXCHANGE ifn result.
33943 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
33944 IFN_ATOMIC_COMPARE_EXCHANGE.
33946 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33949 * opts.c (parse_sanitizer_options): Add missing question mark to
33950 "did you mean" message.
33952 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33954 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
33956 (VMULEUH_UNS): Likewise.
33957 (VMULOUB_UNS): Likewise.
33958 (VMULOUH_UNS): Likewise.
33959 * config/rs6000/rs6000.c (builtin_function_type): Remove
33960 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
33962 2017-03-10 David Malcolm <dmalcolm@redhat.com>
33965 * read-rtl-function.c (function_reader::read_rtx_operand): Update
33966 x with result of extra_parsing_for_operand_code_0.
33967 (function_reader::extra_parsing_for_operand_code_0): Convert
33968 return type from void to rtx, returning x. When reading
33969 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
33970 larger size containing struct block_symbol.
33972 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
33974 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
33975 -mfloat128-hardware without -m64.
33977 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
33980 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
33981 entries to the case statement that marks unsigned arguments to
33982 overloaded functions.
33984 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
33986 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
33987 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
33989 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
33992 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
33993 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
33995 2017-03-10 Martin Liska <mliska@suse.cz>
33999 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
34000 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
34003 2017-03-10 Olivier Hainque <hainque@adacore.com>
34005 * tree-switch-conversion (array_value_type): Start by resetting
34006 candidate type to it's main variant.
34008 2017-03-10 Jakub Jelinek <jakub@redhat.com>
34010 PR rtl-optimization/79909
34011 * combine.c (try_combine): Use simplify_replace_rtx on individual
34012 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
34013 of the whole CALL_INSN_FUNCTION_USAGE.
34015 PR tree-optimization/79972
34016 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
34017 get_range_info on SSA_NAMEs. Formatting fixes.
34019 2017-03-10 Richard Biener <rguenther@suse.de>
34020 Jakub Jelinek <jakub@redhat.com>
34022 PR tree-optimization/77975
34023 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
34024 edge to be constant.
34025 (get_val_for): For constant x return it. Formatting fix.
34026 (loop_niter_by_eval): Avoid pointless looping if the next iteration
34027 would use the same bases as the current one.
34029 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34031 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
34032 instead of vec_select for V1TImode.
34033 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
34035 (VSX_LE_128): Add V1TI to this mode iterator.
34036 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
34037 (*vsx_le_perm_store_<mode>): Likewise.
34038 (pre-reload splitter for VSX stores): Likewise.
34039 (post-reload splitter for VSX stores): Likewise.
34040 (*vsx_xxpermdi2_le_<mode>): Likewise.
34041 (*vsx_lxvd2x2_le_<mode>): Likewise.
34042 (*vsx_stxvd2x2_le_<mode>): Likewise.
34044 2017-03-09 Michael Eager <eager@eagercon.com>
34046 Correct failures with --enable-checking=yes,rtl.
34048 * config/microblaze/microblaze.c (microblaze_expand_shift):
34049 Replace GET_CODE test with CONST_INT_P and INTVAL test with
34050 test for const0_rtx.
34051 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
34052 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
34054 2017-03-09 Richard Biener <rguenther@suse.de>
34056 PR tree-optimization/79977
34057 * graphite-scop-detection.c (scop_detection::merge_sese):
34058 Handle the case of extra exits to blocks dominating the entry.
34060 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
34062 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
34065 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
34067 PR rtl-optimization/79949
34068 * lra-constraints.c (process_alt_operands): Check memory when
34069 trying to predict a cycle. Print about the overall increase.
34071 2017-03-09 Richard Biener <rguenther@suse.de>
34073 PR middle-end/79971
34074 * gimple-expr.c (useless_type_conversion_p): Preserve
34075 TYPE_SATURATING for fixed-point types.
34077 2017-03-09 Richard Biener <rguenther@suse.de>
34080 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
34081 alignment of BLKmode params.
34083 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34086 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
34087 (VALL_NO_V2Q): Likewise.
34089 * config/aarch64/aarch64-simd.md
34090 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
34092 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
34093 VALL_NO_V2Q mode iterator.
34094 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
34096 2017-03-09 Martin Liska <mliska@suse.cz>
34098 PR tree-optimization/79631
34099 * tree-chkp-opt.c (chkp_is_constant_addr): Call
34100 tree_int_cst_sign_bit just for INTEGER constants.
34102 2017-03-09 Martin Liska <mliska@suse.cz>
34106 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
34109 2017-03-09 Marek Polacek <polacek@redhat.com>
34112 * tree.c (inchash::add_expr): Handle TREE_VEC.
34114 2017-03-09 Martin Liska <mliska@suse.cz>
34117 (chkp_narrow_size_and_offset): New function.
34118 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
34119 (void chkp_parse_bit_field_ref): New function.
34120 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
34121 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
34123 2017-03-09 Martin Liska <mliska@suse.cz>
34126 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
34127 (chkp_find_bounds_1): Remove gcc_unreachable.
34129 2017-03-09 Jakub Jelinek <jakub@redhat.com>
34132 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
34133 BUILT_IN_SYNC*, determine the access type from the size suffix and
34134 always build a MEM_REF with that type. Handle forgotten
34135 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
34138 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
34139 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
34140 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
34141 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
34142 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
34143 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
34144 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
34145 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
34146 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
34147 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
34148 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
34149 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
34150 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
34151 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
34152 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
34153 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
34154 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
34155 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
34156 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
34157 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
34158 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
34159 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
34160 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
34161 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
34162 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
34163 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
34164 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
34165 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
34166 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
34167 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
34168 definitions outside of __OPTIMIZE__ guarded section.
34171 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
34172 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
34173 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
34174 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
34177 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34179 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
34180 ("vfenez<mode>"): Add missing constraints.
34182 2017-03-08 Martin Sebor <msebor@redhat.com>
34185 * config/nds32/nds32.c (nds32_option_override):
34186 Fix misspelled diagnostic.
34188 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34191 * gimplify.c (gimplify_omp_for): Replace index var in outer
34192 taskloop statement with an artificial variable and add
34193 OMP_CLAUSE_PRIVATE clause for it.
34195 2017-03-08 Richard Biener <rguenther@suse.de>
34197 PR tree-optimization/79955
34198 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
34199 for accesses that are completely outside of the variable.
34201 2017-03-08 Andrew Haley <aph@redhat.com>
34203 PR tree-optimization/79943
34204 * tree-ssa-loop-split.c (compute_new_first_bound): When
34205 calculating the new upper bound, (END-BEG) should be added, not
34208 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34210 * config/avr/avr.md (setmemhi): Make sure match_dup
34211 operand number comes before match_scratch.
34213 2017-03-08 Richard Biener <rguenther@suse.de>
34215 PR tree-optimization/79920
34216 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
34217 with ncopies == 1 to ...
34218 (vect_transform_slp_perm_load): ... here. Properly compute
34219 all element loads by iterating VF times over the group. Do
34220 not handle ncopies (computed in a broken way) in
34221 vect_create_mask_and_perm.
34223 2017-03-08 Jakub Jelinek <jakub@redhat.com>
34226 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
34227 is a uniform vector, use uniform_vector_p return value instead of
34228 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
34230 2017-03-07 Marek Polacek <polacek@redhat.com>
34232 PR middle-end/79809
34233 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
34234 (alloca_call_type): Likewise.
34236 2017-03-07 Martin Liska <mliska@suse.cz>
34238 * gcov.c (process_args): Put comment to correct location.
34240 2017-03-07 Martin Liska <mliska@suse.cz>
34242 PR middle-end/68270
34243 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
34244 Use array_at_struct_end_p instead of DECL_CHAIN (field).
34245 (chkp_narrow_bounds_for_field): Likewise.
34246 (chkp_parse_array_and_component_ref): Pass one more argument to
34249 2017-03-07 Richard Biener <rguenther@suse.de>
34251 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
34254 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
34256 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
34257 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
34259 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34262 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
34263 to end of description.
34264 (PARAM_MAX_STORES_TO_MERGE): Likewise.
34266 2017-03-07 Jakub Jelinek <jakub@redhat.com>
34268 PR rtl-optimization/79901
34269 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
34271 (*avx512f_<code><mode>3<mask_name>): ... this.
34272 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
34273 iterator instead of VI8_AVX2_AVX512BW.
34275 PR rtl-optimization/79901
34276 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
34277 min/max expander, expand it using expand_vec_cond_expr.
34280 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
34283 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34286 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
34287 to void * for PCH reasons.
34288 * dwarf2out.c (output_loc_operands, output_die): Cast
34289 v.val_vec.array to unsigned char *.
34291 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
34294 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
34297 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
34299 PR rtl-optimization/79571
34300 * lra-constraints.c (process_alt_operands): Calculate static
34301 reject and subtract it from overall when only addresses will be
34304 2017-03-06 Julia Koval <julia.koval@intel.com>
34307 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
34308 incoming stack boundary to 128 for 64-bit targets.
34310 2017-03-06 Richard Biener <rguenther@suse.de>
34312 PR tree-optimization/79894
34313 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
34314 to NULL after folding it.
34316 2017-03-06 Richard Biener <rguenther@suse.de>
34318 PR tree-optimization/79824
34319 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
34320 check disabling peeling for gaps.
34322 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
34324 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
34325 attributes): Document gettimeofday.
34327 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34329 * config/s390/s390.c (s390_option_override_internal): Set
34330 PARAM_MIN_VECT_LOOP_BOUND
34332 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
34334 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
34335 * config/s390/s390.md: Likewise.
34337 2017-03-06 Jakub Jelinek <jakub@redhat.com>
34340 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
34341 (<avx2_avx512>_perm<mode>): Rename to ...
34342 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
34344 (<avx512>_perm<mode>_mask): Rename to ...
34345 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
34347 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
34348 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
34349 instead of VI8F_256_512.
34350 (avx512f_perm<mode>): New define_expand.
34351 (avx512f_perm<mode>_mask): Likewise.
34352 (avx512f_perm<mode>_1<mask_name>): New define_insn.
34353 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
34355 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34357 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
34358 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
34360 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
34362 2017-03-06 Martin Liska <mliska@suse.cz>
34365 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
34366 when having a SSA NAME w/o VAR_DECL assigned to it.
34368 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34370 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
34371 msa_dpsub_<su>_d): Fix MODE for vec_select.
34373 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
34375 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
34377 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
34379 2017-03-06 Richard Biener <rguenther@suse.de>
34381 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
34382 * plugin.c (register_plugin_info): Likewise.
34383 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
34385 2017-03-05 Jakub Jelinek <jakub@redhat.com>
34387 * config/i386/sse.md (sse_storehps, sse_storelps,
34388 avx_<castmode><avxsizesuffix>_<castmode>,
34389 avx512f_<castmode><avxsizesuffix>_<castmode>,
34390 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
34391 in condition that at least one operand is not a MEM.
34393 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34395 PR middle-end/79805
34396 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
34397 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
34399 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
34400 gimple_call_nothrow_p flag based on whether original builtin can throw.
34401 If it can, emit following stmts on the fallthrough edge.
34402 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
34403 don't create new bb if inserting just debug stmts on the edge, try to
34404 insert them on the fallthru bb or just reset debug stmts.
34406 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
34409 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
34410 restore recog_data (including the operand rtxes inside it) around
34411 the call to get_insn_template.
34413 2017-03-03 Martin Sebor <msebor@redhat.com>
34415 PR tree-optimization/79699
34416 * context.c (context::~context): Free MPFR caches to avoid
34417 a memory leak on program exit.
34419 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34421 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
34422 Use wide_int::ulow () instead of .elt (0).
34424 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34426 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
34427 (*pushxf): Limit oF constraint to 32bit targets and add oC
34428 constraint for 64bit targets.
34429 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
34430 (*pushdf): Change rmF constraint to rmC.
34432 2017-03-03 Martin Liska <mliska@suse.cz>
34434 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
34435 Remove unused variable.
34437 2017-03-03 Jakub Jelinek <jakub@redhat.com>
34440 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
34441 is a memory operand, increase num_memory.
34442 (ix86_expand_args_builtin): Likewise.
34444 2017-03-03 Jan Hubicka <jh@suse.cz>
34447 * ipa-devirt.c (maybe_record_node): Properly handle
34448 __cxa_pure_virtual visibility.
34450 2017-03-03 Martin Liska <mliska@suse.cz>
34452 PR tree-optimization/79803
34453 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
34455 (pass_loop_prefetch::execute): Disabled optimization if an
34456 assumption about L1 cache size is not met.
34458 2017-03-03 Martin Liska <mliska@suse.cz>
34460 PR rtl-optimization/79574
34461 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
34462 (hash_scan_set): Likewise.
34463 (dump_hash_table): Likewise.
34464 (hoist_code): Likewise.
34466 2017-03-03 Richard Biener <rguenther@suse.de>
34468 * fixed-value.c (fixed_from_string): Restore use of elt (1)
34469 in place of uhigh ().
34470 (fixed_convert_from_real): Likewise.
34472 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
34475 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
34477 2017-03-03 Richard Biener <rguenther@suse.de>
34479 PR middle-end/79818
34480 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
34481 TYPE_OVERFLOW_UNDEFINED check.
34483 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34485 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
34487 (vector_ae_<mode>_p): Likewise.
34488 (vector_nez_<mode>_p): Likewise.
34489 (vector_ne_v2di_p): Likewise.
34490 (vector_ae_v2di_p): Likewise.
34491 (vector_ne_<mode>_p): Likewise.
34492 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
34494 (vsx_tsqrt<mode>2_fe): Likewise.
34496 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
34499 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
34501 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34503 PR rtl-optimization/79780
34504 * cprop.c (one_cprop_pass): When second and further conditional trap
34505 in a single basic block is turned into an unconditional trap, turn it
34506 into a deleted note to avoid RTL verification failures.
34508 2017-03-02 Richard Biener <rguenther@suse.de>
34510 * fold-const.c (const_binop): Use ulow () instead of elt (0).
34512 2017-03-02 Richard Biener <rguenther@suse.de>
34514 PR tree-optimization/79345
34516 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
34517 param and abort the walk, returning -1 if it is hit.
34518 (walk_aliased_vdefs): Take a limit param and pass it on.
34519 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
34520 defaulting to 0 and return a signed int.
34521 * tree-ssa-uninit.c (struct check_defs_data): New struct.
34522 (check_defs): New helper.
34523 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
34524 about uninitialized memory.
34525 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
34526 bogus uninitialized warning.
34527 (fixed_convert_from_real): Likewise.
34529 2017-03-02 Bin Cheng <bin.cheng@arm.com>
34531 PR tree-optimization/66768
34532 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
34533 iv_use if base object can't be determined.
34535 2017-03-02 Jakub Jelinek <jakub@redhat.com>
34537 PR tree-optimization/79345
34538 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
34539 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
34540 (get_pattern_stats): Initialize it.
34541 * genemit.c (gen_expand): Verify match_scratch numbers come after
34542 match_operand/match_dup numbers.
34543 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
34544 match_scratch numbers.
34545 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
34547 * config/s390/s390.md (trunctdsd2): Likewise.
34549 2017-03-02 Richard Biener <rguenther@suse.de>
34551 * wide-int.h (wide_int_storage::operator=): Implement in terms
34554 2017-03-02 Richard Biener <rguenther@suse.de>
34556 PR tree-optimization/79777
34557 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
34558 the to insert expression to sth existing.
34560 2017-03-01 Martin Sebor <msebor@redhat.com>
34562 PR middle-end/79692
34563 * gimple-ssa-sprintf.c
34564 (directive::known_width_and_precision): New function.
34565 (format_integer): Use it.
34566 (get_mpfr_format_length): Consider the full range of precision
34567 when computing %g output with the # flag. Set the likely byte
34568 count to 3 rather than 1 when precision is indeterminate.
34569 (format_floating): Correct the lower bound of precision.
34571 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34573 * doc/invoke.texi: Document default code model for 64-bit Linux.
34575 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34578 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
34579 udiv rather than div since input pattern is unsigned.
34581 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
34583 * config/i386/i386.c (print_reg): Warn for values of
34584 unsupported size in integer register.
34586 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
34589 * config/rs6000/predicates.md (current_file_function_operand): Do
34590 not allow self calls to be local if the function is replaceable.
34592 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
34595 * config/rs6000/altivec.h (vec_ctz and others): Change the
34596 preprocessor macro that controls conditional compilation from
34597 _ARCH_PWR9 to __POWER9_VECTOR__.
34598 (vec_all_ne): Change parameterization of __altivec_scalar_pred
34599 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
34600 control (instead of _ARCH_PWR9 control) so that template
34601 definition uses power9-specific function.
34602 (vec_any_eq): Likewise.
34603 (vec_all_ne): Change macro definition to use a power9-specific
34604 expansion under #ifdef __POWER9_VECTOR__ control (instead of
34605 _ARCH_PWR9 control).
34606 (vec_any_eq) Likewise.
34607 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
34608 expansion for CMPNEF to remove support for xvcmpnesp instruction.
34609 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
34610 support for xvcmpnedp instruction.
34611 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
34612 macro expansion so that Power9 implementation of vec_all_ne does
34613 not use the AltiVec predicate framework.
34614 (VCMPNEH_P): Likewise.
34615 (VCMPNEW_P): Likewise.
34616 (VCMPNED_P): Likewise.
34617 (VCMPNEFP_P): Likewise.
34618 (VCMPNEDP_P): Likewise.
34619 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
34620 implementation of vec_any_eq to not use AltiVec predicate
34622 (VCMPAEH_P): Likewise.
34623 (VCMPAEW_P): Likewise.
34624 (VCMPAED_P): Likewise.
34625 (VCMPAEFP_P): Likewise.
34626 (VCMPAEDP_P): Likewise.
34627 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
34628 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
34629 not use the AltiVec predicate framework.
34630 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
34631 of vec_any_eq to not use AltiVec predicate framework.
34632 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
34633 support for predefined __POWER9_VECTOR__ macro to indicate that
34634 Power9 instruction selection is enabled.
34635 (altivec_overloaded_builtins): Remove extraneous
34636 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
34637 function argument types RS6000_BTI_bool_V16QI and
34638 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
34639 entry for overloaded function argument types RS6000_BTI_bool_V4SI
34640 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
34641 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
34642 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
34643 Power9 for implementations of vec_cmpne. Change the signature for
34644 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
34645 (representing vec_all_ne) to remove the previously described first
34646 argument of type RS6000_BTI_INTSI, as this was an artifact of
34647 reliance on the AltiVec predicate framework, which is no longer
34648 used in the implementation of these functions. Add
34649 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
34650 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
34651 since, unlike the AltiVec predicate framework implementation, we
34652 do not share function descriptors between vec_alle and vec_anyeq.
34653 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
34654 set of modes that receive special treatment even when
34655 TARGET_P9_VECTOR is true. The special treatment emits code that
34656 does not depend on Power9 instructions.
34657 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
34658 define_expand to not rely on AltiVec predicate framework.
34659 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34661 (vector_ne_v2di_p): Change this define_expand to not rely on
34662 AltiVec predicate framework.
34663 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
34665 (vector_ne_<mode>_p): Change this define_expand to not rely on
34666 AltiVec predicate framework.
34667 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
34669 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
34670 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
34671 define_insn pattern.
34672 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
34673 define_insn pattern because the xvcmpne<VSs>. instruction is not
34675 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
34676 instruction is not supported.
34678 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34680 * config/nvptx/nvptx.c: Include intl.h.
34682 2017-03-01 Martin Jambor <mjambor@suse.cz>
34685 * ipa-prop.h (ipa_bits): Removed field known.
34686 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
34687 to pointers. Adjusted their comments to warn about their sharing.
34688 (ipcp_transformation_summary): Change bits to a vector of pointers.
34689 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
34690 (ipa_get_ipa_bits_for_value): Declare.
34691 * tree-vrp.h (value_range): Mark as GTY((for_user)).
34692 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
34693 (ipa_bits_hash_table): Likewise.
34694 (ipa_vr_ggc_hash_traits): Likewise.
34695 (ipa_vr_hash_table): Likewise.
34696 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
34697 being pointers and vr_known being removed.
34698 (ipa_set_jf_unknown): Likewise.
34699 (ipa_get_ipa_bits_for_value): New function.
34700 (ipa_set_jfunc_bits): Likewise.
34701 (ipa_get_value_range): New overloaded functions.
34702 (ipa_set_jfunc_vr): Likewise.
34703 (ipa_compute_jump_functions_for_edge): Use the above functions to
34704 construct bits and vr parts of jump functions.
34705 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
34706 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34708 (ipcp_grow_transformations_if_necessary): Also allocate
34709 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
34711 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
34712 them. Fix too long lines.
34713 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
34714 vr_known being removed.
34715 (ipa_read_jump_function): Use new setter functions to construct bits
34716 and vr parts of jump functions or set them to NULL.
34717 (write_ipcp_transformation_info): Adjust for bits being pointers.
34718 (read_ipcp_transformation_info): Likewise.
34719 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
34721 Include gt-ipa-prop.h.
34722 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
34724 (ipcp_store_bits_results): Likewise.
34725 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
34726 Do not write to existing jump functions but use a temporary instead.
34728 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34731 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
34732 attempt to use its first operand as BIT_FIELD_REF base.
34734 2017-03-01 Richard Biener <rguenther@suse.de>
34736 PR middle-end/79721
34737 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
34738 interpolating formula in wrapping arithmetic.
34739 (chrec_apply): Convert chrec_evaluate return value to wanted type.
34741 2017-03-01 Jakub Jelinek <jakub@redhat.com>
34743 PR tree-optimization/79734
34744 * tree-vect-generic.c (expand_vector_condition): Optimize
34745 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
34746 Handle VEC_COND_EXPR where comparison has different inner width from
34747 type's inner width.
34749 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
34751 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
34752 markup, and similar issues. Remove @opindex entries for things
34753 that aren't options. Add missing -mmpy-option entries.
34755 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34757 PR tree-optimization/79737
34758 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
34759 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
34760 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
34761 instead of byte_size. Formatting fix.
34762 (shift_bytes_in_array_right): Formatting fix.
34764 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
34767 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
34768 condition on optimize for the leaf function test.
34770 2017-02-28 Martin Liska <mliska@suse.cz>
34773 * read-rtl-function.c (function_reader::handle_unknown_directive):
34774 Bail out when one uses -flto.
34776 2017-02-28 Martin Liska <mliska@suse.cz>
34778 * common.opt: Replace space with tabular for options of <number>
34780 * config/i386/i386.opt: Show <number> value for
34781 -mlarge-data-threshold.
34782 * opts.c (print_filtered_help): Do not display number in hexadecimal
34785 2017-02-28 Martin Liska <mliska@suse.cz>
34787 * common.opt: Fix --help=option -Q for options which are of
34790 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
34792 * config/i386/i386.c (print_reg): Error out for values
34793 of 8-bit size in invalid integer register.
34795 2017-02-28 Martin Sebor <msebor@redhat.com>
34797 PR tree-optimization/79691
34798 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
34800 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34803 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
34804 gcc_unreachable with output_operand_lossage.
34806 2017-02-28 Richard Biener <rguenther@suse.de>
34808 PR tree-optimization/79740
34809 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
34811 (visit_nary_op): Insert the nary into the hashtable if we
34812 pattern-matched sth.
34813 * tree-ssa-pre.c (eliminate_insert): Robustify.
34815 2017-02-28 Richard Biener <rguenther@suse.de>
34817 PR middle-end/79731
34818 * fold-const.c (decode_field_reference): Reject out-of-bound
34821 2017-02-28 Jakub Jelinek <jakub@redhat.com>
34823 * config/i386/i386.c: Include intl.h.
34824 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
34825 instead of just cond ? "..." : "...".
34826 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
34827 * coverage.c (read_counts_file): Likewise.
34828 * omp-offload.c: Include intl.h.
34829 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
34830 of just cond ? "..." : "...".
34831 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
34832 of just cond ? "..." : "...".
34834 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
34837 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
34839 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
34840 'tune for' CPU name.
34841 * config/arm/arm-cpu-data.h: Regenerated.
34843 2017-02-28 Richard Biener <rguenther@suse.de>
34845 PR tree-optimization/79732
34846 * tree-inline.c (expand_call_inline): Do not shadow var.
34848 2017-02-28 Richard Biener <rguenther@suse.de>
34850 PR tree-optimization/79723
34851 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
34852 address-space properly.
34854 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
34856 * doc/optinfo.texi (Optimization groups): Fix option used for
34858 * doc/invoke.texi (-fopt-info): Document "omp".
34859 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
34860 (OPTGROUP_ALL): Add OPTGROUP_OMP.
34861 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
34862 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
34863 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
34865 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
34867 * dumpfile.c (optgroup_options): Instead of "openmp", associate
34868 OPTGROUP_OMP with "omp".
34870 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
34873 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
34874 for arithmetic shift of unsigned V2DI.
34876 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
34878 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
34879 arc/linux.h headers.
34880 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
34881 (LINK_SPEC): Likewise.
34882 (ARC_TLS_EXTRA_START_SPEC): Likewise.
34883 (EXTRA_SPECS): Likewise.
34884 (STARTFILE_SPEC): Likewise.
34885 (ENDFILE_SPEC): Likewise.
34886 (LIB_SPEC): Likewise.
34887 (TARGET_SDATA_DEFAULT): Likewise.
34888 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
34889 (MULTILIB_DEFAULTS): Likewise.
34890 (DWARF2_UNWIND_INFO): Likewise.
34891 * config/arc/big.h: New file.
34892 * config/arc/elf.h: Likewise.
34893 * config/arc/linux.h: Likewise.
34894 * config/arc/t-uClibc: Remove.
34896 2017-02-27 Bin Cheng <bin.cheng@arm.com>
34898 PR tree-optimization/77536
34899 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
34900 (tree_transform_and_unroll_loop): Use above function to compute the
34901 estimated niter of unrolled loop and use it when scaling profile.
34902 Also use count info rather than frequency if it's non-zero.
34903 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
34904 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
34905 (vect_transform_loop): Call above function.
34907 2017-02-27 Richard Biener <rguenther@suse.de>
34909 PR tree-optimization/45397
34910 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
34911 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
34912 (visit_nary_op): Add pattern matching for CSEing sign-changed
34913 or truncated operations with wider ones.
34915 2017-02-27 Richard Biener <rguenther@suse.de>
34917 PR tree-optimization/79690
34918 * tree-vect-stmts.c (vectorizable_store): Use vector type
34919 built from the DR with address-space.
34921 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
34923 * doc/invoke.texi (Optimize Options): Refine the description
34924 of asan-use-after-return.
34926 2017-02-25 Alan Modra <amodra@gmail.com>
34928 PR rtl-optimization/79584
34929 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
34930 base, not ad->base_term, the reg within base. Remove assertion
34931 that ad->base == ad->base_term. Replace gen_int_mode using
34932 bogus mode with const0_rtx.
34934 2017-02-25 Jakub Jelinek <jakub@redhat.com>
34936 PR middle-end/79396
34937 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
34938 FMA_EXPR like tcc_binary or tcc_unary.
34940 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
34943 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
34945 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
34946 (output_loc_operands): Handle DW_OP_call_ref and
34947 DW_OP_GNU_variable_value.
34948 (struct variable_value_struct): New type.
34949 (struct variable_value_hasher): Likewise.
34950 (variable_value_hash): New variable.
34951 (string_types): Remove.
34952 (copy_loc_descr): New function.
34953 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
34954 (prepend_loc_descr_to_each): New function.
34955 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
34956 instead of add_loc_descr_to_each if the first argument is single
34957 location list and the second has multiple.
34958 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
34959 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
34960 when looking for variable value which doesn't have other location info.
34961 (loc_list_from_tree): Formatting fix.
34962 (gen_array_type_die): Simplify DW_AT_string_length handling.
34963 (adjust_string_types): Remove.
34964 (gen_subprogram_die): Don't call adjust_string_types nor test/set
34965 string_types. Call resolve_variable_values.
34966 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
34967 (resolve_addr_in_expr): Likewise. Add A argument.
34968 (copy_deref_exprloc): Remove deref argument. Adjust for the
34969 original expression being DW_OP_GNU_variable_value with optionally
34970 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
34971 optionally after it.
34972 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
34973 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
34974 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
34975 (variable_value_hasher::hash, variable_value_hasher::equal): New
34977 (resolve_variable_value_in_expr, resolve_variable_value,
34978 resolve_variable_values, note_variable_value_in_expr,
34979 note_variable_value): New functions.
34980 (dwarf2out_early_finish): Call note_variable_value on all toplevel
34983 2017-02-24 Jakub Jelinek <jakub@redhat.com>
34986 * opts.h (handle_generated_option): Add GENERATED_P argument.
34987 * opts-common.c (handle_option): Adjust function comment.
34988 (handle_generated_option): Add GENERATED_P argument, pass it to
34990 (control_warning_option): Pass false to handle_generated_option
34992 * opts.c (maybe_default_option): Pass true to handle_generated_option
34994 * optc-gen.awk: Likewise.
34996 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
34998 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
34999 a REG, look at the REG it is a SUBREG of.
35000 (splitter for cmpeqsi_t): Ditto.
35002 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35004 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
35005 the special USEs with the pattern of the insn, not the insn itself.
35007 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
35010 * doc/invoke.texi: Document -mload-store-pairs.
35012 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35013 Sandra Loosemore <sandra@codesourcery.com>
35015 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
35016 argument isn't a CONST_INT.
35017 (nios2_alternate_compare_const): Assert op is a CONST_INT.
35018 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
35019 (nios2_validate_compare): Bypass alternate compare logic if *op2
35020 is not a CONST_INT.
35021 (ldstwm_operation_p): Return false if first_base is not a REG or
35022 if first_offset is not a CONST_INT.
35024 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35026 * config/cris/cris.md: Use correct operand in a define_peephole2.
35028 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35030 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
35032 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
35034 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
35035 this_insn if it is an INSN or JUMP_INSN.
35036 (force_offsettable): Look at base, not at addr.
35037 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
35038 on things that aren't necessarily CONST_INTs.
35040 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
35042 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
35043 -mfpmath=sse is the default also for x86-32 targets with SSE2
35044 instruction set when @option{-ffast-math} is enabled
35046 2017-02-24 Jeff Law <law@redhat.com>
35048 PR rtl-optimizatoin/79286
35049 * ira.c (update_equiv_regs): Drop may_trap_p exception to
35052 2017-02-24 Richard Biener <rguenther@suse.de>
35054 PR tree-optimization/79389
35055 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
35058 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
35060 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
35061 function comment to reflect reality.
35062 (loop_exits_before_overflow): Fix typo in function description.
35064 2017-02-24 Richard Biener <rguenther@suse.de>
35066 PR tree-optimization/79389
35067 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
35068 properly that a threading opportunity exists. Detect conditional
35069 copy/constant propagation opportunities.
35071 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
35073 * config/visium/visium.md (type): Add trap.
35074 (b): New mode attribute.
35075 (*btst): Rename into...
35076 (*btst<mode>): ...this and adjust.
35077 (*cbranchsi4_btst_insn): Rename into...
35078 (*cbranch<mode>4_btst_insn): ...this and adjust.
35079 (trap): New define_insn.
35081 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35083 PR tree-optimization/79389
35084 * ifcvt.c (struct noce_if_info): Add rev_cond field.
35085 (noce_reversed_cond_code): New function.
35086 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
35087 reversed_comparison_code. Formatting fix.
35088 (noce_try_store_flag): Test rev_cond != NULL in addition to
35089 reversed_comparison_code.
35090 (noce_try_store_flag_constants): Likewise.
35091 (noce_try_store_flag_mask): Likewise.
35092 (noce_try_addcc): Use rev_cond if non-NULL instead of
35093 reversed_comparison_code.
35094 (noce_try_cmove_arith): Likewise. Formatting fixes.
35095 (noce_try_minmax, noce_try_abs): Clear rev_cond.
35096 (noce_find_if_block): Initialize rev_cond.
35097 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
35098 instead of false as last argument never attempt to reverse it
35101 2017-02-23 Bin Cheng <bin.cheng@arm.com>
35103 PR tree-optimization/79663
35104 * tree-predcom.c (combine_chains): Process refs in reverse order
35105 only for ZERO length chains, and add explaining comment.
35107 2017-02-23 Jeff Law <law@redhat.com>
35109 PR tree-optimization/79578
35110 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
35111 in call to operand_equal_p.
35113 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
35116 * config/i386/cpuid.h: Fix another undefined behavior.
35118 2017-02-23 Richard Biener <rguenther@suse.de>
35120 PR tree-optimization/79683
35121 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
35122 vector types for data-refs.
35124 2017-02-23 Martin Liska <mliska@suse.cz>
35126 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
35128 2017-02-23 Jakub Jelinek <jakub@redhat.com>
35130 PR middle-end/79665
35131 * internal-fn.c (get_range_pos_neg): Moved to ...
35132 * tree.c (get_range_pos_neg): ... here. No longer static.
35133 * tree.h (get_range_pos_neg): New prototype.
35134 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
35135 are known to be in between 0 and signed maximum inclusive, try to
35136 expand both unsigned and signed divmod and use the cheaper one from
35139 2017-02-22 Jeff Law <law@redhat.com>
35141 PR tree-optimization/79578
35142 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
35143 to compare base operands.
35145 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
35148 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
35149 gpc_reg_operand instead of fpr_reg_operand.
35151 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
35153 * config/mips/mips.c (mips_return_in_memory): Force FP
35154 vector types to be returned in memory for o32 ABI.
35156 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35158 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
35159 instead of DW_TAG_member for static data member declarations and don't
35160 set no_linkage_name for static inline data members.
35161 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
35164 2017-02-22 Martin Liska <mliska@suse.cz>
35166 * doc/invoke.texi: Replace inequality signs with square brackets
35169 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
35172 * lra-constraints.c (simplify_operand_subreg): Handle
35173 WORD_REGISTER_OPERATIONS targets.
35175 2017-02-22 Jakub Jelinek <jakub@redhat.com>
35178 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
35179 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
35180 elimination by swapping fld*.
35182 2017-02-22 Richard Biener <rguenther@suse.de>
35184 PR tree-optimization/79673
35185 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
35186 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
35187 irrelevant address-space qualifiers and avoiding a
35188 ADDR_SPACE_CONVERT_EXPR from fold_convert.
35190 2017-02-22 Richard Biener <rguenther@suse.de>
35192 PR tree-optimization/79666
35193 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
35194 to not symbolically negate if that may introduce undefined
35197 2017-02-22 Martin Liska <mliska@suse.cz>
35200 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
35201 * data-streamer-out.c (streamer_write_gcov_count_stream):
35203 * value-prof.c (stream_out_histogram_value): Make assert more
35204 precise based on type of counter.
35206 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
35209 * config/i386/i386.md (standard_x87sse_constant_load splitter):
35210 Use nonimmediate_operand instead of memory_operand for operand 1.
35211 (float-extend standard_x87sse_constant_load splitter): Ditto.
35213 2017-02-21 Jeff Law <law@redhat.com>
35215 PR tree-optimization/79621
35216 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
35217 blocks with edges to themselves.
35219 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35222 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
35223 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
35224 Use gimple_call_builtin_p.
35227 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
35228 on temporarily removed DEBUG_INSNs.
35230 PR tree-optimization/79649
35231 * tree-loop-distribution.c (classify_partition): Give up on
35232 non-generic address space loads/stores.
35234 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
35236 * doc/loop.texi (Loop manipulation): Remove nonexistent
35237 tree_ssa_loop_version from the documentation.
35238 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
35240 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35243 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
35244 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
35245 * config/rs6000/rs6000.c: Include except.h.
35246 (rs6000_expand_split_stack_prologue): Call
35247 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
35249 2017-02-21 Martin Jambor <mjambor@suse.cz>
35252 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
35253 have been analyzed.
35255 2017-02-21 Martin Jambor <mjambor@suse.cz>
35257 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
35258 for backward compatibility only.
35259 * doc/invoke.texi (Option Summary): Remove all references to
35260 -fipa-cp-alignment.
35262 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
35266 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35268 * lra-constraints.c (curr_insn_transform): Handle
35269 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35271 2017-02-21 Martin Liska <mliska@suse.cz>
35273 * config/i386/i386.opt: Replace -masm-dialect with -masm.
35275 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
35277 PR translation/79638
35278 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
35280 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
35283 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
35284 (arm_function_ok_for_sibcall): Return false for an indirect call by
35285 descriptor if all the argument registers are used.
35286 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
35287 alignment of the function.
35289 2017-02-21 Jakub Jelinek <jakub@redhat.com>
35291 PR tree-optimization/61441
35292 * simplify-rtx.c (simplify_const_unary_operation): For
35293 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
35294 the sNaN unmodified.
35296 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
35298 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
35299 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
35300 instead of SYSTEM_HEADER_DIR.
35302 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
35303 Martin Liška <mliska@suse.cz>
35305 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
35306 Fix typos and grammar, use active voice, and clarify.
35308 2017-02-20 Marek Polacek <polacek@redhat.com>
35310 PR middle-end/79537
35311 * gimplify.c (gimplify_expr): Handle unused *&&L;.
35314 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
35316 2017-02-20 Jakub Jelinek <jakub@redhat.com>
35319 * config/i386/i386.c (ix86_expand_builtin): Handle
35320 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
35321 ix86_builtins_isa[fcode].isa as a requirement of those
35322 flags and any other flag in the bitmask.
35323 (ix86_init_mmx_sse_builtins): Use 0 instead of
35324 ~OPTION_MASK_ISA_64BIT as mask.
35325 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
35326 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
35327 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
35328 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
35330 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35333 * lra-constraints.c (split_reg): Check requested split mode
35334 is supported by the register.
35336 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35338 * lra-constraints.c (simplify_operand_subreg): Remove early
35341 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35344 * lra-constraints.c (curr_insn_transform): Tighten condition
35345 for converting SUBREG reloads from OP_OUT to OP_INOUT.
35347 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
35350 * lra-constraints.c (curr_insn_transform): Handle
35351 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
35353 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
35356 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
35358 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
35360 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
35363 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
35366 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35368 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
35371 2017-02-18 Jakub Jelinek <jakub@redhat.com>
35373 * final.c (last_columnnum, override_columnnum): New variables.
35374 (final_start_function): Set last_columnnum, pass it to begin_prologue
35375 hook and pass 0 to dwarf2out_begin_prologue.
35376 (final_scan_insn): Update override_columnnum. Pass last_columnnum
35377 to source_line debug hook.
35378 (notice_source_line): Compute last_columnnum and for debug_column_info
35379 return true on column changes.
35380 * debug.h (struct gcc_debug_hooks): Add column argument to
35381 source_line and begin_prologue hooks.
35382 (debug_nothing_int_charstar_int_bool): Remove prototype.
35383 (debug_nothing_int_int_charstar,
35384 debug_nothing_int_int_charstar_int_bool): New prototypes.
35385 (dwarf2out_begin_prologue): Add column argument.
35386 * debug.c (do_nothing_debug_hooks): Adjust source_line and
35387 begin_prologue hooks.
35388 (debug_nothing_int_charstar_int_bool): Remove.
35389 (debug_nothing_int_int_charstar,
35390 debug_nothing_int_int_charstar_int_bool): New functions.
35391 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
35392 through to dwarf2out_source_line.
35393 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
35394 (dwarf2out_source_line): Add column argument, emit it if requested.
35395 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
35397 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35398 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
35399 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
35400 through to dwarf2out_begin_prologue.
35401 (vmsdbgout_source_line): Add column argument, pass it through to
35402 dwarf2out_source_line.
35403 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
35404 dbxout_source_line caller.
35405 (dbxout_source_line): Add column argument.
35407 * common.opt (gno-column-info, gcolumn-info): New options.
35408 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
35409 (check_die): Also test for multiple DW_AT_decl_column attributes.
35410 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
35411 DW_AT_decl_column if requested.
35412 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
35414 (gen_variable_die): Likewise.
35415 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
35416 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
35419 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
35420 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
35421 (ix86_handle_option): Handle OPT_m3dnowa.
35422 * doc/invoke.texi (-m3dnowa): Document.
35423 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
35424 -m3dnowa instead of -m3dnow -march=athlon.
35427 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
35428 instead of gcc_assert for K, r and R code checks. Formatting fixes.
35430 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35433 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
35434 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
35435 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
35436 generator for vsx_xxpermdi_<mode>_be.
35437 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
35438 force big-endian semantics.
35439 (vsx_xxpermdi_<mode>_be): New define_expand with same
35440 implementation as previous version of vsx_xxpermdi_<mode>.
35442 2017-02-17 Jakub Jelinek <jakub@redhat.com>
35444 PR tree-optimization/79327
35445 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
35446 variable, its initialization and use.
35448 2017-02-17 Julia Koval <julia.koval@intel.com>
35450 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
35451 (OPTION_MASK_ISA_PKU_UNSET): New.
35452 (ix86_handle_option): Handle -mrdpid.
35453 * config/i386/cpuid.h (bit_RDPID): New.
35454 * config/i386/driver-i386.c (host_detect_local_cpu):
35455 Detect RDPID feature.
35456 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
35457 * config/i386/i386-c.c (ix86_target_macros_internal):
35459 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
35460 (ix86_valid_target_attribute_inner_p): Add "rdpid".
35461 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
35462 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
35463 * config/i386/i386.md (define_insn "rdpid"): New.
35464 * config/i386/i386.opt Add -mrdpid.
35465 * config/i386/immintrin.h (_rdpid_u32): New.
35467 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
35469 PR rtl-optimization/79541
35470 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
35471 instead of transforming it into USE.
35473 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
35475 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
35476 If HONOR_SNANS (SFmode) force the input to a register.
35477 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
35478 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
35479 an frsp or similar insn.
35481 2017-02-17 Martin Liska <mliska@suse.cz>
35483 PR rtl-optimization/79577
35484 * params.def (selsched-max-sched-times): Increase minimum to 1.
35486 2017-02-17 Martin Liska <mliska@suse.cz>
35488 PR rtl-optimization/79574
35489 * gcse.c (want_to_gcse_p): Prevent integer overflow.
35491 2017-02-17 Martin Liska <mliska@suse.cz>
35493 PR tree-optimization/79529
35494 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
35495 ssa_defined_default_def_p to handle cases which are implicitly
35497 * tree-ssa.c (ssa_defined_default_def_p): New function.
35498 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
35499 which are implicitly defined.
35500 * tree-ssa.h (ssa_defined_default_def_p): Declare.
35502 2017-02-17 Richard Biener <rguenther@suse.de>
35504 PR middle-end/79576
35505 * params.def (max-ssa-name-query-depth): Limit to 10.
35507 2017-02-17 Richard Biener <rguenther@suse.de>
35509 PR tree-optimization/79552
35510 * tree-ssa-structalias.c (visit_loadstore): Properly verify
35513 2017-02-17 Richard Biener <rguenther@suse.de>
35516 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
35518 2017-02-17 Marek Polacek <polacek@redhat.com>
35520 PR middle-end/79536
35521 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
35522 (fold_negate_expr): New wrapper.
35524 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
35526 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
35527 Correct terminology and de-emphasize pre-standard behavior.
35529 2017-02-16 Alan Modra <amodra@gmail.com>
35531 PR rtl-optimization/79286
35532 * ira.c (def_dominates_uses): New function.
35533 (update_equiv_regs): Don't create an equivalence for insns that
35534 may trap where the register def does not dominate the use.
35536 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
35538 PR rtl-optimization/78127
35539 * lra.c (lra): Call lra_eliminate before finish the loop after
35542 2017-02-16 Richard Biener <rguenther@suse.de>
35544 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
35546 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
35547 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
35548 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
35549 (isl_val_int_from_wi): New function.
35550 (extract_affine_gmp): Rename to ...
35551 (extract_affine_wi): ... this, take a widest_int.
35552 (extract_affine_int): Just wrap extract_affine_wi.
35553 (add_param_constraints): Use isl_val_int_from_wi.
35554 (add_loop_constraints): Likewise, and extract_affine_wi.
35556 2017-02-15 Jeff Law <law@redhat.com>
35558 PR middle-end/79521
35559 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
35560 ira_init_register_move_cost_if_necessary.
35562 2017-02-15 Martin Sebor <msebor@redhat.com>
35564 PR middle-end/32003
35565 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
35566 removed in a prior commit.
35568 2017-02-15 Bin Cheng <bin.cheng@arm.com>
35570 PR tree-optimization/79347
35571 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
35572 counters during peeling.
35574 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
35576 * Makefile.in (site.exp): Remove "set ISLVER".
35578 2017-02-15 Jakub Jelinek <jakub@redhat.com>
35581 * real.c (real_from_integer): Call real_convert even for decimal.
35583 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
35586 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
35588 2017-02-14 Andrew Pinski <apinski@cavium.com>
35590 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
35591 cores and change the partno/implementer to be correct.
35592 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
35593 the 'B" as the implementer.
35594 * config/aarch64/aarch64-tune.md: Regenerate.
35596 2017-02-14 Carl Love <cel@us.ibm.com>
35598 * config/rs6000/rs6000.c: Add case statement entry to make the
35599 xvcvuxdsp built-in argument unsigned.
35600 * config/rs6000/vsx.md: Fix the source and return operand types so they
35601 match the instruction definitions from the ISA document. Fix typo
35602 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
35605 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
35608 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
35609 member early_clobber_alts.
35610 * lra-lives.c (reg_early_clobber_p): New.
35611 (process_bb_lives): Use it.
35612 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
35613 (debug_operand_data): Initialize early_clobber_alts.
35614 (setup_operand_alternative): Set up early_clobber_alts.
35615 (collect_non_operand_hard_regs): Ditto. Pass early clobber
35616 alternatives to new_insn_reg.
35617 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
35619 (lra_update_insn_regno_info): Pass the new arg.
35621 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35623 PR middle-end/79505
35624 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
35625 (new_oacc_loop_raw): Don't clear already cleared fields.
35628 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
35629 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
35630 _mm512_prefetch_i64gather_ps): New inline functions and macros.
35632 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
35635 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
35637 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
35640 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
35641 the extra instruction to the right place to store 128-bit constant
35644 2017-02-14 Martin Sebor <msebor@redhat.com>
35646 PR middle-end/79448
35647 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
35648 warning for strings of unknown length.
35650 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
35652 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
35654 2017-02-14 Jeff Law <law@redhat.com>
35657 * ira-costs.c (scan_one_insn): Initialize register move costs
35658 for pseudos seen in USE/CLOBBER insns.
35660 PR tree-optimization/79095
35661 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
35662 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
35663 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
35664 if the operands are known to be not equal, then the resulting range
35666 (intersect_ranges): If the new range is ~[0,0] and the old range is
35667 wide, then prefer ~[0,0].
35668 * tree-vrp.c (overflow_comparison_p_1): New function.
35669 (overflow_comparison_p): New function.
35670 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
35671 if NAME is used in an overflow test.
35672 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
35673 overflow check that can be expressed as an equality test, then adjust
35674 ops to be that equality test.
35676 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35678 * config/s390/s390-builtin-types.def: Remove flags argument.
35679 * config/s390/s390.c (s390_init_builtins): Likewise.
35681 2017-02-14 Martin Liska <mliska@suse.cz>
35683 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
35684 vector. Fix trailing white spaces.
35686 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
35688 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
35691 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35693 PR rtl-optimization/68664
35694 * config/arm/arm.c (arm_sched_can_speculate_insn):
35695 New function. Declare prototype.
35696 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35698 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35700 PR rtl-optimization/68664
35701 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
35703 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
35705 2017-02-14 Amit Pawar <amit.pawar@amd.com>
35707 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
35708 max skip bytes for function, loop and jump.
35710 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
35712 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
35713 ABS_EXPR for gimple dump.
35715 2017-02-14 Jakub Jelinek <jakub@redhat.com>
35718 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
35720 PR tree-optimization/79408
35721 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
35722 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
35723 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
35724 also if rhs1 is INTEGER_CST.
35726 2017-02-14 Richard Biener <rguenther@suse.de>
35728 PR middle-end/79432
35729 * tree-into-ssa.c (insert_phi_nodes): When the function can
35730 have abnormal edges rewrite SSA names with broken use-def
35731 dominance out of SSA and register them for PHI insertion.
35733 2017-02-13 Martin Sebor <msebor@redhat.com>
35735 PR middle-end/79496
35736 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
35737 clearing info.nowrite flag when snprintf size argument is a range.
35739 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35741 * cprop.c (cprop_jump): Add missing space in string literal.
35742 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
35743 (get_constraint_for_component_ref): Likewise.
35744 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
35745 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
35746 * lra-constraints.c (process_alt_operands): Likewise.
35747 * ipa-inline.c (inline_small_functions): Likewise.
35748 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
35749 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
35750 * trans-mem.c (diagnose_tm_1_op): Likewise.
35751 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
35752 (grid_parallel_clauses_gridifiable): Likewise.
35754 * config/nvptx/mkoffload.c (process): Add space in between
35757 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
35758 "MOD4_SSE_REGS" and "ALL_REGS".
35760 * spellcheck.c (test_data): Add , in between "foo" and "food".
35762 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35765 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
35766 boundary crossing check and subsequent code generation agree.
35768 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35770 * config/aarch64/aarch64.c (has_memory_op): Delete.
35771 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
35774 2017-02-13 Jakub Jelinek <jakub@redhat.com>
35776 PR rtl-optimization/79388
35777 PR rtl-optimization/79450
35778 * combine.c (distribute_notes): When removing TEM_INSN for which
35779 corresponding dest has last value recorded, invalidate that last
35782 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35784 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
35785 of explicit '@'. Add missing assembly comment marker on branch costs
35788 2017-02-13 Nathan Sidwell <nathan@acm.org>
35790 * gengtype-lex.l (<in_struct>): Add '/'.
35792 2017-02-13 Martin Liska <mliska@suse.cz>
35795 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
35797 2017-02-13 Richard Biener <rguenther@suse.de>
35799 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
35801 * configure: Re-generate.
35802 * config.in: Likewise.
35803 * graphite-dependences.c: Simplify as if
35804 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
35805 * graphite-isl-ast-to-gimple.c: Likewise.
35806 * graphite-optimize-isl.c: Likewise.
35807 * graphite-poly.c: Likewise.
35808 * graphite-sese-to-poly.c: Likewise.
35809 * graphite.h: Likewise.
35810 * toplev.c: Include isl/version.h and use isl_version () for
35811 printing the ISL version.
35812 * doc/install.texi: Update ISL requirement.
35814 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35816 * doc/standards.texi (Standards): Update reference to
35819 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
35821 * doc/extend.texi (Named Address Spaces): sourceware.org now
35823 * doc/install.texi (Binaries): Ditto.
35826 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35828 * doc/cpp.texi: Replace "stringify"/"stringification" with C
35829 standard terminology "stringize"/"stringizing" throughout.
35830 * doc/cppinternals.texi: Likewise.
35832 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
35834 * doc/extend.texi: Fix some spelling mistakes and typos.
35835 * doc/invoke.texi: Likewise.
35837 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35840 * params.def (inline-min-speedup) Change from 10 to 8.
35842 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35844 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
35847 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
35850 * ipa-inline-analysis.c (get_minimal_bb): New function.
35851 (record_modified): Use it.
35852 (remap_edge_change_prob): Handle also ancestor functions.
35854 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
35856 * doc/contrib.texi (Contributors): Remove broken link into
35857 the Mauve CVS repository.
35859 2017-02-11 Jakub Jelinek <jakub@redhat.com>
35861 PR middle-end/79454
35862 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
35863 result computation whenever lhs doesn't have vector mode, not
35864 just when it has BLKmode.
35866 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35868 * doc/makefile.texi (profiledbootstrap): Refer to the
35869 installation instructions only in textual form.
35871 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35874 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
35876 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
35878 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
35879 (Specific): Update mingw-w64 reference.
35881 (Specific): Remove broken link to Renesas RX processor.
35883 2017-02-10 Richard Biener <rguenther@suse.de>
35885 * toplev.c (process_options): Do not mention obsolete graphite
35886 options when printing sorry message about missing graphite support.
35887 Mention -floop-nest-optimize.
35889 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
35891 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
35892 (vtst_p16): Likewise.
35893 (vtstq_p8): Likewise.
35894 (vtstq_p16): Likewise.
35896 (vtstq_p64): Likewise.
35897 * config/arm/arm_neon.h (vgetq_lane_p64): New.
35898 (vset_lane_p64): New.
35899 (vsetq_lane_p64): New.
35901 2017-02-10 Jakub Jelinek <jakub@redhat.com>
35903 PR tree-optimization/79411
35904 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
35905 stmt operands are SSA_NAMEs used in abnormal phis.
35906 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
35909 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35912 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
35915 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
35917 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
35919 2017-02-09 Jakub Jelinek <jakub@redhat.com>
35921 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
35925 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
35926 not arbitrary TREE_CONSTANT.
35929 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
35930 "omp declare target link" attribute unless is_global_var.
35931 * omp-offload.c (find_link_var_op): Likewise.
35933 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
35934 Chung-Lin Tang <cltang@codesourcery.com>
35936 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
35938 (gimplify_adjust_omp_clauses): Don't delete TILE.
35939 (gimplify_omp_for): Deal with TILE.
35940 * internal-fn.c (expand_GOACC_TILE): New function.
35941 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
35943 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
35944 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
35946 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
35947 avoid DIV for outermost collapse var.
35948 (expand_oacc_for): Insert tile element loop as needed. Adjust.
35949 Remove out of date comments, fix whitespace.
35950 * omp-general.c (omp_extract_for_data): Deal with tiling.
35951 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
35952 adjust OLF_DIM_BASE value.
35953 (struct omp_for_data): Add tiling field.
35954 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
35955 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
35956 for auto loops. Remove default auto determining, moved to
35957 oacc_loop_fixed_partitions.
35958 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
35959 stmts, add e_mask field.
35960 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
35961 (oacc_thread_numbers): Use oacc_dim_call.
35962 (oacc_xform_tile): New.
35963 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
35964 (finish_oacc_loop): Adjust for ifns vector.
35965 (oacc_loop_discover_walk): Append loop abstraction sites to list,
35966 add case for GOACC_TILE fns.
35967 (oacc_loop_xform_loop): Delete.
35968 (oacc_loop_process): Iterate over call list directly, and add
35969 handling for GOACC_TILE fns.
35970 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
35972 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
35973 vector partitioning to outer loops. Assign 2 partitions to loops
35974 when available. Add TILE handling.
35975 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
35976 (execite_oacc_device_lower): Process GOACC_TILE fns,
35977 ignore unknown specs.
35978 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
35979 * tree.c (omp_clause_num_ops): Adjust TILE ops.
35980 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
35982 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
35984 * configure.ac (ACX_BUGURL): Update.
35985 * configure: Regenerate.
35987 2017-02-09 Richard Biener <rguenther@suse.de>
35989 PR tree-optimization/69823
35990 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
35991 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
35993 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
35995 * config/arc/arc-c.def: Add __NPS400__ definition.
35996 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
35997 (TARGET_NPS400): Define.
35999 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
36001 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
36003 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
36004 pointer, arch_info.
36005 (arc_cpu_types): Fill the arch_info field with a pointer into the
36006 arc_arch_types table.
36007 (arc_selected_cpu): Declare.
36008 * config/arc/arc.c (arc_selected_cpu): Make global.
36009 (arc_selected_arch): Delete.
36010 (arc_base_cpu): Delete.
36011 (arc_override_options): Remove references to deleted variables,
36012 update access to arch information.
36013 (ARC_OPT): Update access to arch information.
36014 (ARC_OPTX): Likewise.
36015 * config/arc/arc.h (arc_base_cpu): Remove declaration.
36016 (TARGET_ARC600): Update access to arch information.
36017 (TARGET_ARC601): Likewise.
36018 (TARGET_ARC700): Likewise.
36019 (TARGET_EM): Likewise.
36020 (TARGET_HS): Likewise.
36021 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
36024 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
36027 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
36028 condition/operands for integer GE/LE/GEU/LEU operations.
36030 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
36032 PR translation/79397
36033 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
36036 2017-02-08 Martin Jambor <mjambor@suse.cz>
36039 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
36040 whether allocation happened.
36041 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
36042 nothing was allocated.
36044 2017-02-08 Jakub Jelinek <jakub@redhat.com>
36046 PR tree-optimization/79408
36047 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
36048 constant, but SSA_NAME with a known integer range, use the minimum
36049 of that range instead of op1 to determine if modulo can be replaced
36050 with its first operand.
36052 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36054 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
36056 2017-02-08 Richard Biener <rguenther@suse.de>
36058 PR tree-optimization/71824
36059 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36060 Check all loops contained in the merged region.
36062 2017-02-07 Andrew Pinski <apinski@cavium.com>
36064 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
36066 2017-02-07 Andrew Pinski <apinski@cavium.com>
36068 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
36069 (thunderxt88): Likewise.
36070 (thunderxt81): Disable LSE and change v8.1 to v8.
36071 (thunderxt83): Likewise.
36073 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36074 Richard Biener <rguenther@suse.de>
36076 PR middle-end/79399
36077 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
36078 type from int to size_t.
36079 * ira-costs.c (struct_costs_size): Change type from int to size_t.
36081 2017-02-07 Jakub Jelinek <jakub@redhat.com>
36083 PR rtl-optimization/79386
36084 * cprop.c (bypass_conditional_jumps): Initialize
36085 bypass_last_basic_block already before splitting bbs after
36086 unconditional traps...
36087 (bypass_conditional_jumps): ... rather than here.
36090 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
36091 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
36092 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
36093 fix -masm=intel patterns.
36095 2017-02-07 Richard Biener <rguenther@suse.de>
36097 PR tree-optimization/79256
36098 PR middle-end/79278
36099 * builtins.c (get_object_alignment_2): Use min_align_of_type
36100 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
36101 and ADJUST_FIELD_ALIGN.
36103 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
36105 * doc/tm.texi: Regenerate.
36106 * stor-layout.c (layout_decl): Adjust.
36107 (update_alignment_for_field): Likewise.
36108 (place_field): Likewise.
36109 (min_align_of_type): Likewise.
36110 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
36111 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
36112 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
36113 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
36114 * config/frv/frv.c (frv_adjust_field_align): Likewise.
36115 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
36116 * config/i386/i386.c (x86_field_alignment): Likewise.
36117 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
36118 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
36119 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
36120 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
36121 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
36122 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
36126 2017-01-30 Richard Biener <rguenther@suse.de>
36128 PR tree-optimization/79256
36129 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36130 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36133 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
36135 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
36136 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
36137 builtins to SImode and emit a zero-extend, if necessary.
36139 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36141 * docs/invoke.texi (RISC-V Options): Alphabetize.
36143 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36145 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
36148 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
36150 * config/riscv/riscv.c: New file.
36151 * common/config/riscv/riscv-common.c: Likewise.
36152 * config.gcc: Likewise.
36153 * config/riscv/constraints.md: Likewise.
36154 * config/riscv/elf.h: Likewise.
36155 * config/riscv/generic.md: Likewise.
36156 * config/riscv/linux.h: Likewise.
36157 * config/riscv/multilib-generator: Likewise.
36158 * config/riscv/peephole.md: Likewise.
36159 * config/riscv/pic.md: Likewise.
36160 * config/riscv/predicates.md: Likewise.
36161 * config/riscv/riscv-builtins.c: Likewise.
36162 * config/riscv/riscv-c.c: Likewise.
36163 * config/riscv/riscv-ftypes.def: Likewise.
36164 * config/riscv/riscv-modes.def: Likewise.
36165 * config/riscv/riscv-opts.h: Likewise.
36166 * config/riscv/riscv-protos.h: Likewise.
36167 * config/riscv/riscv.h: Likewise.
36168 * config/riscv/riscv.md: Likewise.
36169 * config/riscv/riscv.opt: Likewise.
36170 * config/riscv/sync.md: Likewise.
36171 * config/riscv/t-elf-multilib: Likewise.
36172 * config/riscv/t-linux: Likewise.
36173 * config/riscv/t-linux-multilib: Likewise.
36174 * config/riscv/t-riscv: Likewise.
36175 * configure.ac: Likewise.
36176 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
36177 Waterman as RISC-V maintainers.
36178 * doc/install.texi: Add RISC-V entries.
36179 * doc/invoke.texi: Add RISC-V options section.
36180 * doc/md.texi: Add RISC-V constraints section.
36181 * configure: Regenerated.
36183 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
36186 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
36187 false values to be constant vectors with all 0 or all 1 bits set.
36188 (vcondu<mode><mode>): Likewise.
36189 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
36191 (fpmask_comparison_operator): Update comment.
36192 (vecint_comparison_operator): New predicate.
36193 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
36194 vector conditionals when the true and false values are constant
36195 vectors with all 0 bits or all 1 bits set.
36197 2017-02-06 Martin Sebor <msebor@redhat.com>
36199 PR tree-optimization/79376
36200 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
36202 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
36204 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
36205 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
36206 to simplify split condition.
36208 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36210 * omp-expand.c (oxpand_omp_atomic_fetch_op,
36211 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
36214 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
36216 PR rtl-optimization/68664
36217 * target.def (can_speculate_insn): New hook.
36218 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
36219 * doc/tm.texi: Regenerate.
36220 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
36221 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
36222 (rs6000_sched_can_speculate_insn): New function.
36224 2017-02-06 Jakub Jelinek <jakub@redhat.com>
36226 PR tree-optimization/79284
36227 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
36228 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
36229 vectorizable_mask_load_store, vectorizable_operation,
36230 vect_is_simple_cond, get_same_sized_vectype): Use it instead
36231 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
36232 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
36233 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
36234 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
36235 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
36236 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
36237 is_gimple_assign (stmt). Replace another such test with
36238 is_gimple_assign (stmt).
36240 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36243 * config/avr/avr.c (rtl-iter.h): Include it.
36244 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
36245 (avr_legitimate_combined_insn): ...and implementation.
36247 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
36249 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
36250 * config/s390/s390.c (s390_const_operand_ok)
36251 (s390_canonicalize_comparison, s390_extract_part)
36252 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
36253 (s390_contiguous_bitmask_p, s390_rtx_costs)
36254 (legitimize_pic_address): Likewise.
36255 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
36256 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
36257 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
36258 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
36259 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
36261 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
36263 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
36264 REGNO($0) == REGNO($1).
36266 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36268 * config/s390/linux.h(SIZE_TYPE): Add comment.
36270 2017-02-06 Julian Brown <julian@codesourcery.com>
36271 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36272 Virendra Pathak <virendra.pathak@broadcom.com>
36274 * config/aarch64/aarch64-cores.def: Change the scheduler
36276 * config/aarch64/aarch64.md: Include thunderx2t99.md.
36277 * config/aarch64/thunderx2t99.md: New file.
36279 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36281 * doc/standards.texi (Go Language): Update link to language
36284 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36286 * tree-eh.c (lower_resx): Sanitize profile.
36287 (cleanup_empty_eh_move_lp): Likewise.
36289 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
36292 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
36294 * cfgloopmanip.h (loop_version): Update prototype.
36295 * modulo-sched.c (sms_schedule): Update call of loop_version.
36296 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
36297 * tree-parloops.c (gen_parallel_loop): Likewise.
36298 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
36299 * tree-ssa-loop-split.c (split_loop): Likewise.
36300 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
36301 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
36303 2017-02-05 Martin Liska <mliska@suse.cz>
36306 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
36308 (print_operand_address): Initialize a struct to zero.
36310 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36312 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
36313 garbage collector only in textual form.
36315 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
36317 * doc/extend.texi (x86 specific memory model extensions for
36318 transactional memory): Simplify a phrase.
36320 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
36323 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
36324 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
36325 (atomic_storedi_1): Likewise.
36327 2017-02-04 Jakub Jelinek <jakub@redhat.com>
36329 PR tree-optimization/79338
36330 * tree-parloops.c (gather_scalar_reductions): Don't call
36331 vect_analyze_loop_form for loop->inner before destroying loop's
36334 2017-02-03 Martin Sebor <msebor@redhat.com>
36336 PR tree-optimization/79327
36337 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
36338 when precision has resulted in leading zeros.
36339 (format_integer): Adjust the likely counter to assume an unknown
36340 argument that may be zero is non-zero.
36342 2017-02-03 Jason Merrill <jason@redhat.com>
36345 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
36346 avoid copying non-taken branch.
36348 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36350 PR tree-optimization/79340
36351 * tree-vect-loop.c (vectorizable_reduction): Release
36352 vec_defs elements after safe_splicing them into other vectors.
36355 PR tree-optimization/79327
36356 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
36357 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
36359 (format_integer): Use wide_int_to_tree instead of build_int_cst
36360 + to_?hwi. If argmin is NULL, just set argmin and argmax to
36361 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
36362 of shortest and longest sequence.
36364 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
36366 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
36367 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
36369 2017-02-03 Walter Lee <walt@tilera.com>
36372 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
36373 after initial stackframe link reg save.
36374 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
36376 2017-02-03 Jakub Jelinek <jakub@redhat.com>
36379 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
36380 wu for stxssp alternative.
36382 2017-02-03 Martin Sebor <msebor@redhat.com>
36384 PR tree-optimization/79352
36385 * gimple-fold.c (get_range_strlen): Add argument.
36386 (get_range_strlen): Change return type to bool.
36387 (get_maxval_strlen): Pass in a dummy argument.
36388 * gimple-fold.h (get_range_strlen): Change return type to bool.
36389 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
36390 * tree.h (array_at_struct_end_p): Add argument.
36391 * tree.c (array_at_struct_end_p): Handle it.
36393 2017-02-03 Martin Liska <mliska@suse.cz>
36396 * multiple_target.c (create_dispatcher_calls): Redirect edge
36397 from a caller of a dispatcher.
36398 (expand_target_clones): Make the clones local.
36399 (ipa_target_clone): Do both target clones and resolvers.
36400 (ipa_dispatcher_calls): Remove the pass.
36401 (pass_dispatcher_calls::gate): Likewise.
36402 (make_pass_dispatcher_calls): Likewise.
36403 * passes.def (pass_target_clone): Put as very first IPA early
36406 2017-02-03 Martin Liska <mliska@suse.cz>
36408 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
36409 in case of a function with ifunc attribute.
36411 2017-02-03 Martin Liska <mliska@suse.cz>
36413 * cgraph.c (cgraph_node::dump): Dump function version info.
36414 * symtab.c (symtab_node::dump_base): Add missing new line.
36416 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36418 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
36419 (ifcombine_ifandif): Use it.
36421 2017-02-03 Martin Liska <mliska@suse.cz>
36423 * doc/invoke.texi: Document default value for
36424 use-after-scope-direct-emission-threshold.
36426 2017-02-03 Martin Liska <mliska@suse.cz>
36428 PR tree-optimization/79339
36429 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
36430 (format_floating): Likewise.
36432 2017-02-03 Martin Liska <mliska@suse.cz>
36435 * ipa-prop.c (ipa_node_params_t::insert): Remove current
36437 (ipa_node_params_t::remove): Likewise.
36438 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
36439 initialization from removed ipa_node_params_t::insert.
36440 (ipa_node_params::~ipa_node_params): Move from removed
36441 ipa_node_params_t::release.
36442 * symbol-summary.h (symbol_summary::m_released): New member.
36443 Do not release a summary twice. Do not allow to call finalizer
36444 for types of a summary that live in GGC memory.
36446 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
36448 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
36451 2017-02-02 Martin Sebor <msebor@redhat.com>
36453 PR middle-end/79275
36454 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
36455 (format_string): Tighten up the range of output for non-constant
36456 strings and correct the expected range for wide non-constant strings.
36458 2017-02-02 Martin Sebor <msebor@redhat.com>
36460 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
36462 PR middle-end/32003
36463 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
36465 (-fdump-tree-@var): Add to index and document how to come up
36466 with pass-specific option and dump file names.
36467 (-fdump-passes): Clarify where to look for output.
36469 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
36471 PR middle-end/77445
36472 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
36473 statistics of the analyzed path; allow threading for speed when
36474 any of BBs along the path are optimized for speed.
36476 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
36478 PR middle-end/78468
36479 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
36480 settings of the virtual registers.
36483 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
36485 * explow.c (get_dynamic_stack_size): Take known alignment of stack
36486 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
36489 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36491 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
36492 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
36494 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36496 * config/s390/s390.md: Add missing comments with the expanded
36498 * config/s390/vector.md: Likewise.
36499 * config/s390/vx-builtins.md: Likewise.
36501 2017-02-02 Jakub Jelinek <jakub@redhat.com>
36504 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
36505 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
36506 conditions on a single line.
36508 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36510 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36511 __S390_VX__ to __VX__.
36513 2017-02-01 Andrew Pinski <apinski@cavium.com>
36515 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
36516 stmt_info to record_stmt_cost.
36517 (vect_get_known_peeling_cost): Pass stmt_info if known to
36519 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
36520 cpu_vector_cost field into
36521 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
36522 field into vec_int_stmt_cost and vec_fp_stmt_cost.
36523 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
36524 splitting of scalar_stmt_cost and vec_stmt_cost.
36525 (thunderx_vector_cost): Likewise.
36526 (cortexa57_vector_cost): LIkewise.
36527 (exynosm1_vector_cost): Likewise.
36528 (xgene1_vector_cost): Likewise.
36529 (thunderx2t99_vector_cost): Improve after the splitting of the two
36531 (aarch64_builtin_vectorization_cost): Update for the splitting of
36532 scalar_stmt_cost and vec_stmt_cost.
36534 2017-02-01 Torvald Riegel <triegel@redhat.com>
36535 Richard Henderson <rth@redhat.com>
36537 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
36538 conditional on existance of a fast atomic load.
36539 * optabs-query.c (can_atomic_load_p): New function.
36540 * optabs-query.h (can_atomic_load_p): Declare it.
36541 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
36542 no fast atomic load is available for the particular size of access.
36543 (expand_atomic_compare_and_swap): Likewise.
36544 (expand_atomic_load): Likewise.
36545 (expand_atomic_store): Likewise.
36546 (expand_atomic_fetch_op): Likewise.
36547 * testsuite/lib/target-supports.exp
36548 (check_effective_target_sync_int_128): Remove x86 because it provides
36549 no fast atomic load.
36550 (check_effective_target_sync_int_128_runtime): Likewise.
36552 2017-02-01 Richard Biener <rguenther@suse.de>
36554 * graphite.c: Include tree-vectorizer.h for find_loop_location.
36555 (graphite_transform_loops): Provide opt-info for optimized nests.
36556 * tree-parloop.c (parallelize_loops): Provide opt-info for
36557 parallelized loops.
36559 2017-02-01 Richard Biener <rguenther@suse.de>
36561 PR middle-end/79315
36562 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
36563 was not set before.
36565 2017-02-01 Richard Biener <rguenther@suse.de>
36567 PR tree-optimization/71824
36568 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
36569 Verify the loops are valid in the merged SESE region.
36570 (scop_detection::can_represent_loop_1): Check analyzing the
36571 evolution of the number of iterations in the region succeeds.
36573 2017-01-31 Ian Lance Taylor <iant@golang.org>
36575 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
36576 REG_ARGS_SIZE note to 32-bit push insns and call insn.
36578 2017-01-31 David Malcolm <dmalcolm@redhat.com>
36580 PR preprocessor/79210
36581 * input.c (get_substring_ranges_for_loc): Replace line_width
36582 assertion with error-handling.
36584 2017-01-31 Richard Biener <rguenther@suse.de>
36586 PR tree-optimization/77318
36587 * graphite-sese-to-poly.c (extract_affine): Fix assert.
36588 (create_pw_aff_from_tree): Take loop parameter.
36589 (add_condition_to_pbb): Pass loop of the condition to
36590 create_pw_aff_from_tree.
36592 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36594 * config/s390/s390.c (s390_asan_shadow_offset): New function.
36595 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
36597 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
36601 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
36603 (convert_int_to_float128): Likewise.
36604 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
36605 (convert_int_to_float128): Likewise.
36606 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
36607 (UNSPEC_IEEE128_CONVERT): Likewise.
36608 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
36609 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
36610 Use local variables for IBM extended format.
36611 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
36612 (fix_trunc<mode>si2_fprs): Likewise.
36613 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
36614 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
36615 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
36616 to know that we can now have integers of all sizes in vector
36618 (fix<uns>_<mode>di2_hw): Likewise.
36619 (float<uns>_<mode>si2_hw): Likewise.
36620 (fix_<mode>si2_hw): Likewise.
36621 (fixuns_<mode>si2_hw): Likewise.
36622 (float<uns>_<mode>di2_hw): Likewise.
36623 (float_<mode>di2_hw): Likewise.
36624 (float_<mode>si2_hw): Likewise.
36625 (floatuns_<mode>di2_hw): Likewise.
36626 (floatuns_<mode>si2_hw): Likewise.
36627 (xscvqp<su>wz_<mode>): Delete, no longer used.
36628 (xscvqp<su>dz_<mode>): Likewise.
36629 (xscv<su>dqp_<mode>): Likewise.
36630 (ieee128_mfvsrd_64bit): Likewise.
36631 (ieee128_mfvsrd_32bit): Likewise.
36632 (ieee128_mfvsrwz): Likewise.
36633 (ieee128_mtvsrw): Likewise.
36634 (ieee128_mtvsrd_64bit): Likewise.
36635 (ieee128_mtvsrd_32bit): Likewise.
36637 2017-01-31 Martin Liska <mliska@suse.cz>
36640 * ipa-prop.c (ipa_free_all_node_params): Call release method
36641 instead of ~sumbol_summary to not to trigger double times
36644 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
36646 PR tree-optimization/71691
36647 * bitmap.h (class auto_bitmap): New.
36648 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
36649 is_maybe_undefined instead of ssa_undefined_value_p.
36651 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36653 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
36654 __S390_ARCH_LEVEL__ to __ARCH__.
36656 2017-01-31 Jakub Jelinek <jakub@redhat.com>
36658 PR tree-optimization/79267
36659 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
36660 if should_remove_lhs_p is true.
36662 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
36665 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
36666 (add_alignment_attribute): New.
36667 (base_type_die): Add alignment attribute.
36668 (subrange_type_die): Likewise.
36669 (modified_type_die): Likewise.
36670 (gen_array_type_die): Likewise.
36671 (gen_descr_array_type_die: Likewise.
36672 (gen_enumeration_type_die): Likewise.
36673 (gen_subprogram_die): Likewise.
36674 (gen_variable_die): Likewise.
36675 (gen_field_die): Likewise.
36676 (gen_ptr_to_mbr_type_die): Likewise.
36677 (gen_struct_or_union_type_die): Likewise.
36678 (gen_subroutine_type_die): Likewise.
36679 (gen_typedef_die): Likewise.
36680 (base_type_cmp): Compare alignment attribute.
36682 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36685 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
36686 (setb_unsigned) New pattern for setb with CCUNS.
36687 * config/rs6000/rs6000.c (expand_block_compare): Use a different
36688 subfc./subfe sequence to avoid overflow problems. Generate a
36689 shorter sequence with cmpld/setb for power9.
36690 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
36691 for generating subfc. instruction.
36692 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
36693 now uses this instruction.
36695 2017-01-30 Ian Lance Taylor <iant@google.com>
36698 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
36699 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
36701 2017-01-30 Martin Sebor <msebor@redhat.com>
36703 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
36704 Move constant to the right of a relational operator.
36705 (get_mpfr_format_length, format_character, format_string): Ditto.
36706 (should_warn_p, maybe_warn): Same.
36708 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
36710 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
36713 * asan.c (get_translation_unit_decl): Remove function.
36714 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
36716 2017-01-30 Martin Liska <mliska@suse.cz>
36718 PR gcov-profile/79259
36719 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
36720 -fprofile-generate.
36722 2017-01-30 Martin Liska <mliska@suse.cz>
36725 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
36726 Initialize variables with NULL value.
36728 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
36731 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
36733 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
36735 2017-01-30 Richard Biener <rguenther@suse.de>
36737 PR tree-optimization/79276
36738 * tree-vrp.c (process_assert_insertions): Properly adjust common
36739 when removing a duplicate.
36741 2017-01-30 Richard Biener <rguenther@suse.de>
36743 PR tree-optimization/79256
36744 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
36745 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
36747 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
36749 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
36752 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
36753 ("*r<noxa>sbg_<mode>_sll_bitmask")
36754 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
36755 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
36756 Use contiguous_bitmask_nowrap_operand.
36758 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36761 * config/rs6000/altivec.h (vec_xl): Revise #define.
36762 (vec_xst): Likewise.
36764 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
36766 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
36768 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
36770 PR rtl-optimization/79194
36771 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
36772 traps before call to bypass_conditional_jumps.
36774 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36776 PR tree-optimization/71374
36777 * lra-constraints.c (check_conflict_input_operands): New.
36778 (match_reload): Use it.
36780 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
36783 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
36784 account to calculate conflict_set.
36786 2017-01-27 Bin Cheng <bin.cheng@arm.com>
36788 PR rtl-optimization/78559
36789 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
36790 other_insn in combine.
36792 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
36794 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
36795 uint16_type_node for BT_UINT16.
36797 2017-01-27 David Malcolm <dmalcolm@redhat.com>
36799 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
36800 "RTL Tests" to menu.
36801 (GIMPLE Tests): New node.
36802 (RTL Tests): New node.
36804 2017-01-27 Richard Biener <rguenther@suse.de>
36806 PR tree-optimization/79245
36807 * tree-loop-distribution.c (distribute_loop): Apply cost
36808 modeling also to detected patterns.
36810 2017-01-27 Richard Biener <rguenther@suse.de>
36812 PR tree-optimization/71433
36813 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
36814 (compare_assert_loc): New function.
36815 (process_assert_insertions): Sort and optimize assert locations
36816 to remove duplicates and push down identical assertions on
36817 edges to their destination block.
36819 2017-01-27 Richard Biener <rguenther@suse.de>
36821 PR tree-optimization/79244
36822 * tree-vrp.c (remove_range_assertions): Forcefully propagate
36823 out SSA names even if abnormal.
36825 2017-01-27 Jakub Jelinek <jakub@redhat.com>
36827 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
36828 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
36829 instead of MPFR_RNDN.
36831 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
36834 * arm.c (arm_option_override): Don't call build_target_option_node
36835 until after doing all option overrides.
36836 (arm_valid_target_attribute_tree): Likewise.
36838 2017-01-27 Martin Liska <mliska@suse.cz>
36840 * doc/invoke.texi (-fprofile-arcs): Document profiling support
36841 for {cd}tors and C++ {cd}tors.
36843 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
36845 * config/s390/s390.md ("*setmem_long_and")
36846 ("*setmem_long_and_31z"): Use zero_extend instead of and.
36848 2017-01-26 Martin Sebor <msebor@redhat.com>
36850 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
36853 2017-01-26 Martin Sebor <msebor@redhat.com>
36855 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
36856 HAVE_DFmode before using XFmode or DFmode.
36857 (parse_directive): Avoid using the z length modifier to avoid
36858 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
36860 PR middle-end/78703
36861 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
36862 to accept adjustment as an array.
36863 (get_int_range): New function.
36864 (struct directive): Make width and prec arrays.
36865 (directive::set_width, directive::set_precision): Call get_int_range.
36866 (format_integer, format_floating): Handle width and precision ranges.
36867 (format_string, parse_directive): Same.
36869 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36872 * dwarf2out.c (generate_skeleton_bottom_up): For children with
36873 comdat_type_p set, just clone them, but keep the children in the
36877 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
36878 which have direct callers with -fvar-tracking-assignments enabled
36880 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
36881 inside of type units.
36883 2017-01-26 Martin Sebor <msebor@redhat.com>
36885 PR middle-end/78703
36886 * gimple-ssa-sprintf.c (struct result_range): Add likely and
36888 (struct format_result): Replace number_chars, number_chars_min,
36889 and number_chars_max with a single member of struct result_range.
36891 (format_result::operator+=): Adjust.
36892 (struct fmtresult): Remove bounded. Handle likely and unlikely
36894 (fmtresult::adjust_for_width_or_precision): New function.
36895 (fmtresult:type_max_digits): New function.
36896 (bytes_remaining): Handle likely and unlikely counters.
36897 (min_bytes_remaining): Remove.
36898 (format_percent): Simplify.
36899 (format_integer, format_floating): Set likely and unlikely counters.
36900 (get_string_length, format_character, format_string): Same.
36901 (format_plain, should_warn_p): New function.
36902 (maybe_warn): Call should_warn_p. Update diagnostic messages
36903 and handle those for all directives, including plain strings.
36904 (format_directive): Handle likely and unlikely counters.
36905 Remove unnecessary quoting from diagnostics. Add an informational
36907 (add_bytes): Remove.
36908 (pass_sprintf_length::compute_format_length): Simplify.
36909 (try_substitute_return_value): Handle likely and unlikely counters.
36911 2017-01-26 Carl Love <cel@us.ibm.com>
36913 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
36914 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
36916 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
36919 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
36920 endianess for subregs into account.
36921 * lra-constraints.c (lra_constraints): Do risky transformations
36922 always on the first iteration.
36923 * lra-lives.c (check_pseudos_live_through_calls): Add arg
36924 last_call_used_reg_set.
36925 (process_bb_lives): Define and use last_call_used_reg_set.
36926 * lra.c (lra): Always continue after lra_constraints on the first
36929 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
36931 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
36933 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
36935 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36937 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
36938 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
36939 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
36940 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
36941 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
36942 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
36943 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
36944 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
36945 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
36947 2017-01-26 Marek Polacek <polacek@redhat.com>
36950 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
36951 for the third operand.
36953 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36955 PR middle-end/79236
36956 * omp-low.c (struct omp_context): Add simt_stmt field.
36957 (scan_omp_for): Return omp_context *.
36958 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
36959 context to the _simt_ SIMD stmt.
36960 (lower_omp_for): For combined SIMD with sibling _simt_
36961 SIMD, make sure to use the same decls in _looptemp_
36962 clauses as in the sibling.
36964 2017-01-26 David Sherwood <david.sherwood@arm.com>
36966 PR middle-end/79212
36967 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
36970 2017-01-26 Jakub Jelinek <jakub@redhat.com>
36973 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
36974 emit fld b; fld a; if possible.
36976 * brig-builtins.def: Update copyright years.
36977 * config/arm/arm_acle_builtins.def: Update copyright years.
36979 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
36982 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
36983 constraint instead of o for the stxsd instruction.
36985 2017-01-25 Carl Love <cel@us.ibm.com>
36987 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
36988 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
36990 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
36992 * doc/invoke.texi (C++ Dialect Options): Fix typo.
36994 2017-01-25 Richard Biener <rguenther@suse.de>
36996 PR tree-optimization/69264
36997 * target.def (vector_alignment_reachable): Improve documentation.
36998 * doc/tm.texi: Regenerate.
36999 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
37001 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
37002 earlier changes with respect to TYPE_USER_ALIGN.
37003 (vector_alignment_reachable_p): Likewise. Improve dumping.
37005 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37008 * config/arm/arm.md (xordi3): Force constant operand into a register
37011 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37013 * doc/invoke.texi (-fstore-merging): Correct default optimization
37014 levels at which it is enabled.
37015 (-O): Move -fstore-merging from list to...
37018 2017-01-25 Richard Biener <rguenther@suse.de>
37021 * omp-expand.c: Include debug.h.
37022 (expand_omp_taskreg): Make sure to generate early debug before
37023 outlining anything from a function.
37024 (expand_omp_target): Likewise.
37025 (grid_expand_target_grid_body): Likewise.
37027 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
37030 * asan.c (get_translation_unit_decl): New function.
37031 (asan_add_global): Extract modules file name from globals
37032 TRANSLATION_UNIT_DECL name.
37034 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
37037 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
37038 for long calls with APCS frame and VFP.
37040 2017-01-24 David Malcolm <dmalcolm@redhat.com>
37042 * cfg.c (original_copy_tables_initialized_p): New function.
37043 * cfg.h (original_copy_tables_initialized_p): New decl.
37044 * cfgrtl.c (relink_block_chain): Guard the call to
37045 free_original_copy_tables with a call to
37046 original_copy_tables_initialized_p.
37047 * cgraph.h (symtab_node::native_rtl_p): New decl.
37048 * cgraphunit.c (symtab_node::native_rtl_p): New function.
37049 (symtab_node::needed_p): Don't assert for early assembly output
37050 for __RTL functions.
37051 (cgraph_node::finalize_function): Set "force_output" for __RTL
37053 (cgraph_node::analyze): Bail out early for __RTL functions.
37054 (analyze_functions): Update assertion to support __RTL functions.
37055 (cgraph_node::expand): Bail out early for __RTL functions.
37056 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
37058 * function.h (struct function): Update comment for field
37060 * gimple-expr.c: Include "tree-pass.h".
37061 (gimple_has_body_p): Return false for __RTL functions.
37062 * Makefile.in (OBJS): Add run-rtl-passes.o.
37063 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
37065 (gcc::pass_manager::get_clean_slate): New accessor.
37066 * passes.c: Include "insn-addr.h".
37067 (should_skip_pass_p): Add logging. Update logic for running
37068 "expand" to be compatible with both __GIMPLE and __RTL. Guard
37069 property-provider override so it is only done for gimple passes.
37071 (skip_pass): New function.
37072 (execute_one_pass): Call skip_pass when skipping passes.
37073 * read-md.c (md_reader::read_char): Support filtering
37074 the input to a subset of line numbers.
37075 (md_reader::md_reader): Initialize fields
37076 m_first_line and m_last_line.
37077 (md_reader::read_file_fragment): New function.
37078 * read-md.h (md_reader::read_file_fragment): New decl.
37079 (md_reader::m_first_line): New field.
37080 (md_reader::m_last_line): New field.
37081 * read-rtl-function.c (function_reader::create_function): Only
37082 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
37083 curr_properties. Set DECL_INITIAL to a dummy block.
37084 (read_rtl_function_body_from_file_range): New function.
37085 * read-rtl-function.h (read_rtl_function_body_from_file_range):
37087 * run-rtl-passes.c: New file.
37088 * run-rtl-passes.h: New file.
37090 2017-01-24 Jeff Law <law@redhat.com>
37092 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
37095 2017-01-24 Bin Cheng <bin.cheng@arm.com>
37097 PR tree-optimization/79159
37098 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
37099 (record_nonwrapping_iv): Improve boundary using above function if no
37100 value range information.
37102 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
37103 Martin Jambor <mjambor@suse.cz>
37105 * brig-builtins.def: New file.
37106 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
37107 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
37108 (DEF_HSAIL_SAT_BUILTIN): Likewise.
37109 (DEF_HSAIL_INTR_BUILTIN): Likewise.
37110 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
37111 * builtin-types.def (BT_INT8): New.
37112 (BT_INT16): Likewise.
37113 (BT_UINT8): Likewise.
37114 (BT_UINT16): Likewise.
37115 (BT_FN_ULONG): Likewise.
37116 (BT_FN_UINT_INT): Likewise.
37117 (BT_FN_UINT_ULONG): Likewise.
37118 (BT_FN_UINT_LONG): Likewise.
37119 (BT_FN_UINT_PTR): Likewise.
37120 (BT_FN_ULONG_PTR): Likewise.
37121 (BT_FN_INT8_FLOAT): Likewise.
37122 (BT_FN_INT16_FLOAT): Likewise.
37123 (BT_FN_UINT32_FLOAT): Likewise.
37124 (BT_FN_UINT16_FLOAT): Likewise.
37125 (BT_FN_UINT8_FLOAT): Likewise.
37126 (BT_FN_UINT64_FLOAT): Likewise.
37127 (BT_FN_UINT16_UINT32): Likewise.
37128 (BT_FN_UINT32_UINT16): Likewise.
37129 (BT_FN_UINT16_UINT16_UINT16): Likewise.
37130 (BT_FN_INT_PTR_INT): Likewise.
37131 (BT_FN_UINT_PTR_UINT): Likewise.
37132 (BT_FN_LONG_PTR_LONG): Likewise.
37133 (BT_FN_ULONG_PTR_ULONG): Likewise.
37134 (BT_FN_VOID_UINT64_UINT64): Likewise.
37135 (BT_FN_UINT8_UINT8_UINT8): Likewise.
37136 (BT_FN_INT8_INT8_INT8): Likewise.
37137 (BT_FN_INT16_INT16_INT16): Likewise.
37138 (BT_FN_INT_INT_INT): Likewise.
37139 (BT_FN_UINT_FLOAT_UINT): Likewise.
37140 (BT_FN_FLOAT_UINT_UINT): Likewise.
37141 (BT_FN_ULONG_UINT_UINT): Likewise.
37142 (BT_FN_ULONG_UINT_PTR): Likewise.
37143 (BT_FN_ULONG_ULONG_ULONG): Likewise.
37144 (BT_FN_UINT_UINT_UINT): Likewise.
37145 (BT_FN_VOID_UINT_PTR): Likewise.
37146 (BT_FN_UINT_UINT_PTR: Likewise.
37147 (BT_FN_UINT32_UINT64_PTR): Likewise.
37148 (BT_FN_INT_INT_UINT_UINT): Likewise.
37149 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
37150 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
37151 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
37152 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
37153 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
37154 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
37155 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
37156 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
37157 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
37158 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
37159 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
37160 * doc/frontends.texi: List BRIG FE.
37161 * doc/install.texi (Testing): Add BRIG tesring requirements.
37162 * doc/invoke.texi (Overall Options): Mention BRIG.
37163 * doc/standards.texi (Standards): Doucment BRIG HSA version.
37165 2017-01-24 Richard Biener <rguenther@suse.de>
37167 PR translation/79208
37168 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
37170 2017-01-24 Martin Jambor <mjambor@suse.cz>
37173 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
37174 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
37175 and known_contexts.
37177 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
37179 PR middle-end/79123
37180 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
37181 casts from signed to unsigned really don't have a range.
37183 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
37185 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
37186 GMP_RNDx for compatiblity.
37188 2017-01-24 Martin Liska <mliska@suse.cz>
37191 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
37192 that would prevent us to call alloca with -1 as argument.
37194 2017-01-24 Jakub Jelinek <jakub@redhat.com>
37196 * dwarf2out.c (output_compilation_unit_header, output_file_names):
37197 Avoid -Wformat-security warning.
37199 2017-01-23 Andrew Pinski <apinski@cavium.com>
37201 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
37204 2017-01-23 Martin Sebor <msebor@redhat.com>
37206 PR middle-end/78703
37207 * gimple-ssa-sprintf.c (warn_level): New global.
37208 (format_integer): Use it here and throughout the rest of the file.
37209 Use the same switch to compute sign as base.
37210 (maybe_warn): New function.
37211 (format_directive): Factor out warnings into maybe_warn.
37212 Add debugging output. Use warn_level.
37213 (add_bytes): Use warn_level.
37214 (pass_sprintf_length::compute_format_length): Add debugging output.
37215 (try_substitute_return_value): Same.
37216 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
37218 PR middle-end/78703
37219 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
37220 (struct fmtresult, format_integer, format_floating): Adjust.
37221 (fmtresult::fmtresult): Set max correctly in two argument ctor.
37222 (get_string_length, format_string,format_directive): Same.
37223 (pass_sprintf_length::compute_format_length): Same.
37224 (try_substitute_return_value): Simplify slightly.
37226 PR middle-end/78703
37227 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
37228 (fmtresult::operator+=): Outlined.
37229 (struct fmtresult): Add ctors.
37230 (struct conversion_spec): Rename...
37231 (struct directive): ...to this. Add and remove data members.
37232 (directive::set_width, directive::set_precision): New functions.
37233 (format_percent): Use fmtresult ctor.
37234 (get_width_and_precision): Remove.
37235 (format_integer): Make naming changes. Avoid computing width and
37237 (format_floating): Same. Adjust indentation.
37238 (format_character, format_none): New functions.
37239 (format_string): Moved character handling to format_character.
37240 (format_directive): Remove arguments, change return type.
37241 (parse_directive): New function.
37242 (pass_sprintf_length::compute_format_length): Move directive
37243 parsing to parse_directive.
37245 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37247 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
37248 (assign_assembler_name_if_needed): ... this.
37249 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
37250 (assign_assembler_name_if_needed): ... this.
37251 (free_lang_data_in_cgraph): Adjust callers.
37252 * cgraphunit.c (cgraph_node::analyze): Likewise.
37253 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
37256 2017-01-23 Richard Biener <rguenther@suse.de>
37258 PR tree-optimization/79088
37259 PR tree-optimization/79188
37260 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
37261 resetting loop bounds after last path deletion. Reset loop
37262 bounds of the target loop, make code match the comments.
37263 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
37264 Make sure loops need no fixups.
37266 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
37268 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
37269 exponent support with double type for first argument.
37270 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
37271 type returned by __builtin_vec_extract_sig,
37272 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
37273 functions from "vector int" to "vector unsigned int" or from
37274 "vector long long int" to "vector unsigned long long int".
37275 Changed type returned by __builtin_vec_extract_exp,
37276 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
37277 functions from "vector int" to "vector unsigned int" or from
37278 "vector long long int" to "vector unsigned long long int".
37279 Changed return type of __builtin_vec_test_data_class,
37280 __builtin_vec_test_data_class_sp, and
37281 __builtin_vec_test_data_class_dp from "vector int" to
37282 "vector bool int" or from "vector long long int" to "vector bool
37283 long long int" and changed second argument type from "unsigned
37284 int" to "int". Added new overloaded function forms "vector float
37285 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
37286 "vector float __builtin_vec_insert_exp_sp (vector float, vector
37287 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
37288 double, vector unsigned long long int)" and "vector double
37289 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
37290 long int)". Changed return type of
37291 __builtin_scalar_test_data_class and
37292 __builtin_scalar_test_data_class_sp and
37293 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
37294 int" and changed second argument from "unsigned int" to "int".
37295 Changed type returned by __builtin_scalar_test_neg,
37296 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
37297 from "int" to "bool int". Added new overloaded function form
37298 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
37299 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
37300 exponent double-precision with floating point first argument.
37301 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
37302 documentation of scalar_test_data_class, scalar_test_neg,
37303 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
37304 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
37305 vec_test_data_class built-in functions to reflect refinements in
37306 their type signatures.
37308 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
37310 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
37312 (aarch64_elf_asm_destructor): Likewise.
37314 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
37316 PR rtl-optimization/78634
37317 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
37318 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
37319 * ifcvt.c (noce_try_cmove): Add missing cost check.
37321 PR rtl-optimization/71724
37322 * combine.c (if_then_else_cond): Look for situations where it is
37323 beneficial to undo the work of one of the recursive calls.
37325 2017-01-23 Bin Cheng <bin.cheng@arm.com>
37327 PR tree-optimization/70754
37328 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
37329 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
37330 combined stmt before it if not NULL.
37331 (combine_chains): Process refs reversely and compute dominance point
37334 2017-01-23 Martin Liska <mliska@suse.cz>
37336 PR tree-optimization/79196
37337 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
37338 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
37340 (strlen_optimize_stmt): Call the renamed function.
37342 2017-01-23 Michael Matz <matz@suse.de>
37344 PR tree-optimization/78384
37345 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
37347 2017-01-23 Richard Biener <rguenther@suse.de>
37349 PR tree-optimization/79186
37350 * tree-vrp.c (register_new_assert_for): Make sure we've seen
37351 both incoming edges before moving an assert.
37353 2017-01-23 Martin Jambor <mjambor@suse.cz>
37355 * ipa-prop.c (load_from_param_1): Removed.
37356 (load_from_unmodified_param): Bits from load_from_param_1 put back
37358 (load_from_param): Removed.
37359 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
37360 with stmt. Reverted back to use of load_from_unmodified_param.
37362 2017-01-23 Martin Jambor <mjambor@suse.cz>
37365 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
37366 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
37367 field a pointer to garbage collected vector, mark lattices and
37368 ipcp_orig_node with GTY((skip)).
37369 (ipa_get_param_count): Adjust to descriptors being a pointer.
37370 (ipa_get_param): Likewise.
37371 (ipa_get_type): Likewise.
37372 (ipa_get_param_move_cost): Likewise.
37373 (ipa_set_param_used): Likewise.
37374 (ipa_get_controlled_uses): Likewise.
37375 (ipa_set_controlled_uses): Likewise.
37376 (ipa_is_param_used): Likewise.
37377 (ipa_node_params_t): Move into garbage collector. New methods insert
37379 (ipa_node_params_sum): Annotate wth GTY(()).
37380 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
37382 (ipa_load_from_parm_agg): Adjust declaration.
37383 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
37384 * ipa-profile.c (ipa_profile): Likewise.
37385 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
37386 (ipa_populate_param_decls): Make descriptors parameter garbage
37388 (ipa_dump_param): Adjust to descriptors being a pointer.
37389 (ipa_alloc_node_params): Likewise.
37390 (ipa_initialize_node_params): Likewise.
37391 (load_from_param_1): Make descriptors parameter garbage collected.
37392 (load_from_unmodified_param): Likewise.
37393 (load_from_param): Likewise.
37394 (ipa_load_from_parm_agg): Likewise.
37395 (ipa_node_params::~ipa_node_params): Removed.
37396 (ipa_free_all_node_params): Remove call to delete operator.
37397 (ipa_node_params_t::insert): New.
37398 (ipa_node_params_t::remove): Likewise.
37399 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
37400 copy known_csts and known_contexts vectors.
37401 (ipa_read_node_info): Adjust to descriptors being a pointer.
37402 (ipcp_modif_dom_walker): Make m_descriptors field garbage
37404 (ipcp_transform_function): Make descriptors variable garbage
37407 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
37409 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
37410 * config/i386/avx512dqintrin.h: Ditto.
37411 * config/i386/avx512fintrin.h: Ditto.
37412 * config/i386/i386.c: Handle new builtins.
37413 * config/i386/i386-builtin.def: Add new builtins.
37414 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
37415 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
37417 2017-01-23 Jakub Jelinek <jakub@redhat.com>
37418 Martin Liska <mliska@suse.cz>
37420 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
37421 * asan.c (asan_expand_poison_ifn): Support stores and use
37422 appropriate ASAN report function.
37423 * internal-fn.c (expand_ASAN_POISON_USE): New function.
37424 * internal-fn.def (ASAN_POISON_USE): Declare.
37425 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
37426 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
37427 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
37428 ASAN_POISON calls w/o LHS.
37429 * tree-ssa.c (execute_update_addresses_taken): Create clobber
37430 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
37431 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
37432 * gimplify.c (asan_poison_variables): Add attribute
37433 use_after_scope_memory to variables that really needs to live
37435 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
37436 having the attribute.
37438 2017-01-23 Martin Liska <mliska@suse.cz>
37440 * asan.c (create_asan_shadow_var): New function.
37441 (asan_expand_poison_ifn): Likewise.
37442 * asan.h (asan_expand_poison_ifn): New declaration.
37443 * internal-fn.c (expand_ASAN_POISON): Likewise.
37444 * internal-fn.def (ASAN_POISON): New builtin.
37445 * sanopt.c (pass_sanopt::execute): Expand
37446 asan_expand_poison_ifn.
37447 * tree-inline.c (copy_decl_for_dup_finish): Make function
37449 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
37450 * tree-ssa.c (is_asan_mark_p): New function.
37451 (execute_update_addresses_taken): Rewrite local variables
37452 (identified just by use-after-scope as addressable) into SSA.
37454 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37456 * doc/install.texi (Specific): opensource.apple.com uses https
37457 now. Remove trailing slash.
37459 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
37461 * README.Portability: Remove note on an Irix compatibility issue.
37463 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
37465 * gcov.c (INCLUDE_ALGORITHM): Define.
37466 (INCLUDE_VECTOR): Define.
37467 No longer include <vector> and <algorithm> directly.
37469 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
37471 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
37473 * doc/invoke.texi (Code Gen Options): Ditto.
37475 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
37478 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
37480 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
37482 rtl-optimization/79125
37483 * cprop.c (local_cprop_pass): Handle cases where we make an
37484 unconditional trap.
37486 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
37490 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
37491 read from, for big endian.
37493 2017-01-20 Jiong Wang <jiong.wang@arm.com>
37495 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
37496 register pauth builtins for LP64 only.
37498 2017-01-20 Marek Polacek <polacek@redhat.com>
37501 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
37504 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37506 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
37508 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
37509 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
37510 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
37512 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37515 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
37516 in big-endian mode when they are not a single duplicated value.
37518 2017-01-20 Richard Biener <rguenther@suse.de>
37520 * BASE-VER: Bump to 7.0.1.
37522 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
37524 * omp-low.c (omplow_simd_context): New struct. Use it...
37525 (lower_rec_simd_input_clauses): ...here and...
37526 (lower_rec_input_clauses): ...here to hold common data. Adjust all
37527 references to idx, lane, max_vf, is_simt.
37529 2017-01-20 Graham Markall <graham.markall@embecosm.com>
37531 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
37534 2017-01-20 Martin Jambor <mjambor@suse.cz>
37536 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
37537 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
37539 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
37540 (GTFILES): Rename hsa.c to hsa-common.c.
37541 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
37542 * hsa-dump.c: Likewise.
37543 * hsa-gen.c: Likewise.
37544 * hsa-regalloc.c: Likewise.
37545 * ipa-hsa.c: Likewise.
37546 * omp-expand.c: Likewise.
37547 * omp-low.c: Likewise.
37548 * toplev.c: Likewise.
37550 2017-01-20 Marek Polacek <polacek@redhat.com>
37553 * doc/invoke.texi: Document -Wduplicated-branches.
37554 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
37555 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
37556 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
37557 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
37558 return 0 only when not OEP_LEXICOGRAPHIC.
37559 (fold_build_cleanup_point_expr): Use the expression
37560 location when building CLEANUP_POINT_EXPR.
37561 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
37562 * tree.c (add_expr): Handle error_mark_node.
37564 2017-01-20 Martin Liska <mliska@suse.cz>
37567 * tree-profile.c (init_ic_make_global_vars): Do not call
37569 (gimple_init_gcov_profiler): Likewise.
37571 2017-01-20 Martin Liska <mliska@suse.cz>
37574 * cgraph.h (maybe_create_reference): Remove argument and
37576 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
37578 * ipa-cp.c (create_specialized_node): Likewise.
37579 * symtab.c (symtab_node::maybe_create_reference): Handle
37580 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
37582 2017-01-20 Martin Liska <mliska@suse.cz>
37584 * read-rtl-function.c (function_reader::create_function): Use
37585 build_decl instread of build_decl_stat.
37587 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
37589 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
37590 * config/i386/avx512dqintrin.h: Ditto.
37591 * config/i386/avx512fintrin.h: Ditto.
37592 * config/i386/i386-builtin-types.def: Add new types.
37593 * config/i386/i386.c: Handle new types.
37594 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
37595 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
37596 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
37597 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
37598 (__builtin_ia32_kshiftridi): New.
37599 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
37601 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
37605 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
37606 define to rs6000_init_stack_protect_guard.
37607 (rs6000_init_stack_protect_guard): New function.
37609 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37610 Yunqiang Su <yunqiang.su@imgtec.com>
37612 * config.gcc (supported_defaults): Add madd4.
37613 (with_madd4): Add validation.
37614 (all_defaults): Add madd4.
37615 * config/mips/mips.opt (mmadd4): New option.
37616 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37618 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37620 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
37621 (ISA_HAS_FUSED_MADD4): Likewise.
37622 * doc/invoke.texi (-mmadd4): Document the new option.
37623 * doc/install.texi (--with-madd4): Document the new option.
37625 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37627 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
37628 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
37629 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
37630 (aarch64_init_pauth_hint_builtins): New.
37631 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
37632 (aarch64_expand_builtin): Expand new builtins.
37634 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37636 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
37637 * combine-stack-adj.c (no_unhandled_cfa): Handle
37638 REG_CFA_TOGGLE_RA_MANGLE.
37639 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
37640 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
37641 info for return address signing.
37642 (aarch64_expand_epilogue): Likewise.
37644 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37646 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
37647 * config/aarch64/aarch64-protos.h
37648 (aarch64_return_address_signing_enabled): New declaration.
37649 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
37651 (aarch64_expand_prologue): Sign return address before it's pushed onto
37653 (aarch64_expand_epilogue): Authenticate return address fetched from
37655 (aarch64_override_options): Sanity check for ILP32 and ISA level.
37656 (aarch64_attributes): New function attributes for "sign-return-address".
37657 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
37658 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
37659 ("*do_return"): Generate combined instructions according to key index.
37660 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
37661 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
37663 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
37664 * config/aarch64/aarch64.opt (msign-return-address=): New.
37665 * doc/extend.texi (AArch64 Function Attributes): Documents
37666 "sign-return-address=".
37667 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
37669 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
37671 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
37672 overall option summary.
37674 2017-01-19 Jiong Wang <jiong.wang@arm.com>
37676 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
37677 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
37678 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
37679 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
37681 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
37683 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
37684 -mpower9-minmax by default for -mcpu=power9.
37685 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
37686 128-bit floating point.
37688 2017-01-20 Alan Modra <amodra@gmail.com>
37690 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
37691 optimizing for size.
37693 2017-01-20 Alan Modra <amodra@gmail.com>
37696 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
37697 for strcmp and strncmp from corresponding builtin decl.
37699 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37701 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
37702 instead of i386/rtems-64.h.
37703 * config/i386/rtems-64.h: Remove.
37705 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
37709 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
37711 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
37713 2017-01-19 Tamar Christina <tamar.christina@arm.com>
37715 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
37716 Change int to HOST_WIDE_INT.
37717 * config/aarch64/aarch64-protos.h
37718 (aarch64_simd_gen_const_vector_dup): Likewise.
37719 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
37721 2017-01-19 David Malcolm <dmalcolm@redhat.com>
37723 * langhooks-def.h (lhd_type_for_size): New decl.
37724 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
37725 * langhooks.c (lhd_type_for_size): New function, taken from
37728 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
37730 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
37731 define_bypass for CR latency.
37732 (power9-cracked-alu): Update bypass latency and remove power9-branch.
37733 (power9-alu2): Add define_bypass for CR latency.
37735 (power9-mul): Update insn latency.
37736 (power9-mul-compare): Update insn latency, bypass latency and remove
37739 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37741 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
37743 * config/aarch64/aarch64.md
37744 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
37745 aarch64_nopcrelative_literal_loads.
37746 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
37748 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
37750 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
37751 TARGET_LOONGSON_3A.
37752 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
37754 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
37757 * config.gcc (supported_defaults): Add lxc1-sxc1.
37758 (with_lxc1_sxc1): Add validation.
37759 (all_defaults): Add lxc1-sxc1.
37760 * config/mips/mips.opt (mlxc1-sxc1): New option.
37761 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
37763 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
37764 __mips_no_lxc1_sxc1.
37765 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
37766 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
37767 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
37769 2017-01-19 Richard Biener <rguenther@suse.de>
37771 PR tree-optimization/72488
37772 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
37773 sure to restore SSA info.
37774 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
37776 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
37778 PR rtl-optimization/79121
37779 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
37780 of the inner type when shifting an extended value.
37782 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
37785 * symtab.c (symtab_node::equal_address_to): Fix comparing of
37786 interposable aliases.
37788 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
37791 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
37792 Use the evmergelohi instruction.
37793 (mov_si<mode>_e500_subreg4_2_le): Likewise.
37794 (mov_sitf_e500_subreg8_2_be): Likewise.
37795 (mov_sitf_e500_subreg12_2_le): Likewise.
37796 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
37797 (mov_si<mode>_e500_subreg4_2_be): Likewise.
37798 (mov_sitf_e500_subreg8_2_le): Likewise.
37799 (mov_sitf_e500_subreg12_2_be): Likewise.
37801 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37803 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
37804 attribute from vecsimple to vecperm.
37805 (altivec_vbpermq2): Likewise.
37807 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37810 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
37812 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37813 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
37814 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
37815 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
37816 case where N arg is SIZE_MAX.
37817 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
37818 (cmpstrsi): Add pattern.
37820 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
37822 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
37823 __builtin_vec_revb builtins.
37824 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
37825 built-in functions to support generation of the ISA 3.0 XXBR<x>
37826 vector byte reverse instructions.
37827 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
37828 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
37829 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
37830 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
37831 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
37832 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
37833 (P9V_BUILTIN_VEC_REVB): Likewise.
37834 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
37835 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
37836 (p9_xxbrq_v16qi): Likewise.
37837 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
37838 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
37839 (p9_xxbrh_v8hi): Likewise.
37840 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
37841 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
37842 vec_revb built-in functions.
37844 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
37846 PR rtl-optimization/78952
37847 * config/i386/i386.md (any_extract): New code iterator.
37848 (*insvqi_2): Use any_extract for source operand.
37849 (*insvqi_3): Use any_shiftrt for source operand.
37851 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
37853 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
37855 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
37857 2017-01-18 Matthias Klose <doko@ubuntu.com>
37859 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
37861 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37863 * config/rs6000/altivec.h (vec_bperm): Change #define.
37864 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
37865 (altivec_vbpermq2): New define_insn.
37866 (altivec_vbpermd): Likewise.
37867 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
37868 function interface.
37869 (VBPERMD): Likewise.
37870 (VBPERM): New polymorphic function interface.
37871 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
37872 Add entries for P9V_BUILTIN_VEC_VBPERM.
37873 * doc/extend.texi: Add interfaces for vec_bperm.
37875 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
37877 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
37878 first letter of error messages.
37879 (s390_resolve_overloaded_builtin): Likewise.
37880 * config/s390/s390.c (s390_expand_builtin): Likewise.
37881 (s390_invalid_arg_for_unprototyped_fn): Likewise.
37882 (s390_valid_target_attribute_inner_p): Likewise.
37883 * config/s390/s390.md ("tabort"): Likewise.
37885 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
37887 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
37888 (ISA_AVOID_DIV_HILO): New macro.
37889 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
37890 (ISA_HAS_DDIV): Likewise.
37892 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37894 * doc/invoke.texi (fabi-version): Correct number of occurrences.
37896 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37898 * doc/invoke.texi (fabi-version): Spelling fix.
37900 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37903 * doc/invoke.texi (fabi-version): Mention mangling fix for
37906 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
37909 * doc/invoke.texi (fabi-version): Document discriminator mangling.
37911 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
37914 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
37915 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
37917 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
37919 (stack_protect_test): Ditto.
37920 * config/rs6000/rs6000.opt (mstack-protector-guard=,
37921 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
37923 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
37924 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
37925 -mstack-protector-guard-offset=.
37926 (RS/6000 and PowerPC Options): Ditto.
37928 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
37930 * config/i386/i386.h (MASK_CLASS_P): New define.
37931 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
37932 there are no registers from different register sets also when
37933 mask registers are used. Update function comment.
37934 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
37935 to (*k/*r) and (*k/*km) alternatives.
37937 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
37939 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
37940 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
37941 (EH_RETURN_HANDLER_RTX): New define.
37942 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
37943 Force frame pointer in EH return functions.
37944 (aarch64_expand_epilogue): Add barrier for eh_return.
37945 (aarch64_final_eh_return_addr): Remove.
37946 (aarch64_eh_return_handler_rtx): New function.
37947 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
37949 (aarch64_eh_return_handler_rtx): New prototype.
37951 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37953 * config/rs6000/altivec.h (vec_rlmi): New #define.
37954 (vec_vrlnm): Likewise.
37955 (vec_rlnm): Likewise.
37956 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
37957 (UNSPEC_VRLNM): Likewise.
37958 (VIlong): New mode iterator.
37959 (altivec_vrl<VI_char>mi): New define_insn.
37960 (altivec_vrl<VI_char>nm): Likewise.
37961 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
37963 (VRLDNM): Likewise.
37964 (RLNM): New polymorphic function entry.
37965 (VRLWMI): New monomorphic function entry.
37966 (VRLDMI): Likewise.
37967 (RLMI): New polymorphic function entry.
37968 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
37969 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
37970 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
37973 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37976 * dwarf2out.c (field_byte_offset): Restore the
37977 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
37978 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
37979 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
37982 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37985 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
37987 2017-01-17 Jakub Jelinek <jakub@redhat.com>
37990 * dwarf2out.c (add_data_member_location_attribute): For constant
37991 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
37992 instead of DW_AT_data_member_location, DW_AT_bit_offset and
37993 DW_AT_byte_size attributes.
37995 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
37997 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
37998 after forcing to constant memory when the code model is medium.
38000 2017-01-17 Julia Koval <julia.koval@intel.com>
38003 * config/i386/avx512fintrin.h
38004 (_mm512_i32gather_ps): Change __addr type to void const*.
38005 (_mm512_mask_i32gather_ps): Ditto.
38006 (_mm512_i32gather_pd): Ditto.
38007 (_mm512_mask_i32gather_pd): Ditto.
38008 (_mm512_i64gather_ps): Ditto.
38009 (_mm512_mask_i64gather_ps): Ditto.
38010 (_mm512_i64gather_pd): Ditto.
38011 (_mm512_mask_i64gather_pd): Ditto.
38012 (_mm512_i32gather_epi32): Ditto.
38013 (_mm512_mask_i32gather_epi32): Ditto.
38014 (_mm512_i32gather_epi64): Ditto.
38015 (_mm512_mask_i32gather_epi64): Ditto.
38016 (_mm512_i64gather_epi32): Ditto.
38017 (_mm512_mask_i64gather_epi32): Ditto.
38018 (_mm512_i64gather_epi64): Ditto.
38019 (_mm512_mask_i64gather_epi64): Ditto.
38020 (_mm512_i32scatter_ps): Change __addr type to void*.
38021 (_mm512_mask_i32scatter_ps): Ditto.
38022 (_mm512_i32scatter_pd): Ditto.
38023 (_mm512_mask_i32scatter_pd): Ditto.
38024 (_mm512_i64scatter_ps): Ditto.
38025 (_mm512_mask_i64scatter_ps): Ditto.
38026 (_mm512_i64scatter_pd): Ditto.
38027 (_mm512_mask_i64scatter_pd): Ditto.
38028 (_mm512_i32scatter_epi32): Ditto.
38029 (_mm512_mask_i32scatter_epi32): Ditto.
38030 (_mm512_i32scatter_epi64): Ditto.
38031 (_mm512_mask_i32scatter_epi64): Ditto.
38032 (_mm512_i64scatter_epi32): Ditto.
38033 (_mm512_mask_i64scatter_epi32): Ditto.
38034 (_mm512_i64scatter_epi64): Ditto.
38035 (_mm512_mask_i64scatter_epi64): Ditto.
38036 * config/i386/avx512pfintrin.h
38037 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
38038 (_mm512_mask_prefetch_i32gather_ps): Ditto.
38039 (_mm512_mask_prefetch_i64gather_pd): Ditto.
38040 (_mm512_mask_prefetch_i64gather_ps): Ditto.
38041 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
38042 (_mm512_prefetch_i32scatter_ps): Ditto.
38043 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
38044 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
38045 (_mm512_prefetch_i64scatter_pd): Ditto.
38046 (_mm512_prefetch_i64scatter_ps): Ditto.
38047 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
38048 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
38049 * config/i386/avx512vlintrin.h
38050 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
38051 (_mm_mmask_i32gather_ps): Ditto.
38052 (_mm256_mmask_i32gather_pd): Ditto.
38053 (_mm_mmask_i32gather_pd): Ditto.
38054 (_mm256_mmask_i64gather_ps): Ditto.
38055 (_mm_mmask_i64gather_ps): Ditto.
38056 (_mm256_mmask_i64gather_pd): Ditto.
38057 (_mm_mmask_i64gather_pd): Ditto.
38058 (_mm256_mmask_i32gather_epi32): Ditto.
38059 (_mm_mmask_i32gather_epi32): Ditto.
38060 (_mm256_mmask_i32gather_epi64): Ditto.
38061 (_mm_mmask_i32gather_epi64): Ditto.
38062 (_mm256_mmask_i64gather_epi32): Ditto.
38063 (_mm_mmask_i64gather_epi32): Ditto.
38064 (_mm256_mmask_i64gather_epi64): Ditto.
38065 (_mm_mmask_i64gather_epi64): Ditto.
38066 (_mm256_i32scatter_ps): Change __addr type to void*.
38067 (_mm256_mask_i32scatter_ps): Ditto.
38068 (_mm_i32scatter_ps): Ditto.
38069 (_mm_mask_i32scatter_ps): Ditto.
38070 (_mm256_i32scatter_pd): Ditto.
38071 (_mm256_mask_i32scatter_pd): Ditto.
38072 (_mm_i32scatter_pd): Ditto.
38073 (_mm_mask_i32scatter_pd): Ditto.
38074 (_mm256_i64scatter_ps): Ditto.
38075 (_mm256_mask_i64scatter_ps): Ditto.
38076 (_mm_i64scatter_ps): Ditto.
38077 (_mm_mask_i64scatter_ps): Ditto.
38078 (_mm256_i64scatter_pd): Ditto.
38079 (_mm256_mask_i64scatter_pd): Ditto.
38080 (_mm_i64scatter_pd): Ditto.
38081 (_mm_mask_i64scatter_pd): Ditto.
38082 (_mm256_i32scatter_epi32): Ditto.
38083 (_mm256_mask_i32scatter_epi32): Ditto.
38084 (_mm_i32scatter_epi32): Ditto.
38085 (_mm_mask_i32scatter_epi32): Ditto.
38086 (_mm256_i32scatter_epi64): Ditto.
38087 (_mm256_mask_i32scatter_epi64): Ditto.
38088 (_mm_i32scatter_epi64): Ditto.
38089 (_mm_mask_i32scatter_epi64): Ditto.
38090 (_mm256_i64scatter_epi32): Ditto.
38091 (_mm256_mask_i64scatter_epi32): Ditto.
38092 (_mm_i64scatter_epi32): Ditto.
38093 (_mm_mask_i64scatter_epi32): Ditto.
38094 (_mm256_i64scatter_epi64): Ditto.
38095 (_mm256_mask_i64scatter_epi64): Ditto.
38096 (_mm_i64scatter_epi64): Ditto.
38097 (_mm_mask_i64scatter_epi64): Ditto.
38098 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
38099 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
38100 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
38101 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
38102 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
38103 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
38104 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
38105 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
38106 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
38107 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
38108 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
38109 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
38110 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
38111 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
38112 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
38113 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
38114 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
38115 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
38116 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
38117 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
38118 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
38119 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
38120 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
38121 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
38122 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
38123 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
38124 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
38125 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
38126 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
38127 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
38128 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
38129 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
38130 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
38131 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
38132 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
38133 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
38134 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
38135 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
38136 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
38137 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
38138 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
38139 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
38140 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
38141 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
38142 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
38143 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
38144 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
38145 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
38146 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
38147 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
38148 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
38149 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
38150 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
38151 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
38152 definitions accordingly.
38154 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
38155 Kuan-Lin Chen <kuanlinchentw@gmail.com>
38158 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
38161 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
38164 * ira-conflicts.c (ira_build_conflicts): Update total conflict
38165 hard regs for inner regno.
38167 2017-01-17 Martin Liska <mliska@suse.cz>
38170 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
38171 assumption and add comment.
38173 2017-01-17 Nathan Sidwell <nathan@acm.org>
38175 * ipa-visibility.c (localize_node): New function, broken out of ...
38176 (function_and_variable_visibility): ... here. Call it.
38178 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
38180 PR middle-end/77445
38181 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
38182 correctly set frequency of oudgoing edge.
38183 (duplicate_thread_path): Fix profile updating.
38185 2017-01-17 Jakub Jelinek <jakub@redhat.com>
38188 * configure.ac: Add GCC_BASE_VER.
38189 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
38190 version from BASE-VER file.
38191 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
38192 (gcc.o): Depend on $(BASEVER).
38193 * common.opt (dumpfullversion): New option.
38194 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
38195 * doc/invoke.texi: Document -dumpfullversion.
38196 * doc/install.texi: Document --with-gcc-major-version-only.
38197 * configure: Regenerated.
38199 2017-01-17 Richard Biener <rguenther@suse.de>
38201 PR tree-optimization/71433
38202 * tree-vrp.c (register_new_assert_for): Merge same asserts
38203 on all incoming edges.
38204 (process_assert_insertions_for): Handle insertions at the
38207 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
38209 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
38210 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
38212 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
38215 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
38218 2017-01-17 Alan Modra <amodra@gmail.com>
38221 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
38222 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
38223 symbolic stack limit when pic.
38225 2017-01-16 Martin Sebor <msebor@redhat.com>
38227 PR tree-optimization/78608
38228 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
38230 2017-01-16 Jeff Law <law@redhat.com>
38233 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
38234 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
38235 for several include directories that may be relative to sysroot.
38236 * config/i386/x-mingw32 (gplus_includedir): Define.
38237 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
38238 (native_system_includedir): Likewise.
38239 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
38240 override if TARGET_SYSTEM_ROOT is defined.
38241 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
38243 PR tree-optimization/79090
38244 PR tree-optimization/33562
38245 PR tree-optimization/61912
38246 PR tree-optimization/77485
38247 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
38248 and computed trims into the dump file.
38250 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
38252 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
38254 2017-01-16 Jakub Jelinek <jakub@redhat.com>
38257 * gimplify.c (gimplify_init_constructor): If want_value and
38258 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
38262 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
38263 sequence. Formatting fixes.
38264 (doloop_optimize): Formatting fixes.
38267 * gcc.c (debug_level_greater_than_spec_func): New function.
38268 (static_spec_functions): Add debug-level-gt spec function.
38269 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
38271 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38272 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
38273 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
38274 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
38275 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
38276 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
38278 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
38280 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
38281 QImode fixups to general and mask registers only.
38283 2017-01-16 Carl Love <cel@us.ibm.com>
38285 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
38286 for built-in functions
38287 vector signed char vec_nabs (vector signed char)
38288 vector signed short vec_nabs (vector signed short)
38289 vector signed int vec_nabs (vector signed int)
38290 vector signed long long vec_nabs (vector signed long long)
38291 vector float vec_nabs (vector float)
38292 vector double vec_nabs (vector double)
38293 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
38295 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
38296 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
38297 * doc/extend.texi: Update the documentation file for the new built-in
38300 2017-01-16 Martin Sebor <msebor@redhat.com>
38302 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
38305 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38307 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
38308 UNSPEC_VSX__XXSPLTD to require special splat handling.
38310 2017-01-16 David Malcolm <dmalcolm@redhat.com>
38313 * system.h: Poison strndup.
38315 2017-01-16 Alan Modra <amodra@gmail.com>
38318 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
38321 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
38323 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
38325 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
38327 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
38328 call recog here. Assert that INSN_CODE (insn) is non-negative.
38330 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38333 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
38335 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
38336 in the currently scheduled RTL fragment.
38338 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
38340 PR rtl-optimization/78751
38341 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
38344 2017-01-14 Jeff Law <law@redhat.com>
38346 PR tree-optimization/79090
38347 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
38348 variable length stores.
38349 (compute_trims): Delete dead assignment to *trim_tail.
38350 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
38353 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
38355 PR rtl-optimization/78626
38356 PR rtl-optimization/78727
38357 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
38358 of a block, and split such blocks after everything else is finished.
38360 2017-01-14 Alan Modra <amodra@gmail.com>
38363 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
38364 target legitimate_combined_insn.
38365 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
38366 (rs6000_legitimate_combined_insn): New function.
38367 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
38369 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
38370 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
38371 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
38373 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
38375 * doc/frontends.texi (G++ and GCC): Remove references to Java.
38377 2017-01-13 Jeff Law <law@redhat.com>
38379 PR tree-optimization/33562
38380 PR tree-optimization/61912
38381 PR tree-optimization/77485
38382 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
38384 (delete_dead_assignment): Likewise.
38385 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
38386 statement to delete_dead_call and delete_dead_assignment.
38388 2017-01-13 David Malcolm <dmalcolm@redhat.com>
38391 * substring-locations.c (format_warning_va): Strengthen case 1 so
38392 that both endpoints of the substring must be within the format
38393 range for just the substring to be printed.
38395 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
38397 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
38398 * config/i386/i386.c (ix86_target_string): Add missing options
38399 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
38400 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
38401 flags_other and ix86_target_other to flags2_other. Display unknown
38403 (ix86_valid_target_attribute_inner_p): Add missing options and
38404 reorder options by implied ISAs, as in ix86_target_string.
38406 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38408 * hash-table.h (hash_table::too_empty_p): New function.
38409 (hash_table::expand): Use it.
38410 (hash_table::traverse): Likewise.
38411 (hash_table::empty_slot): Use sizeof (value_type) instead of
38412 sizeof (PTR) to convert bytes to elements. Shrink the table
38413 if the current size is excessive for the current number of
38416 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
38418 * ira-costs.c (record_reg_classes): Break from the inner loop
38419 early once alt_fail is known to be true. Update outer loop
38420 handling accordingly.
38422 2017-01-13 Jeff Law <law@redhat.com>
38424 * tree-ssa-dse.c (decrement_count): New function.
38425 (increment_start_addr, maybe_trim_memstar_call): Likewise.
38426 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
38427 when we know the partially dead statement is a mem* function.
38429 PR tree-optimization/61912
38430 PR tree-optimization/77485
38431 * tree-ssa-dse.c: Include expr.h.
38432 (maybe_trim_constructor_store): New function.
38433 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
38435 PR tree-optimization/33562
38436 PR tree-optimization/61912
38437 PR tree-optimization/77485
38438 * doc/invoke.texi: Document new dse-max-object-size param.
38439 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
38440 * tree-ssa-dse.c: Include params.h.
38441 (dse_store_status): New enum.
38442 (initialize_ao_ref_for_dse): New, partially extracted from
38444 (valid_ao_ref_for_dse, normalize_ref): New.
38445 (setup_live_bytes_from_ref, compute_trims): Likewise.
38446 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
38447 (maybe_trim_partially_dead_store): Likewise.
38448 (maybe_trim_complex_store): Likewise.
38449 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
38450 Track what bytes live from the original store. Return tri-state
38451 for dead, partially dead or live.
38452 (dse_dom_walker): Add constructor, destructor and new private members.
38453 (delete_dead_call, delete_dead_assignment): New extracted from
38455 (dse_optimize_stmt): Make a member of dse_dom_walker.
38456 Use initialize_ao_ref_for_dse.
38458 PR tree-optimization/33562
38459 PR tree-optimization/61912
38460 PR tree-optimization/77485
38461 * sbitmap.h (bitmap_count_bits): Prototype.
38462 (bitmap_clear_range, bitmap_set_range): Likewise.
38463 * sbitmap.c (bitmap_clear_range): New function.
38464 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
38466 2017-01-13 Martin Liska <mliska@suse.cz>
38469 * function.c (set_cfun): Add new argument force.
38470 * function.h (set_cfun): Likewise.
38471 * ipa-inline-transform.c (inline_call): Use the function when
38472 strict alising from is dropped for function we inline to.
38474 2017-01-13 Richard Biener <rguenther@suse.de>
38476 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
38477 for dumping GIMPLE INTEGER_CSTs.
38479 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
38481 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
38482 to 201112L since C++17.
38484 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
38487 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
38488 if -fsanitize=kernel-address is present.
38490 2017-01-13 Richard Biener <rguenther@suse.de>
38492 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
38493 as _Literal ( type ) number in case usual suffixes do not
38494 preserve all information.
38496 2017-01-13 Richard Biener <rguenther@suse.de>
38498 PR tree-optimization/77283
38499 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
38500 and ssa-iterators.h.
38501 (is_feasible_trace): Implement a cost model based on joiner
38504 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
38507 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
38508 char or short to __float128/_Float128 directly.
38510 2017-01-12 Martin Sebor <msebor@redhat.com>
38512 to -Wformat-overflow.
38513 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
38514 (min_bytes_remaining): Same.
38515 (get_string_length): Same.
38516 (format_string): Same.
38517 (format_directive): Same.
38519 (pass_sprintf_length::handle_gimple_call): Same.
38521 2017-01-12 Jakub Jelinek <jakub@redhat.com>
38523 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
38524 info.nowrite calls with no lhs that can't throw. Return bool
38525 whether gsi_remove has been called or not.
38526 (pass_sprintf_length::handle_gimple_call): Return bool whether
38527 try_substitute_return_value called gsi_remove. Formatting fix.
38528 (pass_sprintf_length::execute): Don't use gsi_remove if
38529 handle_gimple_call returned true.
38532 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
38533 be removed due to side-effects, don't remove following barrier nor
38534 turn the successor edge into fallthru edge.
38536 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38539 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
38540 element-reversing loads and stores as not swappable.
38542 2017-01-12 Nathan Sidwell <nathan@acm.org>
38543 Nicolai Stange <nicstange@gmail.com>
38545 * combine.c (try_combine): Don't ignore result of overlap checking
38546 loop. Combine overlap & asm check into single loop.
38548 2017-01-12 Richard Biener <rguenther@suse.de>
38550 * tree-pretty-print.c (dump_generic_node): Provide -gimple
38551 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
38553 2017-01-12 Richard Biener <rguenther@suse.de>
38555 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
38556 and TS_TARGET_OPTION directly derive from TS_BASE.
38557 * tree-core.h (tree_optimization_option): Derive from tree_base.
38558 (tree_target_option): Likewise.
38560 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
38562 * config/i386/i386.c (memory_address_length): Increase len
38563 only when rip_relative_addr_p returns false.
38565 2017-01-11 Julia Koval <julia.koval@intel.com>
38567 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
38568 (OPTION_MASK_ISA_SGX_SET): New.
38569 (ix86_handle_option): Handle OPT_msgx.
38570 * config.gcc: Added sgxintrin.h.
38571 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
38572 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
38573 * config/i386/i386.c (ix86_target_string): Add -msgx.
38575 (ix86_option_override_internal): Handle new options.
38576 (ix86_valid_target_attribute_inner_p): Add sgx.
38577 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
38578 * config/i386/i386.opt: Add msgx.
38579 * config/i386/sgxintrin.h: New file.
38580 * config/i386/x86intrin.h: Add sgxintrin.h.
38582 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38585 * fold-const.c (maybe_nonzero_address): Return 1 for function
38587 (tree_single_nonzero_warnv_p): Don't handle function local objects
38591 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
38594 2017-01-11 David Malcolm <dmalcolm@redhat.com>
38597 * opts.c: Include "spellcheck.h"
38598 (struct string_fragment): New struct.
38599 (struct edit_distance_traits<const string_fragment &>): New
38601 (get_closest_sanitizer_option): New function.
38602 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
38604 2017-01-11 Jakub Jelinek <jakub@redhat.com>
38606 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
38608 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
38609 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
38610 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
38611 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
38612 for initial die_offset if dwarf_split_debug_info.
38613 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
38614 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
38616 (output_skeleton_debug_sections): Formatting fix. Use
38617 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
38618 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
38620 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
38622 * config/arm/cortex-a53.md: Add bypasses for
38623 cortex_a53_r2f_cvt.
38624 (cortex_a53_r2f): Only use for transfers.
38625 (cortex_a53_f2r): Likewise.
38626 (cortex_a53_r2f_cvt): Add reservation for conversions.
38627 (cortex_a53_f2r_cvt): Likewise.
38629 2017-01-11 Tamar Christina <tamar.christina@arm.com>
38631 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
38632 to all inlined functions, change static to extern.
38634 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
38637 * config/arm/arm.c (legitimize_pic_address): Handle reference to
38639 (arm_assemble_integer): Likewise.
38641 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
38643 * config.gcc: Use new awk script to check CPU, FPU and architecture
38644 parameters for --with-... options.
38645 * config/arm/parsecpu.awk: New file
38646 * config/arm/arm-cpus.in: New file.
38647 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
38649 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
38651 * config/arm/t-arm: Update dependency rules.
38652 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
38653 of processing .def files.
38654 * config/arm/genopt.sh: Deleted.
38655 * config/arm/gentune.sh: Deleted.
38656 * config/arm/arm-cores.def: Deleted.
38657 * config/arm/arm-arches.def: Deleted.
38658 * config/arm/arm-fpus.def: Deleted.
38659 * config/arm/arm-tune.md: Regenerated.
38660 * config/arm/arm-tables.opt: Regenerated.
38661 * config/arm/arm-cpu.h: New generated file.
38662 * config/arm/arm-cpu-data.h: New generated file.
38663 * config/arm/arm-cpu-cdata.h: New generated file.
38665 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
38668 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
38670 (input_varpool_node): Unpack dynamically_initialized bit.
38672 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
38674 PR rtl-optimization/79032
38675 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
38676 the alignment of the adjusted memory reference against that of MODE,
38677 instead of the alignment of the original memory reference.
38679 2017-01-11 Martin Jambor <mjambor@suse.cz>
38681 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
38683 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
38684 decorated functions.
38686 2017-01-11 Richard Biener <rguenther@suse.de>
38688 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
38689 set range/nonnull info for PHI results. Do not set it on
38690 stmts marked for removal.
38692 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
38694 * expr.c (store_field): In the bitfield case, fetch the return value
38695 from the registers before applying a single big-endian adjustment.
38696 Always do a final load for a BLKmode value not larger than a word.
38698 2017-01-10 David Malcolm <dmalcolm@redhat.com>
38701 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
38702 that we correctly handle column numbers greater than
38703 LINE_MAP_MAX_COLUMN_NUMBER.
38705 2017-01-10 Martin Sebor <msebor@redhat.com>
38707 PR middle-end/78245
38708 * gimple-ssa-sprintf.c (get_destination_size): Call
38709 {init,fini}object_sizes.
38710 * tree-object-size.c (addr_object_size): Adjust.
38711 (pass_through_call): Adjust.
38712 (pass_object_sizes::execute): Adjust.
38713 * tree-object-size.h (fini_object_sizes): Declare.
38715 2017-01-10 Martin Sebor <msebor@redhat.com>
38717 PR tree-optimization/78775
38718 * builtins.c (get_size_range): Move...
38719 * calls.c: ...to here.
38720 (alloc_max_size): Accept zero argument.
38721 (operand_signed_p): Remove.
38722 (maybe_warn_alloc_args_overflow): Call get_size_range.
38723 * calls.h (get_size_range): Declare.
38725 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
38727 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
38728 from TI's devices.csv file as of September 2016.
38729 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
38731 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
38733 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
38734 * doc/invoke.texi: Likewise.
38735 * doc/md.texi: Likewise.
38736 * doc/objc.texi: Likewise.
38738 2017-01-10 Joshua Conner <joshconner@google.com>
38740 * config/arm/fuchsia-elf.h: New file.
38741 * config/fuchsia.h: New file.
38742 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
38743 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
38745 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
38747 2016-01-10 Richard Biener <rguenther@suse.de>
38749 PR tree-optimization/79034
38750 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
38751 Propagate out degenerate PHIs in the joiner.
38753 2017-01-10 Martin Liska <mliska@suse.cz>
38755 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
38756 (sort_congruence_classes_by_decl_uid): Likewise.
38757 (sort_congruence_class_groups_by_decl_uid): Likewise.
38758 (sem_item_optimizer::merge_classes): Sort class, groups in these
38759 classes and members in the groups by DECL_UID of declarations.
38760 This would make merge operations stable.
38762 2017-01-10 Martin Liska <mliska@suse.cz>
38764 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
38765 usage of m_classes_vec.
38766 (sem_item_optimizer::~sem_item_optimizer): Likewise.
38767 (sem_item_optimizer::get_group_by_hash): Likewise.
38768 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
38769 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
38770 (sem_item_optimizer::verify_classes): Likewise.
38771 (sem_item_optimizer::process_cong_reduction): Likewise.
38772 (sem_item_optimizer::dump_cong_classes): Likewise.
38773 (sem_item_optimizer::merge_classes): Likewise.
38774 * ipa-icf.h (congruence_class_hash): Rename from
38775 congruence_class_group_hash. Remove declaration of m_classes_vec.
38777 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
38779 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
38780 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
38781 * config.gcc: Add avx512vpopcntdqintrin.h.
38782 * config/i386/avx512vpopcntdqintrin.h: New.
38783 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
38784 * config/i386/i386-builtin-types.def: Add new types.
38785 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
38786 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
38787 __builtin_ia32_vpopcountq_v8di_mask): New.
38788 * config/i386/i386-c.c (ix86_target_macros_internal): Define
38789 __AVX512VPOPCNTDQ__.
38790 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
38791 (PTA_AVX512VPOPCNTDQ): Define.
38792 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
38793 TARGET_AVX512VPOPCNTDQ_P): Define.
38794 * config/i386/i386.opt: Add mavx512vpopcntdq.
38795 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
38796 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
38798 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38800 PR middle-end/77484
38801 * predict.def (PRED_CALL): Set to 67.
38803 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
38805 * expr.c (store_field): In the bitfield case, if the value comes from
38806 a function call and is of an aggregate type returned in registers, do
38807 not modify the field mode; extract the value in all cases if the mode
38808 is BLKmode and the size is not larger than a word.
38810 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
38813 * config/i386/cpuid.h: Fix undefined behavior.
38815 2017-01-04 Jeff Law <law@redhat.com>
38817 PR tree-optimization/79007
38818 PR tree-optimization/67955
38819 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
38820 conservative for pt.null when flag_non_call_exceptions is on.
38822 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38824 PR translation/79019
38825 PR translation/79020
38826 * params.def (PARAM_INLINE_MIN_SPEEDUP,
38827 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
38828 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
38830 * config/avr/avr.opt (maccumulate-args): Likewise.
38831 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
38832 * common.opt (freport-bug): Likewise.
38833 * cif-code.def (CIF_FINAL_ERROR): Likewise.
38834 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
38835 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
38836 translatable string.
38837 * config/i386/i386.c (function_value_32): Likewise.
38838 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
38839 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
38841 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
38842 * common/config/msp430/msp430-common.c (msp430_handle_option):
38844 * symtab.c (symtab_node::verify_base): Likewise.
38845 * opts.c (set_debug_level): Likewise.
38846 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
38847 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
38848 missing whitespace to translatable strings.
38849 * config/avr/avr.md (bswapsi2): Fix typo in comment.
38850 * config/sh/superh.h: Likewise.
38851 * config/i386/xopintrin.h: Likewise.
38852 * config/i386/znver1.md: Likewise.
38853 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
38854 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
38855 * double-int.h (struct double_int): Likewise.
38856 * double-int.c (div_and_round_double): Likewise.
38857 * wide-int.cc: Likewise.
38858 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
38859 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
38860 * cfgcleanup.c (crossjumps_occured): Renamed to ...
38861 (crossjumps_occurred): ... this.
38862 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
38865 PR tree-optimization/78899
38866 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
38867 returning bool return struct loop *, NULL for failure and the new
38869 (versionable_outer_loop_p): Don't version outer loop if it has
38870 dont_vectorized bit set.
38871 (tree_if_conversion): When versioning outer loop, ensure
38872 tree_if_conversion is performed also on the inner loop of the
38873 non-vectorizable outer loop copy.
38874 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
38875 LOOP_VECTORIZED in inner loop of the scalar outer loop and
38876 prevent vectorization of it.
38877 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
38878 the outer loop vectorization of the non-scalar version is attempted
38879 before vectorization of the inner loop in scalar version. If
38880 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
38881 vectorization of its inner loop.
38882 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
38883 has 2 inner loops, rename also on edges from bb whose single pred
38884 is outer_loop->header. Fix typo in function comment.
38886 2017-01-09 Martin Sebor <msebor@redhat.com>
38889 * asan.c (asan_emit_stack_protection): Increase local buffer size
38890 to avoid snprintf truncation warning.
38892 2017-01-09 Andrew Pinski <apinski@cavium.com>
38894 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
38895 to reference thunderx2t99 for the tuning structure
38896 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
38898 (thunderx2t99_extra_costs): This.
38899 * config/aarch64/aarch64-tune.md: Regenerate.
38900 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
38901 (vulcan_addrcost_table): This.
38902 (vulcan_regmove_cost): Rename to ...
38903 (thunderx2t99_regmove_cost): This.
38904 (vulcan_vector_cost): Rename to ...
38905 (thunderx2t99_vector_cost): this.
38906 (vulcan_branch_cost): Rename to ...
38907 (thunderx2t99_branch_cost): This.
38908 (vulcan_tunings): Rename to ...
38909 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
38910 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
38912 2017-01-09 Martin Jambor <mjambor@suse.cz>
38916 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
38917 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
38918 (propagate_vr_accross_jump_function): Use the above function for all
38919 value range computations for pass-through jump functions and type
38920 converasion from explicit value range values.
38921 (ipcp_propagate_stage): Do not attempt to deduce types of formal
38922 parameters from TYPE_ARG_TYPES.
38923 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
38924 (ipa_write_node_info): Stream type of the actual argument.
38925 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
38927 2017-01-09 Martin Liska <mliska@suse.cz>
38930 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
38931 (lookup_compiler): Do not show error message with have_E.
38933 2017-01-09 Jakub Jelinek <jakub@redhat.com>
38935 PR tree-optimization/78938
38936 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
38937 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
38938 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
38939 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
38942 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38944 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
38947 2017-01-09 Richard Biener <rguenther@suse.de>
38949 PR tree-optimization/78997
38950 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
38951 name condition properly.
38953 2017-01-09 Richard Biener <rguenther@suse.de>
38956 * dwarf2out.c (is_cxx): New overload with context.
38957 (is_naming_typedef_decl): Use it.
38959 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38961 * invoke.texi (Option Summary): Correct spacing in option lists
38962 and add line breaks to fix over-long lines.
38964 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38966 PR middle-end/17660
38968 * extend.texi (Common Variable Attributes): Add xref to GCC
38969 Internals manual to explain mode attribute keywords.
38971 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
38974 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
38975 and Preprocessor Options.
38976 (Options for Linking): Document -pthread here....
38977 (RS/6000 and PowerPC Options): ...not here.
38978 (Solaris 2 Options): ...or here.
38979 * doc/cppopts.texi: Document -pthread.
38981 2017-01-08 Martin Sebor <msebor@redhat.com>
38983 PR middle-end/77708
38984 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
38985 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
38986 New member functions.
38987 (format_directive): Used them.
38989 (pass_sprintf_length::handle_gimple_call): Same.
38990 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
38991 to avoid truncation for any argument.
38992 (extract_affine_mul): Same.
38993 * tree.c (get_file_function_name): Same.
38995 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
38997 PR middle-end/77484
38998 * predict.def (PRED_INDIR_CALL): Set to 86.
39000 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39002 PR preprocessor/54124
39003 * doc/cppopts.texi: Reformat -d subtable to list the full name
39004 of the options. Add cross-reference to the docs for the general
39005 compiler -d options.
39006 * doc/invoke.texi (Developer Options): Add cross-reference to the
39007 preprocessor-specific -d option documentation.
39009 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39011 PR preprocessor/13498
39012 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
39013 redudant material, and reflect new command-line options.
39014 (System Headers): Likewise.
39016 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
39018 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
39019 -isystem, and -idirafter. Copy-edit.
39020 * doc/cppopts.texi: Copy-edit. Remove contradiction about
39021 default for -ftrack-macro-expansion. Delete obsolete and
39022 badly-formatted implementation details about -fdebug-cpp output.
39023 * doc/cppwarnopts.texi: Copy-edit.
39025 2017-01-07 David Malcolm <dmalcolm@redhat.com>
39028 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
39029 that the transition from a max line width >= 1<<10 to narrower
39030 lines works correctly.
39032 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
39034 * doc/options.texi (PerFunction): New.
39035 * opt-functions.awk (switch_flags): Map both Optimization and
39036 PerFunction to CL_OPTIMIZATION.
39037 * opth-gen.awk: Test for PerFunction flag along with
39039 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
39040 it only when the latter is present. Skip those that don't in
39041 the hash function generator.
39042 * common.opt (fvar-tracking): Mark as PerFunction instead of
39044 (fvar-tracking-assignments): Likewise.
39045 (fvar-tracking-assignments-toggle): Likewise.
39046 (fvar-tracking-uninit): Likewise.
39048 2017-01-07 Jakub Jelinek <jakub@redhat.com>
39050 PR translation/79018
39051 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
39054 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
39057 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
39058 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
39059 TARGET_LONG_JUMP_TABLE_OFFSETS.
39060 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
39061 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
39062 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
39063 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
39064 * config/m68k/m68k.md (tablejump expander): Likewise.
39065 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
39066 TARGET_LONG_JUMP_TABLE_OFFSETS.
39067 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
39068 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
39070 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
39071 David Holsgrove <david.holsgrove@xilinx.com>
39073 * common/config/microblaze/microblaze-common.c
39074 (TARGET_EXCEPT_UNWIND_INFO): Remove.
39075 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
39077 * config/microblaze/microblaze.c (microblaze_must_save_register)
39078 (microblaze_expand_epilogue, microblaze_return_addr): Handle
39080 (microblaze_eh_return): New function.
39081 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
39082 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
39083 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
39084 * config/microblaze/microblaze.md (eh_return): New pattern.
39086 2017-01-06 Jakub Jelinek <jakub@redhat.com>
39088 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
39089 GCC_DIAGNOSTIC_STRINGIFY): Define.
39091 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
39093 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39095 * config/arm/arm.md (<mcrr>): New.
39097 * config/arm/arm.c (arm_arch5te): New.
39098 (arm_option_override): Set arm_arch5te.
39099 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
39101 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
39102 (arm_mcrr_qualifiers): ... this. New.
39103 (MRRC_QUALIFIERS): Define to...
39104 (arm_mrrc_qualifiers): ... this. New.
39105 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
39107 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
39108 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
39109 (MRRCI, mrrc, MRRC): New.
39110 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
39111 VUNSPEC_MRRC2): New.
39113 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39115 * config/arm/arm.md (<mcr>): New.
39117 * config/arm/arm.c (arm_coproc_builtin_available): Add
39118 support for mcr, mrc, mcr2 and mrc2.
39119 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
39120 (arm_mcr_qualifiers): ... this. New.
39121 (MRC_QUALIFIERS): Define to ...
39122 (arm_mrc_qualifiers): ... this. New.
39123 (MCR_QUALIFIERS): Define to ...
39124 (arm_mcr_qualifiers): ... this. New.
39125 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
39127 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
39128 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
39129 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
39130 VUNSPEC_MRC2): New.
39132 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39134 * config/arm/arm.md (*ldc): New.
39138 * config/arm/arm.c (arm_coproc_builtin_available): Add
39139 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
39140 (arm_coproc_ldc_stc_legitimate_address): New.
39141 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
39142 'qualifier_const_pointer'.
39143 (LDC_QUALIFIERS): Define to...
39144 (arm_ldc_qualifiers): ... this. New.
39145 (STC_QUALIFIERS): Define to...
39146 (arm_stc_qualifiers): ... this. New.
39147 * config/arm/arm-protos.h
39148 (arm_coproc_ldc_stc_legitimate_address): New.
39149 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
39150 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
39151 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
39152 stc2, stcl, stc2l): New.
39153 * config/arm/constraints.md (Uz): New.
39154 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
39155 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
39156 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
39157 VUNSPEC_STC2L): New.
39159 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39161 * config/arm/arm.md (<cdp>): New.
39162 * config/arm/arm.c (neon_const_bounds): Rename this ...
39163 (arm_const_bounds): ... this.
39164 (arm_coproc_builtin_available): New.
39165 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
39166 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
39167 (CDP_QUALIFIERS): Define to...
39168 (arm_cdp_qualifiers): ... this. New.
39170 (arm_expand_builtin_args): Add case for 6 arguments.
39171 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
39172 (arm_const_bounds): ... this.
39173 (arm_coproc_builtin_available): New.
39174 * config/arm/arm_acle.h (__arm_cdp): New.
39176 * config/arm/arm_acle_builtins.def (cdp): New.
39178 * config/arm/iterators.md (CDPI,CDP,cdp): New.
39179 * config/arm/neon.md: Rename all 'neon_const_bounds' to
39180 'arm_const_bounds'.
39181 * config/arm/types.md (coproc): New.
39182 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
39183 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
39184 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
39185 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
39187 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39189 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
39190 (UBINOP_QUALIFIERS): New.
39192 (acle_builtin_data): New. Change comment.
39193 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
39194 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
39195 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
39196 arm_acle_builtins.def.
39197 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
39198 (arm_init_acle_builtins): New.
39199 (CRC32_BUILTIN): Remove.
39200 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
39201 crc32cb, crc32ch and crc32cw.
39202 (arm_init_crc32_builtins): Remove.
39203 (arm_init_builtins): Use arm_init_acle_builtins rather
39204 than arm_init_crc32_builtins.
39205 (arm_expand_acle_builtin): New.
39206 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
39207 * config/arm/arm_acle_builtins.def: New.
39209 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
39211 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
39212 (arm_builtin_datum): ... this.
39213 (arm_init_neon_builtin): Rename to ...
39214 (arm_init_builtin): ... this. Add a new parameters PREFIX
39215 and USE_SIG_IN_NAME.
39216 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
39217 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
39218 'arm_builtin_datum'.
39219 (arm_init_vfp_builtins): Likewise.
39220 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
39221 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
39222 (arm_expand_neon_args): Rename to ...
39223 (arm_expand_builtin_args): ... this. Rename builtin_arg
39224 enum values and differentiate between ARG_BUILTIN_MEMORY
39225 and ARG_BUILTIN_NEON_MEMORY.
39226 (arm_expand_neon_builtin_1): Rename to ...
39227 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
39228 values, arm_expand_builtin_args and add bool parameter NEON.
39229 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
39230 (arm_expand_vfp_builtin): Likewise.
39231 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
39233 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39235 PR middle-end/77484
39236 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
39237 * predict.c (tree_estimate_probability_bb): Reverse direction of
39238 polymorphic call predictor.
39240 2017-01-06 David Malcolm <dmalcolm@redhat.com>
39242 * passes.c (execute_one_pass): Split out pass-skipping logic into...
39243 (determine_pass_name_match): ...this new function and...
39244 (should_skip_pass_p): ...this new function.
39246 2017-01-06 Nathan Sidwell <nathan@acm.org>
39248 * ipa-visibility.c (function_and_variable_visibility): Reformat
39249 comments and long lines. Remove extrneous if.
39250 * symtab.c (symtab_node::make_decl_local): Fix code format.
39251 (symtab_node::set_section_for_node): Fix comment typo.
39253 2017-01-06 Martin Liska <mliska@suse.cz>
39256 * lra-constraints.c: Rename invariant to lra_invariant.
39257 * predict.c (set_even_probabilities): Initialize e to NULL.
39259 2017-01-05 Martin Sebor <msebor@redhat.com>
39261 PR tree-optimization/78910
39262 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
39263 (format_integer): Correct off-by-one error in the handling
39264 of precision with negative numbers in signed conversions..
39266 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
39268 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
39270 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39272 PR tree-optimization/71016
39273 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
39274 factor_out_conditional_conversion. Formatting fix.
39275 (factor_out_conditional_conversion): Add cond_stmt argument.
39276 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
39277 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
39280 2017-01-05 David Malcolm <dmalcolm@redhat.com>
39282 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
39283 read-rtl-function.o, and selftest-rtl.o.
39284 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
39285 (selftest::aarch64_test_loading_full_dump): New function.
39286 (selftest::aarch64_run_selftests): New function.
39287 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
39288 selftest::aarch64_run_selftests.
39289 * config/i386/i386.c
39290 (selftest::ix86_test_loading_dump_fragment_1): New function.
39291 (selftest::ix86_test_loading_call_insn): New function.
39292 (selftest::ix86_test_loading_full_dump): New function.
39293 (selftest::ix86_test_loading_unspec): New function.
39294 (selftest::ix86_run_selftests): Call the new functions.
39295 * emit-rtl.c (maybe_set_max_label_num): New function.
39296 * emit-rtl.h (maybe_set_max_label_num): New decl.
39297 * function.c (instantiate_decls): Guard call to
39298 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
39299 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
39301 * gensupport.c (gen_reader::gen_reader): Pass "false"
39302 for new "compact" param of rtx_reader.
39303 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
39304 rather than an empty string for NULL strings.
39305 * read-md.c: Potentially include config.h rather than bconfig.h.
39306 Wrap include of errors.h with #ifdef GENERATOR_FILE.
39307 (have_error): New global, copied from errors.c.
39308 (md_reader::read_name): Rename to...
39309 (md_reader::read_name_1): ...this, adding "out_loc" param,
39310 and converting "missing name or number" to returning false, rather
39312 (md_reader::read_name): Reimplement in terms of read_name_1.
39313 (md_reader::read_name_or_nil): New function.
39314 (md_reader::read_string): Handle "(nil)" by returning NULL.
39315 (md_reader::md_reader): Add new param "compact".
39316 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
39317 (md_reader::read_file): New method.
39318 * read-md.h (md_reader::md_reader): Add new param "compact".
39319 (md_reader::read_file): New method.
39320 (md_reader::is_compact): New accessor.
39321 (md_reader::read_name): Convert return type from void to file_location.
39322 (md_reader::read_name_or_nil): New decl.
39323 (md_reader::read_name_1): New decl.
39324 (md_reader::m_compact): New field.
39325 (noop_reader::noop_reader): Pass "false" for new "compact" param
39327 (rtx_reader::rtx_reader): Add new "compact" param.
39328 (rtx_reader::read_rtx_operand): Make virtual and convert return
39329 type from void to rtx.
39330 (rtx_reader::read_until): New decl.
39331 (rtx_reader::handle_any_trailing_information): New virtual function.
39332 (rtx_reader::postprocess): New virtual function.
39333 (rtx_reader::finalize_string): New virtual function.
39334 (rtx_reader::m_in_call_function_usage): New field.
39335 (rtx_reader::m_reuse_rtx_by_id): New field.
39336 * read-rtl-function.c: New file.
39337 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
39338 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
39339 (selftest::verify_three_block_rtl_cfg): New decl.
39340 * read-rtl-function.h: New file.
39341 * read-rtl.c: Potentially include config.h rather than bconfig.h.
39342 For host, include function.h, memmodel.h, and emit-rtl.h.
39343 (one_time_initialization): New function.
39344 (struct compact_insn_name): New struct.
39345 (compact_insn_names): New array.
39346 (find_code): Handle insn codes in compact dumps.
39347 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
39348 (bind_subst_iter_and_attr): Likewise.
39349 (add_condition_to_string): Likewise.
39350 (add_condition_to_rtx): Likewise.
39351 (apply_attribute_uses): Likewise.
39352 (add_current_iterators): Likewise.
39353 (apply_iterators): Likewise.
39354 (initialize_iterators): Guard usage of apply_subst_iterator with
39355 #ifdef GENERATOR_FILE.
39356 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
39357 (md_reader::read_mapping): Likewise.
39358 (add_define_attr_for_define_subst): Likewise.
39359 (add_define_subst_attr): Likewise.
39360 (read_subst_mapping): Likewise.
39361 (check_code_iterator): Likewise.
39362 (rtx_reader::read_rtx): Likewise. Move one-time initialization
39364 (one_time_initialization): New function.
39365 (rtx_reader::read_until): New method.
39366 (read_flags): New function.
39367 (parse_reg_note_name): New function.
39368 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
39369 Handle reuse_rtx ids.
39370 Wrap iterator lookup within #ifdef GENERATOR_FILE.
39371 Add parsing support for RTL dumps, mirroring the special-cases in
39372 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
39373 values, and calling handle_any_trailing_information.
39374 (rtx_reader::read_rtx_operand): Convert return type from void
39375 to rtx, returning return_rtx. Handle case 'e'. Call
39376 finalize_string on XSTR and XTMPL fields.
39377 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
39378 "(nil)" values were omitted. Call the postprocess vfunc on the
39380 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
39381 class ctor. Initialize m_in_call_function_usage. Call
39382 one_time_initialization.
39383 * rtl-tests.c (selftest::test_uncond_jump): Call
39384 set_new_first_and_last_insn.
39385 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
39386 * selftest-rtl.c: New file.
39387 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
39388 (selftest::get_insn_by_uid): New decl.
39389 * selftest-run-tests.c (selftest::run_tests): Call
39390 read_rtl_function_c_tests.
39391 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
39392 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
39395 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
39397 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
39398 operands in a special way. Assert that pos+len <= mode precision.
39400 2017-01-05 Jakub Jelinek <jakub@redhat.com>
39402 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
39403 3 argument Alias with unlimited for the negative form.
39404 (fno-vect-cost-model): Removed.
39406 2017-01-05 Martin Liska <mliska@suse.cz>
39408 * hsa-gen.c (gen_hsa_divmod): New function.
39409 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
39411 2017-01-05 Martin Liska <mliska@suse.cz>
39414 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
39417 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39419 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
39420 small constant length operands.
39422 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39424 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
39425 between loop iterations.
39427 2017-01-05 Martin Liska <mliska@suse.cz>
39430 * gimplify.c (gimplify_decl_expr): Compare to
39431 asan_poisoned_variables instread of checking flags.
39432 (gimplify_target_expr): Likewise.
39433 (gimplify_expr): Likewise.
39434 (gimplify_function_tree): Conditionally initialize
39435 asan_poisoned_variables.
39437 2017-01-04 Jeff Law <law@redhat.com>
39439 PR tree-optimizatin/78812
39440 * rtl.h (contains_mem_rtx_p): Prototype.
39441 * ifcvt.c (containts_mem_rtx_p): Move from here to...
39442 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
39443 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
39444 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
39445 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
39447 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39449 * input.c (assert_char_at_range): Default-initialize actual_range.
39451 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39453 * df-scan.c (df_ref_create_structure): Make regno unsigned,
39454 to match the caller.
39456 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39458 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
39459 insns after final jump in test to emit dummy move.
39461 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39463 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
39464 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
39466 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
39468 * multiple_target.c (create_dispatcher_calls): Init e_next.
39469 * tree-ssa-loop-split.c (split_loop): Init border.
39470 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
39473 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
39478 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
39479 (altivec_register_operand): Do not return true if the operand
39480 contains a SUBREG mixing SImode and SFmode.
39481 (vsx_register_operand): Likewise.
39482 (vsx_reg_sfsubreg_ok): New predicate.
39483 (vfloat_operand): Do not return true if the operand contains a
39484 SUBREG mixing SImode and SFmode.
39485 (vint_operand): Likewise.
39486 (vlogical_operand): Likewise.
39487 (gpc_reg_operand): Likewise.
39488 (int_reg_operand): Likewise.
39489 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
39490 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
39491 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
39493 (rs6000_emit_move_si_sf_subreg): New helper function.
39494 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
39495 fixup SUBREGs involving SImode and SFmode.
39496 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
39497 numbers for the new peephole2 optimization.
39498 (peephole2 for SFmode unions): New peephole2 to optimize cases in
39499 the GLIBC math library that do AND/IOR/XOR operations on single
39500 precision floating point.
39501 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
39502 target macros to say whether we need to avoid SUBREGs mixing
39504 (TARGET_ALLOW_SF_SUBREG): Likewise.
39505 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
39506 (UNSPEC_SI_FROM_SF): Likewise.
39507 (iorxor): Change spacing.
39508 (and_ior_xor): New iterator for AND, IOR, and XOR.
39509 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
39510 (movdi_from_sf_zero_ext): Likewise.
39511 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
39512 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
39513 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
39514 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
39515 (fms<mode>4): Likewise.
39516 (fnma<mode>4): Likewise.
39517 (fnms<mode>4): Likewise.
39518 (nfma<mode>4): Likewise.
39519 (nfms<mode>4): Likewise.
39521 2017-01-04 Marek Polacek <polacek@redhat.com>
39524 * doc/invoke.texi: Document -Wpointer-compare.
39526 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39528 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
39531 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
39532 descriptions for -gdwarf-5 and emit them as uleb128 instead of
39535 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
39538 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
39539 documentation of the powerpc_popcntb_ok attribute.
39540 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
39541 code to issue warning messages if a requested CPU configuration is
39542 not supported by the binary (assembler and loader) toolchain.
39543 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
39544 made to define a built-in function that has been disabled.
39545 (paired_init_builtins): Add assertion to prevent ICE if attempt is
39546 made to define a built-in function that has been disabled.
39547 (altivec_init_builtins): Add comment explaining why definition
39548 of the DST built-in functions is not preceded by an assertion
39549 check. Add assertions to prevent ICE if attempts are made to
39550 define an altivec predicate or an abs* built-in function that has
39552 (htm_init_builtins): Add comment explaining why definition of the
39553 htm built-in functions is not preceded by an assertion check.
39555 2017-01-04 Jeff Law <law@redhat.com>
39557 PR tree-optimizatin/67955
39558 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
39559 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
39560 the points-to solution does not include pt_null. Use DECL_PT_UID
39563 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
39565 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
39566 Use gen_int_mode instead of gen_lopwart for const_int operands.
39568 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39570 PR tree-optimization/71563
39571 * match.pd: Simplify X << Y into X if Y is known to be 0 or
39572 out of range value - has low bits known to be zero.
39574 2017-01-04 Alan Modra <amodra@gmail.com>
39576 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
39577 * configure: Regenerate.
39578 * config.in: Regenerate.
39580 2017-01-04 Jakub Jelinek <jakub@redhat.com>
39583 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
39584 a substring of the message, but strcmp with the whole message. Ifdef
39585 ENABLE_NLS, translate the message first using dgettext.
39587 2017-01-03 Jeff Law <law@redhat.com>
39589 PR tree-optimizatin/78856
39590 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
39591 (mark_threaded_blocks): Remove code to truncate thread paths that
39592 cross multiple loop headers. Instead invalidate the cached loop
39593 iteration information and handle case of a thread path walking
39594 into an irreducible region.
39596 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
39599 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
39600 assertions. Add support for doing the signbit if the IEEE 128-bit
39601 floating point value is in a GPR.
39602 * config/rs6000/rs6000.md (Fsignbit): Delete.
39603 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
39604 Update the length attribute if the value is in a GPR.
39605 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
39606 the sign or zero extension instruction, since the value is always 0/1.
39607 (signbit<mode>2_dm2): Delete using <Fsignbit>.
39610 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
39611 extracting SImode to a GPR register so that we can generate a
39612 store, limit the vector to be in a traditional Altivec register
39613 for the vextuwrx instruction.
39615 2017-01-03 Ian Lance Taylor <iant@google.com>
39617 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
39619 2017-01-03 Martin Sebor <msebor@redhat.com>
39621 PR tree-optimization/78696
39622 * gimple-ssa-sprintf.c (format_floating): Correct handling of
39623 precision. Use MPFR for %f for greater fidelity. Correct handling
39625 (pass_sprintf_length::compute_format_length): Set width and precision
39626 specified by asrerisk to void_node for vararg functions.
39627 (try_substitute_return_value): Adjust dump output.
39629 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
39631 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
39633 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
39635 * doc/invoke.texi (SPARC options): Document -mlra as the default.
39636 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
39637 -mlra/-mno-lra was passed to the compiler.
39639 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
39641 PR rtl-optimization/65618
39642 * emit-rtl.c (try_split): Move initialization of "before" and
39643 "after" to just before the call to emit_insn_after_setloc.
39645 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
39647 * doc/md.texi (Standard Names): Remove reference to Java frontend.
39649 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
39651 * dwarf2out.c (gen_enumeration_type_die): When
39652 -gno-strict-dwarf, add a DW_AT_encoding attribute.
39654 2017-01-03 Jakub Jelinek <jakub@redhat.com>
39656 PR tree-optimization/78965
39657 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
39658 Change first argument from const call_info & to call_info &. For %n
39659 set info.nowrite to false.
39661 PR middle-end/78901
39662 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
39663 possibly throwing calls.
39665 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
39666 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
39667 and fns handling, rather than in a separate case SSA_NAME.
39669 2017-01-02 Jeff Law <law@redhat.com>
39671 * config/darwin-driver.c (darwin_driver_init): Const-correctness
39672 fixes for first_period and second_period variables.
39674 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
39677 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
39678 (*insvqi_1): New insn pattern.
39679 (*insvqi_1_mem_rex64): Ditto.
39680 (*insvqi_2): Ditto.
39681 (*insvqi_3): Rename from *insvqi.
39683 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
39685 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
39687 * doc/cfg.texi (Edges): Remove reference to Java.
39688 (Maintaining the CFG): Ditto.
39690 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39692 PR middle-end/77674
39693 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
39694 transparent aliases.
39696 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
39698 PR middle-end/77484
39699 * predict.def (PRED_CALL): Update hitrate.
39700 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
39701 * predict.c (tree_estimate_probability_bb): Split CALL predictor
39702 into direct/indirect/polymorphic variants.
39704 2017-01-01 Jakub Jelinek <jakub@redhat.com>
39706 Update copyright years.
39708 * gcc.c (process_command): Update copyright notice dates.
39709 * gcov-dump.c (print_version): Ditto.
39710 * gcov.c (print_version): Ditto.
39711 * gcov-tool.c (print_version): Ditto.
39712 * gengtype.c (create_file): Ditto.
39713 * doc/cpp.texi: Bump @copying's copyright year.
39714 * doc/cppinternals.texi: Ditto.
39715 * doc/gcc.texi: Ditto.
39716 * doc/gccint.texi: Ditto.
39717 * doc/gcov.texi: Ditto.
39718 * doc/install.texi: Ditto.
39719 * doc/invoke.texi: Ditto.
39721 Copyright (C) 2017 Free Software Foundation, Inc.
39723 Copying and distribution of this file, with or without modification,
39724 are permitted in any medium without royalty provided the copyright
39725 notice and this notice are preserved.