1 2017-03-17 Martin Jambor <mjambor@suse.cz>
3 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
4 (List of -O2 options): Likewise.
5 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
8 2017-03-17 Tom de Vries <tom@codesourcery.com>
10 * gcov-dump.c (print_usage): Print bug_report_url.
12 2017-03-17 Richard Biener <rguenther@suse.de>
15 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
16 (parser::peek): Likewise.
18 2017-03-17 Richard Biener <rguenther@suse.de>
20 PR tree-optimization/80048
21 * sese.c (free_sese_info): Properly release rename_map and
22 copied_bb_map elements.
24 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
26 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
27 Add linked-list forward and backlinks. Insert on
28 construction, remove on destruction.
29 (class pass_store_merging): Add m_stores_head field.
30 (pass_store_merging::terminate_and_process_all_chains):
31 Iterate over m_stores_head list.
32 (pass_store_merging::terminate_all_aliasing_chains):
34 (pass_store_merging::execute): Check for debug stmts first.
35 Push new chains onto the m_stores_head stack.
37 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
40 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
41 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
42 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
44 2017-03-16 Jeff Law <law@redhat.com>
46 PR tree-optimization/71437
47 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
48 member function. Implementation moved into after_dom_children
49 member function and into the threader's thread_outgoing_edges
51 (dom_opt_dom_walker::after_dom_children): Simplify by moving
52 some code into new thread_outgoing_edges.
53 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
54 definition. Simplify marker handling (do it here). Assume we always
55 have the available expression and the const/copies tables.
56 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
58 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
59 * tree-vrp.c (equiv_stack): No longer file scoped.
60 (vrp_dom_walker): New class.
61 (vrp_dom_walker::before_dom_children): New member function.
62 (vrp_dom_walker::after_dom_children): Likewise.
63 (identify_jump_threads): Setup domwalker. Use it rather than
64 walking edges in a random order by hand. Simplify setup/finalization.
65 (finalize_jump_threads): Remove.
66 (vrp_finalize): Do not call identify_jump_threads here.
67 (execute_vrp): Do it here instead and call thread_through_all_blocks
70 PR tree-optimization/71437
71 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
73 (simplify_stmt_for_jump_threading): Add basic_block argument. All
75 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
76 (dom_opt_dom_walker::thread_across_edge): Remove
77 handle_dominating_asserts argument. All callers changed.
78 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
79 changes. Remove calls to lhs_of_dominating_assert. Other
80 uses of handle_dominating_asserts turn into unconditional code
81 (simplify_control_stmt_condition_1): Likewise.
82 (simplify_control_stmt_condition): Likewise.
83 (thread_through_normal_block, thread_across_edge): Likewise.
84 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
85 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
86 object if it is not an SSA_NAME.
87 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
88 before calling into the VRP specific simplifiers.
89 (identify_jump_threads): Remove handle_dominating_asserts
92 2017-03-16 Jakub Jelinek <jakub@redhat.com>
95 * tree-diagnostic.c (default_tree_printer): No longer static.
96 * tree-diagnostic.h (default_tree_printer): New prototype.
98 2017-03-16 Tamar Christina <tamar.christina@arm.com>
100 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
101 Change ins into fmov.
103 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
105 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
106 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
107 Use h_con constraint for operand 1.
108 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
109 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
111 2017-03-15 Jeff Law <law@redhat.com>
113 PR tree-optimization/71437
114 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
115 (record_temporary_equivalences): Use it.
117 PR tree-optimization/71437
118 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
119 tree-ssa-scopedtables.
120 (lookup_avail_expr, build_and_record_new_cond): Likewise.
121 (record_conditions, record_cond, vuse_eq): Likewise.
122 (record_edge_info): Adjust to API tweak of record_conditions.
123 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
124 (record_temporary_equivalences, optimize_stmt): Likewise.
125 (eliminate_redundant_computations): Likewise.
126 (record_equivalences_from_stmt): Likewise.
127 * tree-ssa-scopedtables.c: Include options.h and params.h.
128 (vuse_eq): New function, moved from tree-ssa-dom.c
129 (build_and_record_new_cond): Likewise.
130 (record_conditions): Likewise. Accept vector of conditions rather
131 than edge_equivalence structure for first argument.
132 for the first argument.
133 (avail_exprs_stack::lookup_avail_expr): New member function, moved
135 (avail_exprs_stack::record_cond): Likewise.
136 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
138 (avail_exprs_stack): Add new member functions lookup_avail_expr
140 (record_conditions): Declare.
142 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
145 * lra-constraints.c (process_alt_operands): Increase reject for
146 reloading an input/output operand.
148 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
151 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
152 insns to convert from signed/unsigned char/short to IEEE 128-bit
154 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
156 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
159 * config/i386/i386.c (ix86_vector_duplicate_value): Create
160 subreg of inner mode for values already in registers.
162 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
164 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
165 iteration reg is used after the loop.
167 2017-03-14 Martin Sebor <msebor@redhat.com>
169 PR tree-optimization/79800
170 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
171 precision in negative-positive range.
172 (format_floating): Call non-const overload with adjusted precision.
174 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
177 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
180 2017-03-14 Martin Sebor <msebor@redhat.com>
183 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
184 * builtins.def (aligned_alloc): Use it.
187 * Makefile.in (GTFILES): Add calls.c.
188 * calls.c: Include "gt-calls.h".
190 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
192 PR rtl-optimization/79728
193 * regs.h (struct target_regs): New field
194 x_contains_allocatable_regs_of_mode.
195 (contains_allocatable_regs_of_mode): New macro.
196 * reginfo.c (init_reg_sets_1): Initialize it, and change
197 contains_reg_of_mode so it includes global regs as well.
198 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
199 rather than contains_regs_of_mode.
201 2017-03-14 Martin Liska <mliska@suse.cz>
203 * doc/invoke.texi: Document options that can't be combined with
204 -fcheck-pointer-bounds.
206 2017-03-14 Martin Liska <mliska@suse.cz>
209 * doc/invoke.texi (-Wchkp): Document the option.
211 2017-03-14 Martin Liska <mliska@suse.cz>
213 * Makefile.in: Install gcov-dump.
215 2017-03-14 Martin Liska <mliska@suse.cz>
217 * multiple_target.c (expand_target_clones): Bail out for
218 an invalid attribute.
220 2017-03-14 Richard Biener <rguenther@suse.de>
222 * alias.c (struct alias_set_entry): Pack properly.
223 * cfgloop.h (struct loop): Likewise.
224 * cse.c (struct set): Likewise.
225 * ipa-utils.c (struct searchc_env): Likewise.
226 * loop-invariant.c (struct invariant): Likewise.
227 * lra-remat.c (struct cand): Likewise.
228 * recog.c (struct change_t): Likewise.
229 * rtl.h (struct address_info): Likewise.
230 * symbol-summary.h (function_summary): Likewise.
231 * tree-loop-distribution.c (struct partition): Likewise.
232 * tree-object-size.c (struct object_size_info): Likewise.
233 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
234 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
235 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
236 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
237 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
238 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
239 (struct _stmt_vec_info): Likewise.
241 2017-03-14 Martin Liska <mliska@suse.cz>
244 * multiple_target.c (create_dispatcher_calls): Check that
245 a target can create a function dispatcher.
247 2017-03-14 Martin Liska <mliska@suse.cz>
250 * multiple_target.c (expand_target_clones): Drop local.local
251 flag for default implementation.
253 2017-03-14 Richard Biener <rguenther@suse.de>
255 PR tree-optimization/80030
256 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
258 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
260 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
261 gcc_fallthrough() instead of __attribute__((fallthrough));
263 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
265 * doc/gcc.texi: Remove "up" link to (DIR).
266 * doc/gccint.texi: Ditto.
268 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
270 * doc/install.texi (Specific) <avr>: Remove reference to
273 2017-03-13 Jeff Law <law@redhat.com>
275 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
276 attribute rather than comments.
278 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
279 match_scratch operand is highest.
281 2017-03-13 Martin Liska <mliska@suse.cz>
284 * ipa-pure-const.c (warn_function_noreturn): If the declarations
285 is a CHKP clone, use original declaration.
287 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
289 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
290 (arc_conditional_register_usage): Use a different allocation order
291 when optimizing for size.
292 * common/config/arc/arc-common.c (arc_option_optimization_table):
293 Section anchors default on when optimizing for size.
295 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
297 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
299 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
301 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
302 * config/arc/arc.md (cpu_facility): Add cd variant.
303 (*movqi_insn): Add code density variant.
304 (*movhi_insn): Likewise.
305 (*movqi_insn): Likewise.
306 (*addsi3_mixed): Likewise.
307 (subsi3_insn): Likewise.
309 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
311 * config/arc/arc.md (movsi_cond_exec): Update constraint.
313 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
315 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
316 expressions with MINUS and UNARY ops.
318 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
321 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
323 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
324 between vec_select and vector argument.
325 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
326 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
327 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
328 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
329 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
330 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
332 2017-03-13 Richard Biener <rguenther@suse.de>
335 * params.def (vect-max-peeling-for-alignment): Fix typo.
337 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
339 * doc/install.texi (Specific) <mips-*-*>: Remove description of
340 issue that only occurred with binutils below 2.18.
342 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
344 * doc/install.texi (Specific) <cris-axis-elf>: No longer
345 refer to binutils 2.11/2.12 minimum.
347 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
349 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
350 ftp.kernel.org and simplify binutils requirement.
352 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
354 * doc/invoke.texi (Warning Options): Fix spelling of link-time
356 (Optimize Options): Ditto. Also remove redundancy.
358 2017-03-10 David Malcolm <dmalcolm@redhat.com>
361 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
363 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
364 to G_ to avoid double translation.
366 2017-03-10 David Malcolm <dmalcolm@redhat.com>
369 * auto-profile.c (get_combined_location): Convert leading
370 character of diagnostics to lower case and remove trailing period.
371 (read_profile): Likewise for various diagnostics.
372 * config/arm/arm.c (arm_option_override): Remove trailing period
373 from various diagnostics.
374 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
375 (msp430_expand_delay_cycles): Likewise.
377 2017-03-10 David Malcolm <dmalcolm@redhat.com>
380 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
381 full command-line argument, rather than just "str".
382 (aarch64_validate_march): Likewise.
383 (aarch64_validate_mtune): Likewise.
385 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
387 PR rtl-optimization/78911
388 * lra-assigns.c (must_not_spill_p): New function.
391 2017-03-10 Jakub Jelinek <jakub@redhat.com>
393 PR tree-optimization/79981
394 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
395 ATOMIC_COMPARE_EXCHANGE ifn result.
396 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
397 IFN_ATOMIC_COMPARE_EXCHANGE.
399 2017-03-10 David Malcolm <dmalcolm@redhat.com>
402 * opts.c (parse_sanitizer_options): Add missing question mark to
403 "did you mean" message.
405 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
407 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
409 (VMULEUH_UNS): Likewise.
410 (VMULOUB_UNS): Likewise.
411 (VMULOUH_UNS): Likewise.
412 * config/rs6000/rs6000.c (builtin_function_type): Remove
413 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
415 2017-03-10 David Malcolm <dmalcolm@redhat.com>
418 * read-rtl-function.c (function_reader::read_rtx_operand): Update
419 x with result of extra_parsing_for_operand_code_0.
420 (function_reader::extra_parsing_for_operand_code_0): Convert
421 return type from void to rtx, returning x. When reading
422 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
423 larger size containing struct block_symbol.
425 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
427 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
428 -mfloat128-hardware without -m64.
430 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
433 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
434 entries to the case statement that marks unsigned arguments to
435 overloaded functions.
437 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
439 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
440 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
442 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
445 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
446 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
448 2017-03-10 Martin Liska <mliska@suse.cz>
452 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
453 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
456 2017-03-10 Olivier Hainque <hainque@adacore.com>
458 * tree-switch-conversion (array_value_type): Start by resetting
459 candidate type to it's main variant.
461 2017-03-10 Jakub Jelinek <jakub@redhat.com>
463 PR rtl-optimization/79909
464 * combine.c (try_combine): Use simplify_replace_rtx on individual
465 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
466 of the whole CALL_INSN_FUNCTION_USAGE.
468 PR tree-optimization/79972
469 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
470 get_range_info on SSA_NAMEs. Formatting fixes.
472 2017-03-10 Richard Biener <rguenther@suse.de>
473 Jakub Jelinek <jakub@redhat.com>
475 PR tree-optimization/77975
476 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
478 (get_val_for): For constant x return it. Formatting fix.
479 (loop_niter_by_eval): Avoid pointless looping if the next iteration
480 would use the same bases as the current one.
482 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
484 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
485 instead of vec_select for V1TImode.
486 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
488 (VSX_LE_128): Add V1TI to this mode iterator.
489 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
490 (*vsx_le_perm_store_<mode>): Likewise.
491 (pre-reload splitter for VSX stores): Likewise.
492 (post-reload splitter for VSX stores): Likewise.
493 (*vsx_xxpermdi2_le_<mode>): Likewise.
494 (*vsx_lxvd2x2_le_<mode>): Likewise.
495 (*vsx_stxvd2x2_le_<mode>): Likewise.
497 2017-03-09 Michael Eager <eager@eagercon.com>
499 Correct failures with --enable-checking=yes,rtl.
501 * config/microblaze/microblaze.c (microblaze_expand_shift):
502 Replace GET_CODE test with CONST_INT_P and INTVAL test with
504 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
505 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
507 2017-03-09 Richard Biener <rguenther@suse.de>
509 PR tree-optimization/79977
510 * graphite-scop-detection.c (scop_detection::merge_sese):
511 Handle the case of extra exits to blocks dominating the entry.
513 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
515 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
518 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
520 PR rtl-optimization/79949
521 * lra-constraints.c (process_alt_operands): Check memory when
522 trying to predict a cycle. Print about the overall increase.
524 2017-03-09 Richard Biener <rguenther@suse.de>
527 * gimple-expr.c (useless_type_conversion_p): Preserve
528 TYPE_SATURATING for fixed-point types.
530 2017-03-09 Richard Biener <rguenther@suse.de>
533 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
534 alignment of BLKmode params.
536 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
539 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
540 (VALL_NO_V2Q): Likewise.
542 * config/aarch64/aarch64-simd.md
543 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
545 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
546 VALL_NO_V2Q mode iterator.
547 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
549 2017-03-09 Martin Liska <mliska@suse.cz>
551 PR tree-optimization/79631
552 * tree-chkp-opt.c (chkp_is_constant_addr): Call
553 tree_int_cst_sign_bit just for INTEGER constants.
555 2017-03-09 Martin Liska <mliska@suse.cz>
559 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
562 2017-03-09 Marek Polacek <polacek@redhat.com>
565 * tree.c (inchash::add_expr): Handle TREE_VEC.
567 2017-03-09 Martin Liska <mliska@suse.cz>
570 (chkp_narrow_size_and_offset): New function.
571 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
572 (void chkp_parse_bit_field_ref): New function.
573 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
574 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
576 2017-03-09 Martin Liska <mliska@suse.cz>
579 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
580 (chkp_find_bounds_1): Remove gcc_unreachable.
582 2017-03-09 Jakub Jelinek <jakub@redhat.com>
585 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
586 BUILT_IN_SYNC*, determine the access type from the size suffix and
587 always build a MEM_REF with that type. Handle forgotten
588 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
591 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
592 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
593 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
594 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
595 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
596 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
597 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
598 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
599 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
600 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
601 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
602 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
603 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
604 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
605 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
606 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
607 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
608 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
609 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
610 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
611 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
612 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
613 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
614 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
615 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
616 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
617 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
618 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
619 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
620 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
621 definitions outside of __OPTIMIZE__ guarded section.
624 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
625 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
626 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
627 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
630 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
632 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
633 ("vfenez<mode>"): Add missing constraints.
635 2017-03-08 Martin Sebor <msebor@redhat.com>
638 * config/nds32/nds32.c (nds32_option_override):
640 2017-03-08 Jakub Jelinek <jakub@redhat.com>
643 * gimplify.c (gimplify_omp_for): Replace index var in outer
644 taskloop statement with an artificial variable and add
645 OMP_CLAUSE_PRIVATE clause for it.
647 2017-03-08 Richard Biener <rguenther@suse.de>
649 PR tree-optimization/79955
650 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
651 for accesses that are completely outside of the variable.
653 2017-03-08 Andrew Haley <aph@redhat.com>
655 PR tree-optimization/79943
656 * tree-ssa-loop-split.c (compute_new_first_bound): When
657 calculating the new upper bound, (END-BEG) should be added, not
660 2017-03-08 Jakub Jelinek <jakub@redhat.com>
662 * config/avr/avr.md (setmemhi): Make sure match_dup
663 operand number comes before match_scratch.
665 2017-03-08 Richard Biener <rguenther@suse.de>
667 PR tree-optimization/79920
668 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
669 with ncopies == 1 to ...
670 (vect_transform_slp_perm_load): ... here. Properly compute
671 all element loads by iterating VF times over the group. Do
672 not handle ncopies (computed in a broken way) in
673 vect_create_mask_and_perm.
675 2017-03-08 Jakub Jelinek <jakub@redhat.com>
678 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
679 is a uniform vector, use uniform_vector_p return value instead of
680 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
682 2017-03-07 Marek Polacek <polacek@redhat.com>
685 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
686 (alloca_call_type): Likewise.
688 2017-03-07 Martin Liska <mliska@suse.cz>
690 * gcov.c (process_args): Put comment to correct location.
692 2017-03-07 Martin Liska <mliska@suse.cz>
695 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
696 Use array_at_struct_end_p instead of DECL_CHAIN (field).
697 (chkp_narrow_bounds_for_field): Likewise.
698 (chkp_parse_array_and_component_ref): Pass one more argument to
701 2017-03-07 Richard Biener <rguenther@suse.de>
703 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
706 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
708 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
709 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
711 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
714 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
715 to end of description.
716 (PARAM_MAX_STORES_TO_MERGE): Likewise.
718 2017-03-07 Jakub Jelinek <jakub@redhat.com>
720 PR rtl-optimization/79901
721 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
723 (*avx512f_<code><mode>3<mask_name>): ... this.
724 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
725 iterator instead of VI8_AVX2_AVX512BW.
727 PR rtl-optimization/79901
728 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
729 min/max expander, expand it using expand_vec_cond_expr.
732 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
735 2017-03-06 Jakub Jelinek <jakub@redhat.com>
738 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
739 to void * for PCH reasons.
740 * dwarf2out.c (output_loc_operands, output_die): Cast
741 v.val_vec.array to unsigned char *.
743 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
746 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
749 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
751 PR rtl-optimization/79571
752 * lra-constraints.c (process_alt_operands): Calculate static
753 reject and subtract it from overall when only addresses will be
756 2017-03-06 Julia Koval <julia.koval@intel.com>
759 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
760 incoming stack boundary to 128 for 64-bit targets.
762 2017-03-06 Richard Biener <rguenther@suse.de>
764 PR tree-optimization/79894
765 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
766 to NULL after folding it.
768 2017-03-06 Richard Biener <rguenther@suse.de>
770 PR tree-optimization/79824
771 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
772 check disabling peeling for gaps.
774 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
776 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
777 attributes): Document gettimeofday.
779 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
781 * config/s390/s390.c (s390_option_override_internal): Set
782 PARAM_MIN_VECT_LOOP_BOUND
784 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
786 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
787 * config/s390/s390.md: Likewise.
789 2017-03-06 Jakub Jelinek <jakub@redhat.com>
792 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
793 (<avx2_avx512>_perm<mode>): Rename to ...
794 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
796 (<avx512>_perm<mode>_mask): Rename to ...
797 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
799 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
800 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
801 instead of VI8F_256_512.
802 (avx512f_perm<mode>): New define_expand.
803 (avx512f_perm<mode>_mask): Likewise.
804 (avx512f_perm<mode>_1<mask_name>): New define_insn.
805 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
807 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
809 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
810 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
812 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
814 2017-03-06 Martin Liska <mliska@suse.cz>
817 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
818 when having a SSA NAME w/o VAR_DECL assigned to it.
820 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
822 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
823 msa_dpsub_<su>_d): Fix MODE for vec_select.
825 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
827 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
829 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
831 2017-03-06 Richard Biener <rguenther@suse.de>
833 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
834 * plugin.c (register_plugin_info): Likewise.
835 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
837 2017-03-05 Jakub Jelinek <jakub@redhat.com>
839 * config/i386/sse.md (sse_storehps, sse_storelps,
840 avx_<castmode><avxsizesuffix>_<castmode>,
841 avx512f_<castmode><avxsizesuffix>_<castmode>,
842 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
843 in condition that at least one operand is not a MEM.
845 2017-03-03 Jakub Jelinek <jakub@redhat.com>
848 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
849 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
851 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
852 gimple_call_nothrow_p flag based on whether original builtin can throw.
853 If it can, emit following stmts on the fallthrough edge.
854 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
855 don't create new bb if inserting just debug stmts on the edge, try to
856 insert them on the fallthru bb or just reset debug stmts.
858 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
861 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
862 restore recog_data (including the operand rtxes inside it) around
863 the call to get_insn_template.
865 2017-03-03 Martin Sebor <msebor@redhat.com>
867 PR tree-optimization/79699
868 * context.c (context::~context): Free MPFR caches to avoid
869 a memory leak on program exit.
871 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
873 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
874 Use wide_int::ulow () instead of .elt (0).
876 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
878 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
879 (*pushxf): Limit oF constraint to 32bit targets and add oC
880 constraint for 64bit targets.
881 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
882 (*pushdf): Change rmF constraint to rmC.
884 2017-03-03 Martin Liska <mliska@suse.cz>
886 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
887 Remove unused variable.
889 2017-03-03 Jakub Jelinek <jakub@redhat.com>
892 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
893 is a memory operand, increase num_memory.
894 (ix86_expand_args_builtin): Likewise.
896 2017-03-03 Jan Hubicka <jh@suse.cz>
899 * ipa-devirt.c (maybe_record_node): Properly handle
900 __cxa_pure_virtual visibility.
902 2017-03-03 Martin Liska <mliska@suse.cz>
904 PR tree-optimization/79803
905 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
907 (pass_loop_prefetch::execute): Disabled optimization if an
908 assumption about L1 cache size is not met.
910 2017-03-03 Martin Liska <mliska@suse.cz>
912 PR rtl-optimization/79574
913 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
914 (hash_scan_set): Likewise.
915 (dump_hash_table): Likewise.
916 (hoist_code): Likewise.
918 2017-03-03 Richard Biener <rguenther@suse.de>
920 * fixed-value.c (fixed_from_string): Restore use of elt (1)
921 in place of uhigh ().
922 (fixed_convert_from_real): Likewise.
924 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
927 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
929 2017-03-03 Richard Biener <rguenther@suse.de>
932 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
933 TYPE_OVERFLOW_UNDEFINED check.
935 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
937 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
939 (vector_ae_<mode>_p): Likewise.
940 (vector_nez_<mode>_p): Likewise.
941 (vector_ne_v2di_p): Likewise.
942 (vector_ae_v2di_p): Likewise.
943 (vector_ne_<mode>_p): Likewise.
944 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
946 (vsx_tsqrt<mode>2_fe): Likewise.
948 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
951 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
953 2017-03-02 Jakub Jelinek <jakub@redhat.com>
955 PR rtl-optimization/79780
956 * cprop.c (one_cprop_pass): When second and further conditional trap
957 in a single basic block is turned into an unconditional trap, turn it
958 into a deleted note to avoid RTL verification failures.
960 2017-03-02 Richard Biener <rguenther@suse.de>
962 * fold-const.c (const_binop): Use ulow () instead of elt (0).
964 2017-03-02 Richard Biener <rguenther@suse.de>
966 PR tree-optimization/79345
968 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
969 param and abort the walk, returning -1 if it is hit.
970 (walk_aliased_vdefs): Take a limit param and pass it on.
971 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
972 defaulting to 0 and return a signed int.
973 * tree-ssa-uninit.c (struct check_defs_data): New struct.
974 (check_defs): New helper.
975 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
976 about uninitialized memory.
977 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
978 bogus uninitialized warning.
979 (fixed_convert_from_real): Likewise.
981 2017-03-02 Bin Cheng <bin.cheng@arm.com>
983 PR tree-optimization/66768
984 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
985 iv_use if base object can't be determined.
987 2017-03-02 Jakub Jelinek <jakub@redhat.com>
989 PR tree-optimization/79345
990 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
991 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
992 (get_pattern_stats): Initialize it.
993 * genemit.c (gen_expand): Verify match_scratch numbers come after
994 match_operand/match_dup numbers.
995 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
996 match_scratch numbers.
997 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
999 * config/s390/s390.md (trunctdsd2): Likewise.
1001 2017-03-02 Richard Biener <rguenther@suse.de>
1003 * wide-int.h (wide_int_storage::operator=): Implement in terms
1006 2017-03-02 Richard Biener <rguenther@suse.de>
1008 PR tree-optimization/79777
1009 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
1010 the to insert expression to sth existing.
1012 2017-03-01 Martin Sebor <msebor@redhat.com>
1015 * gimple-ssa-sprintf.c
1016 (directive::known_width_and_precision): New function.
1017 (format_integer): Use it.
1018 (get_mpfr_format_length): Consider the full range of precision
1019 when computing %g output with the # flag. Set the likely byte
1020 count to 3 rather than 1 when precision is indeterminate.
1021 (format_floating): Correct the lower bound of precision.
1023 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1025 * doc/invoke.texi: Document default code model for 64-bit Linux.
1027 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1030 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
1031 udiv rather than div since input pattern is unsigned.
1033 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
1035 * config/i386/i386.c (print_reg): Warn for values of
1036 unsupported size in integer register.
1038 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
1041 * config/rs6000/predicates.md (current_file_function_operand): Do
1042 not allow self calls to be local if the function is replaceable.
1044 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
1047 * config/rs6000/altivec.h (vec_ctz and others): Change the
1048 preprocessor macro that controls conditional compilation from
1049 _ARCH_PWR9 to __POWER9_VECTOR__.
1050 (vec_all_ne): Change parameterization of __altivec_scalar_pred
1051 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
1052 control (instead of _ARCH_PWR9 control) so that template
1053 definition uses power9-specific function.
1054 (vec_any_eq): Likewise.
1055 (vec_all_ne): Change macro definition to use a power9-specific
1056 expansion under #ifdef __POWER9_VECTOR__ control (instead of
1057 _ARCH_PWR9 control).
1058 (vec_any_eq) Likewise.
1059 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
1060 expansion for CMPNEF to remove support for xvcmpnesp instruction.
1061 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
1062 support for xvcmpnedp instruction.
1063 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
1064 macro expansion so that Power9 implementation of vec_all_ne does
1065 not use the AltiVec predicate framework.
1066 (VCMPNEH_P): Likewise.
1067 (VCMPNEW_P): Likewise.
1068 (VCMPNED_P): Likewise.
1069 (VCMPNEFP_P): Likewise.
1070 (VCMPNEDP_P): Likewise.
1071 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
1072 implementation of vec_any_eq to not use AltiVec predicate
1074 (VCMPAEH_P): Likewise.
1075 (VCMPAEW_P): Likewise.
1076 (VCMPAED_P): Likewise.
1077 (VCMPAEFP_P): Likewise.
1078 (VCMPAEDP_P): Likewise.
1079 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
1080 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
1081 not use the AltiVec predicate framework.
1082 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
1083 of vec_any_eq to not use AltiVec predicate framework.
1084 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
1085 support for predefined __POWER9_VECTOR__ macro to indicate that
1086 Power9 instruction selection is enabled.
1087 (altivec_overloaded_builtins): Remove extraneous
1088 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
1089 function argument types RS6000_BTI_bool_V16QI and
1090 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
1091 entry for overloaded function argument types RS6000_BTI_bool_V4SI
1092 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
1093 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
1094 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
1095 Power9 for implementations of vec_cmpne. Change the signature for
1096 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
1097 (representing vec_all_ne) to remove the previously described first
1098 argument of type RS6000_BTI_INTSI, as this was an artifact of
1099 reliance on the AltiVec predicate framework, which is no longer
1100 used in the implementation of these functions. Add
1101 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
1102 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
1103 since, unlike the AltiVec predicate framework implementation, we
1104 do not share function descriptors between vec_alle and vec_anyeq.
1105 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
1106 set of modes that receive special treatment even when
1107 TARGET_P9_VECTOR is true. The special treatment emits code that
1108 does not depend on Power9 instructions.
1109 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
1110 define_expand to not rely on AltiVec predicate framework.
1111 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
1113 (vector_ne_v2di_p): Change this define_expand to not rely on
1114 AltiVec predicate framework.
1115 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
1117 (vector_ne_<mode>_p): Change this define_expand to not rely on
1118 AltiVec predicate framework.
1119 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
1121 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
1122 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
1123 define_insn pattern.
1124 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
1125 define_insn pattern because the xvcmpne<VSs>. instruction is not
1127 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
1128 instruction is not supported.
1130 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1132 * config/nvptx/nvptx.c: Include intl.h.
1134 2017-03-01 Martin Jambor <mjambor@suse.cz>
1137 * ipa-prop.h (ipa_bits): Removed field known.
1138 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
1139 to pointers. Adjusted their comments to warn about their sharing.
1140 (ipcp_transformation_summary): Change bits to a vector of pointers.
1141 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
1142 (ipa_get_ipa_bits_for_value): Declare.
1143 * tree-vrp.h (value_range): Mark as GTY((for_user)).
1144 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
1145 (ipa_bits_hash_table): Likewise.
1146 (ipa_vr_ggc_hash_traits): Likewise.
1147 (ipa_vr_hash_table): Likewise.
1148 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
1149 being pointers and vr_known being removed.
1150 (ipa_set_jf_unknown): Likewise.
1151 (ipa_get_ipa_bits_for_value): New function.
1152 (ipa_set_jfunc_bits): Likewise.
1153 (ipa_get_value_range): New overloaded functions.
1154 (ipa_set_jfunc_vr): Likewise.
1155 (ipa_compute_jump_functions_for_edge): Use the above functions to
1156 construct bits and vr parts of jump functions.
1157 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
1158 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
1160 (ipcp_grow_transformations_if_necessary): Also allocate
1161 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
1163 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
1164 them. Fix too long lines.
1165 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
1166 vr_known being removed.
1167 (ipa_read_jump_function): Use new setter functions to construct bits
1168 and vr parts of jump functions or set them to NULL.
1169 (write_ipcp_transformation_info): Adjust for bits being pointers.
1170 (read_ipcp_transformation_info): Likewise.
1171 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
1173 Include gt-ipa-prop.h.
1174 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
1176 (ipcp_store_bits_results): Likewise.
1177 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
1178 Do not write to existing jump functions but use a temporary instead.
1180 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1183 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
1184 attempt to use its first operand as BIT_FIELD_REF base.
1186 2017-03-01 Richard Biener <rguenther@suse.de>
1189 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
1190 interpolating formula in wrapping arithmetic.
1191 (chrec_apply): Convert chrec_evaluate return value to wanted type.
1193 2017-03-01 Jakub Jelinek <jakub@redhat.com>
1195 PR tree-optimization/79734
1196 * tree-vect-generic.c (expand_vector_condition): Optimize
1197 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
1198 Handle VEC_COND_EXPR where comparison has different inner width from
1201 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
1203 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
1204 markup, and similar issues. Remove @opindex entries for things
1205 that aren't options. Add missing -mmpy-option entries.
1207 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1209 PR tree-optimization/79737
1210 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
1211 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
1212 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
1213 instead of byte_size. Formatting fix.
1214 (shift_bytes_in_array_right): Formatting fix.
1216 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
1219 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
1220 condition on optimize for the leaf function test.
1222 2017-02-28 Martin Liska <mliska@suse.cz>
1225 * read-rtl-function.c (function_reader::handle_unknown_directive):
1226 Bail out when one uses -flto.
1228 2017-02-28 Martin Liska <mliska@suse.cz>
1230 * common.opt: Replace space with tabular for options of <number>
1232 * config/i386/i386.opt: Show <number> value for
1233 -mlarge-data-threshold.
1234 * opts.c (print_filtered_help): Do not display number in hexadecimal
1237 2017-02-28 Martin Liska <mliska@suse.cz>
1239 * common.opt: Fix --help=option -Q for options which are of
1242 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
1244 * config/i386/i386.c (print_reg): Error out for values
1245 of 8-bit size in invalid integer register.
1247 2017-02-28 Martin Sebor <msebor@redhat.com>
1249 PR tree-optimization/79691
1250 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
1252 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1255 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
1256 gcc_unreachable with output_operand_lossage.
1258 2017-02-28 Richard Biener <rguenther@suse.de>
1260 PR tree-optimization/79740
1261 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
1263 (visit_nary_op): Insert the nary into the hashtable if we
1264 pattern-matched sth.
1265 * tree-ssa-pre.c (eliminate_insert): Robustify.
1267 2017-02-28 Richard Biener <rguenther@suse.de>
1270 * fold-const.c (decode_field_reference): Reject out-of-bound
1273 2017-02-28 Jakub Jelinek <jakub@redhat.com>
1275 * config/i386/i386.c: Include intl.h.
1276 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
1277 instead of just cond ? "..." : "...".
1278 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
1279 * coverage.c (read_counts_file): Likewise.
1280 * omp-offload.c: Include intl.h.
1281 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
1282 of just cond ? "..." : "...".
1283 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
1284 of just cond ? "..." : "...".
1286 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
1289 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
1291 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
1292 'tune for' CPU name.
1293 * config/arm/arm-cpu-data.h: Regenerated.
1295 2017-02-28 Richard Biener <rguenther@suse.de>
1297 PR tree-optimization/79732
1298 * tree-inline.c (expand_call_inline): Do not shadow var.
1300 2017-02-28 Richard Biener <rguenther@suse.de>
1302 PR tree-optimization/79723
1303 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
1304 address-space properly.
1306 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
1308 * doc/optinfo.texi (Optimization groups): Fix option used for
1310 * doc/invoke.texi (-fopt-info): Document "omp".
1311 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
1312 (OPTGROUP_ALL): Add OPTGROUP_OMP.
1313 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
1314 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
1315 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
1317 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
1319 * dumpfile.c (optgroup_options): Instead of "openmp", associate
1320 OPTGROUP_OMP with "omp".
1322 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
1325 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
1326 for arithmetic shift of unsigned V2DI.
1328 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
1330 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
1331 arc/linux.h headers.
1332 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
1333 (LINK_SPEC): Likewise.
1334 (ARC_TLS_EXTRA_START_SPEC): Likewise.
1335 (EXTRA_SPECS): Likewise.
1336 (STARTFILE_SPEC): Likewise.
1337 (ENDFILE_SPEC): Likewise.
1338 (LIB_SPEC): Likewise.
1339 (TARGET_SDATA_DEFAULT): Likewise.
1340 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
1341 (MULTILIB_DEFAULTS): Likewise.
1342 (DWARF2_UNWIND_INFO): Likewise.
1343 * config/arc/big.h: New file.
1344 * config/arc/elf.h: Likewise.
1345 * config/arc/linux.h: Likewise.
1346 * config/arc/t-uClibc: Remove.
1348 2017-02-27 Bin Cheng <bin.cheng@arm.com>
1350 PR tree-optimization/77536
1351 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
1352 (tree_transform_and_unroll_loop): Use above function to compute the
1353 estimated niter of unrolled loop and use it when scaling profile.
1354 Also use count info rather than frequency if it's non-zero.
1355 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
1356 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
1357 (vect_transform_loop): Call above function.
1359 2017-02-27 Richard Biener <rguenther@suse.de>
1361 PR tree-optimization/45397
1362 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
1363 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
1364 (visit_nary_op): Add pattern matching for CSEing sign-changed
1365 or truncated operations with wider ones.
1367 2017-02-27 Richard Biener <rguenther@suse.de>
1369 PR tree-optimization/79690
1370 * tree-vect-stmts.c (vectorizable_store): Use vector type
1371 built from the DR with address-space.
1373 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
1375 * doc/invoke.texi (Optimize Options): Refine the description
1376 of asan-use-after-return.
1378 2017-02-25 Alan Modra <amodra@gmail.com>
1380 PR rtl-optimization/79584
1381 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
1382 base, not ad->base_term, the reg within base. Remove assertion
1383 that ad->base == ad->base_term. Replace gen_int_mode using
1384 bogus mode with const0_rtx.
1386 2017-02-25 Jakub Jelinek <jakub@redhat.com>
1389 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
1390 FMA_EXPR like tcc_binary or tcc_unary.
1392 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
1395 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
1397 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
1398 (output_loc_operands): Handle DW_OP_call_ref and
1399 DW_OP_GNU_variable_value.
1400 (struct variable_value_struct): New type.
1401 (struct variable_value_hasher): Likewise.
1402 (variable_value_hash): New variable.
1403 (string_types): Remove.
1404 (copy_loc_descr): New function.
1405 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
1406 (prepend_loc_descr_to_each): New function.
1407 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
1408 instead of add_loc_descr_to_each if the first argument is single
1409 location list and the second has multiple.
1410 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
1411 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
1412 when looking for variable value which doesn't have other location info.
1413 (loc_list_from_tree): Formatting fix.
1414 (gen_array_type_die): Simplify DW_AT_string_length handling.
1415 (adjust_string_types): Remove.
1416 (gen_subprogram_die): Don't call adjust_string_types nor test/set
1417 string_types. Call resolve_variable_values.
1418 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
1419 (resolve_addr_in_expr): Likewise. Add A argument.
1420 (copy_deref_exprloc): Remove deref argument. Adjust for the
1421 original expression being DW_OP_GNU_variable_value with optionally
1422 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
1423 optionally after it.
1424 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
1425 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
1426 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
1427 (variable_value_hasher::hash, variable_value_hasher::equal): New
1429 (resolve_variable_value_in_expr, resolve_variable_value,
1430 resolve_variable_values, note_variable_value_in_expr,
1431 note_variable_value): New functions.
1432 (dwarf2out_early_finish): Call note_variable_value on all toplevel
1435 2017-02-24 Jakub Jelinek <jakub@redhat.com>
1438 * opts.h (handle_generated_option): Add GENERATED_P argument.
1439 * opts-common.c (handle_option): Adjust function comment.
1440 (handle_generated_option): Add GENERATED_P argument, pass it to
1442 (control_warning_option): Pass false to handle_generated_option
1444 * opts.c (maybe_default_option): Pass true to handle_generated_option
1446 * optc-gen.awk: Likewise.
1448 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1450 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
1451 a REG, look at the REG it is a SUBREG of.
1452 (splitter for cmpeqsi_t): Ditto.
1454 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1456 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
1457 the special USEs with the pattern of the insn, not the insn itself.
1459 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
1462 * doc/invoke.texi: Document -mload-store-pairs.
1464 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1465 Sandra Loosemore <sandra@codesourcery.com>
1467 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
1468 argument isn't a CONST_INT.
1469 (nios2_alternate_compare_const): Assert op is a CONST_INT.
1470 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
1471 (nios2_validate_compare): Bypass alternate compare logic if *op2
1473 (ldstwm_operation_p): Return false if first_base is not a REG or
1474 if first_offset is not a CONST_INT.
1476 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1478 * config/cris/cris.md: Use correct operand in a define_peephole2.
1480 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1482 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
1484 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
1486 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
1487 this_insn if it is an INSN or JUMP_INSN.
1488 (force_offsettable): Look at base, not at addr.
1489 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
1490 on things that aren't necessarily CONST_INTs.
1492 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
1494 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
1495 -mfpmath=sse is the default also for x86-32 targets with SSE2
1496 instruction set when @option{-ffast-math} is enabled
1498 2017-02-24 Jeff Law <law@redhat.com>
1500 PR rtl-optimizatoin/79286
1501 * ira.c (update_equiv_regs): Drop may_trap_p exception to
1504 2017-02-24 Richard Biener <rguenther@suse.de>
1506 PR tree-optimization/79389
1507 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
1510 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
1512 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
1513 function comment to reflect reality.
1514 (loop_exits_before_overflow): Fix typo in function description.
1516 2017-02-24 Richard Biener <rguenther@suse.de>
1518 PR tree-optimization/79389
1519 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
1520 properly that a threading opportunity exists. Detect conditional
1521 copy/constant propagation opportunities.
1523 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
1525 * config/visium/visium.md (type): Add trap.
1526 (b): New mode attribute.
1527 (*btst): Rename into...
1528 (*btst<mode>): ...this and adjust.
1529 (*cbranchsi4_btst_insn): Rename into...
1530 (*cbranch<mode>4_btst_insn): ...this and adjust.
1531 (trap): New define_insn.
1533 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1535 PR tree-optimization/79389
1536 * ifcvt.c (struct noce_if_info): Add rev_cond field.
1537 (noce_reversed_cond_code): New function.
1538 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
1539 reversed_comparison_code. Formatting fix.
1540 (noce_try_store_flag): Test rev_cond != NULL in addition to
1541 reversed_comparison_code.
1542 (noce_try_store_flag_constants): Likewise.
1543 (noce_try_store_flag_mask): Likewise.
1544 (noce_try_addcc): Use rev_cond if non-NULL instead of
1545 reversed_comparison_code.
1546 (noce_try_cmove_arith): Likewise. Formatting fixes.
1547 (noce_try_minmax, noce_try_abs): Clear rev_cond.
1548 (noce_find_if_block): Initialize rev_cond.
1549 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
1550 instead of false as last argument never attempt to reverse it
1553 2017-02-23 Bin Cheng <bin.cheng@arm.com>
1555 PR tree-optimization/79663
1556 * tree-predcom.c (combine_chains): Process refs in reverse order
1557 only for ZERO length chains, and add explaining comment.
1559 2017-02-23 Jeff Law <law@redhat.com>
1561 PR tree-optimization/79578
1562 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
1563 in call to operand_equal_p.
1565 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
1568 * config/i386/cpuid.h: Fix another undefined behavior.
1570 2017-02-23 Richard Biener <rguenther@suse.de>
1572 PR tree-optimization/79683
1573 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
1574 vector types for data-refs.
1576 2017-02-23 Martin Liska <mliska@suse.cz>
1578 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
1580 2017-02-23 Jakub Jelinek <jakub@redhat.com>
1583 * internal-fn.c (get_range_pos_neg): Moved to ...
1584 * tree.c (get_range_pos_neg): ... here. No longer static.
1585 * tree.h (get_range_pos_neg): New prototype.
1586 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
1587 are known to be in between 0 and signed maximum inclusive, try to
1588 expand both unsigned and signed divmod and use the cheaper one from
1591 2017-02-22 Jeff Law <law@redhat.com>
1593 PR tree-optimization/79578
1594 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
1595 to compare base operands.
1597 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
1600 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
1601 gpc_reg_operand instead of fpr_reg_operand.
1603 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
1605 * config/mips/mips.c (mips_return_in_memory): Force FP
1606 vector types to be returned in memory for o32 ABI.
1608 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1610 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
1611 instead of DW_TAG_member for static data member declarations and don't
1612 set no_linkage_name for static inline data members.
1613 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
1616 2017-02-22 Martin Liska <mliska@suse.cz>
1618 * doc/invoke.texi: Replace inequality signs with square brackets
1621 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1623 PR tree-optimization/68644
1624 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
1626 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
1629 * lra-constraints.c (simplify_operand_subreg): Handle
1630 WORD_REGISTER_OPERATIONS targets.
1632 2017-02-22 Jakub Jelinek <jakub@redhat.com>
1635 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
1636 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
1637 elimination by swapping fld*.
1639 2017-02-22 Richard Biener <rguenther@suse.de>
1641 PR tree-optimization/79673
1642 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
1643 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
1644 irrelevant address-space qualifiers and avoiding a
1645 ADDR_SPACE_CONVERT_EXPR from fold_convert.
1647 2017-02-22 Richard Biener <rguenther@suse.de>
1649 PR tree-optimization/79666
1650 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
1651 to not symbolically negate if that may introduce undefined
1654 2017-02-22 Martin Liska <mliska@suse.cz>
1657 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
1658 * data-streamer-out.c (streamer_write_gcov_count_stream):
1660 * value-prof.c (stream_out_histogram_value): Make assert more
1661 precise based on type of counter.
1663 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
1666 * config/i386/i386.md (standard_x87sse_constant_load splitter):
1667 Use nonimmediate_operand instead of memory_operand for operand 1.
1668 (float-extend standard_x87sse_constant_load splitter): Ditto.
1670 2017-02-21 Jeff Law <law@redhat.com>
1672 PR tree-optimization/79621
1673 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
1674 blocks with edges to themselves.
1676 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1679 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
1680 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
1681 Use gimple_call_builtin_p.
1684 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
1685 on temporarily removed DEBUG_INSNs.
1687 PR tree-optimization/79649
1688 * tree-loop-distribution.c (classify_partition): Give up on
1689 non-generic address space loads/stores.
1691 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
1693 * doc/loop.texi (Loop manipulation): Remove nonexistent
1694 tree_ssa_loop_version from the documentation.
1695 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
1697 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1700 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
1701 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
1702 * config/rs6000/rs6000.c: Include except.h.
1703 (rs6000_expand_split_stack_prologue): Call
1704 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
1706 2017-02-21 Martin Jambor <mjambor@suse.cz>
1709 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
1712 2017-02-21 Martin Jambor <mjambor@suse.cz>
1714 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
1715 for backward compatibility only.
1716 * doc/invoke.texi (Option Summary): Remove all references to
1719 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
1723 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1725 * lra-constraints.c (curr_insn_transform): Handle
1726 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1728 2017-02-21 Martin Liska <mliska@suse.cz>
1730 * config/i386/i386.opt: Replace -masm-dialect with -masm.
1732 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
1734 PR translation/79638
1735 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
1737 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
1740 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
1741 (arm_function_ok_for_sibcall): Return false for an indirect call by
1742 descriptor if all the argument registers are used.
1743 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
1744 alignment of the function.
1746 2017-02-21 Jakub Jelinek <jakub@redhat.com>
1748 PR tree-optimization/61441
1749 * simplify-rtx.c (simplify_const_unary_operation): For
1750 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
1751 the sNaN unmodified.
1753 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1755 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
1756 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
1757 instead of SYSTEM_HEADER_DIR.
1759 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
1760 Martin Liška <mliska@suse.cz>
1762 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
1763 Fix typos and grammar, use active voice, and clarify.
1765 2017-02-20 Marek Polacek <polacek@redhat.com>
1768 * gimplify.c (gimplify_expr): Handle unused *&&L;.
1771 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
1773 2017-02-20 Jakub Jelinek <jakub@redhat.com>
1776 * config/i386/i386.c (ix86_expand_builtin): Handle
1777 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
1778 ix86_builtins_isa[fcode].isa as a requirement of those
1779 flags and any other flag in the bitmask.
1780 (ix86_init_mmx_sse_builtins): Use 0 instead of
1781 ~OPTION_MASK_ISA_64BIT as mask.
1782 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
1783 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
1784 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
1785 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
1787 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1790 * lra-constraints.c (split_reg): Check requested split mode
1791 is supported by the register.
1793 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1795 * lra-constraints.c (simplify_operand_subreg): Remove early
1798 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1801 * lra-constraints.c (curr_insn_transform): Tighten condition
1802 for converting SUBREG reloads from OP_OUT to OP_INOUT.
1804 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
1807 * lra-constraints.c (curr_insn_transform): Handle
1808 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
1810 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
1813 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
1815 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
1817 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
1820 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
1823 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1825 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
1828 2017-02-18 Jakub Jelinek <jakub@redhat.com>
1830 * final.c (last_columnnum, override_columnnum): New variables.
1831 (final_start_function): Set last_columnnum, pass it to begin_prologue
1832 hook and pass 0 to dwarf2out_begin_prologue.
1833 (final_scan_insn): Update override_columnnum. Pass last_columnnum
1834 to source_line debug hook.
1835 (notice_source_line): Compute last_columnnum and for debug_column_info
1836 return true on column changes.
1837 * debug.h (struct gcc_debug_hooks): Add column argument to
1838 source_line and begin_prologue hooks.
1839 (debug_nothing_int_charstar_int_bool): Remove prototype.
1840 (debug_nothing_int_int_charstar,
1841 debug_nothing_int_int_charstar_int_bool): New prototypes.
1842 (dwarf2out_begin_prologue): Add column argument.
1843 * debug.c (do_nothing_debug_hooks): Adjust source_line and
1844 begin_prologue hooks.
1845 (debug_nothing_int_charstar_int_bool): Remove.
1846 (debug_nothing_int_int_charstar,
1847 debug_nothing_int_int_charstar_int_bool): New functions.
1848 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
1849 through to dwarf2out_source_line.
1850 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
1851 (dwarf2out_source_line): Add column argument, emit it if requested.
1852 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
1854 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1855 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
1856 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
1857 through to dwarf2out_begin_prologue.
1858 (vmsdbgout_source_line): Add column argument, pass it through to
1859 dwarf2out_source_line.
1860 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
1861 dbxout_source_line caller.
1862 (dbxout_source_line): Add column argument.
1864 * common.opt (gno-column-info, gcolumn-info): New options.
1865 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
1866 (check_die): Also test for multiple DW_AT_decl_column attributes.
1867 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
1868 DW_AT_decl_column if requested.
1869 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
1871 (gen_variable_die): Likewise.
1872 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
1873 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
1876 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
1877 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
1878 (ix86_handle_option): Handle OPT_m3dnowa.
1879 * doc/invoke.texi (-m3dnowa): Document.
1880 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
1881 -m3dnowa instead of -m3dnow -march=athlon.
1884 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
1885 instead of gcc_assert for K, r and R code checks. Formatting fixes.
1887 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1890 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
1891 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
1892 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
1893 generator for vsx_xxpermdi_<mode>_be.
1894 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
1895 force big-endian semantics.
1896 (vsx_xxpermdi_<mode>_be): New define_expand with same
1897 implementation as previous version of vsx_xxpermdi_<mode>.
1899 2017-02-17 Jakub Jelinek <jakub@redhat.com>
1901 PR tree-optimization/79327
1902 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
1903 variable, its initialization and use.
1905 2017-02-17 Julia Koval <julia.koval@intel.com>
1907 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
1908 (OPTION_MASK_ISA_PKU_UNSET): New.
1909 (ix86_handle_option): Handle -mrdpid.
1910 * config/i386/cpuid.h (bit_RDPID): New.
1911 * config/i386/driver-i386.c (host_detect_local_cpu):
1912 Detect RDPID feature.
1913 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
1914 * config/i386/i386-c.c (ix86_target_macros_internal):
1916 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
1917 (ix86_valid_target_attribute_inner_p): Add "rdpid".
1918 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
1919 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
1920 * config/i386/i386.md (define_insn "rdpid"): New.
1921 * config/i386/i386.opt Add -mrdpid.
1922 * config/i386/immintrin.h (_rdpid_u32): New.
1924 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
1926 PR rtl-optimization/79541
1927 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
1928 instead of transforming it into USE.
1930 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
1932 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
1933 If HONOR_SNANS (SFmode) force the input to a register.
1934 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
1935 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
1936 an frsp or similar insn.
1938 2017-02-17 Martin Liska <mliska@suse.cz>
1940 PR rtl-optimization/79577
1941 * params.def (selsched-max-sched-times): Increase minimum to 1.
1943 2017-02-17 Martin Liska <mliska@suse.cz>
1945 PR rtl-optimization/79574
1946 * gcse.c (want_to_gcse_p): Prevent integer overflow.
1948 2017-02-17 Martin Liska <mliska@suse.cz>
1950 PR tree-optimization/79529
1951 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
1952 ssa_defined_default_def_p to handle cases which are implicitly
1954 * tree-ssa.c (ssa_defined_default_def_p): New function.
1955 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
1956 which are implicitly defined.
1957 * tree-ssa.h (ssa_defined_default_def_p): Declare.
1959 2017-02-17 Richard Biener <rguenther@suse.de>
1962 * params.def (max-ssa-name-query-depth): Limit to 10.
1964 2017-02-17 Richard Biener <rguenther@suse.de>
1966 PR tree-optimization/79552
1967 * tree-ssa-structalias.c (visit_loadstore): Properly verify
1970 2017-02-17 Richard Biener <rguenther@suse.de>
1973 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
1975 2017-02-17 Marek Polacek <polacek@redhat.com>
1978 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
1979 (fold_negate_expr): New wrapper.
1981 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
1983 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
1984 Correct terminology and de-emphasize pre-standard behavior.
1986 2017-02-16 Alan Modra <amodra@gmail.com>
1988 PR rtl-optimization/79286
1989 * ira.c (def_dominates_uses): New function.
1990 (update_equiv_regs): Don't create an equivalence for insns that
1991 may trap where the register def does not dominate the use.
1993 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
1995 PR rtl-optimization/78127
1996 * lra.c (lra): Call lra_eliminate before finish the loop after
1999 2017-02-16 Richard Biener <rguenther@suse.de>
2001 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
2003 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
2004 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
2005 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
2006 (isl_val_int_from_wi): New function.
2007 (extract_affine_gmp): Rename to ...
2008 (extract_affine_wi): ... this, take a widest_int.
2009 (extract_affine_int): Just wrap extract_affine_wi.
2010 (add_param_constraints): Use isl_val_int_from_wi.
2011 (add_loop_constraints): Likewise, and extract_affine_wi.
2013 2017-02-15 Jeff Law <law@redhat.com>
2016 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
2017 ira_init_register_move_cost_if_necessary.
2019 2017-02-15 Martin Sebor <msebor@redhat.com>
2022 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
2023 removed in a prior commit.
2025 2017-02-15 Bin Cheng <bin.cheng@arm.com>
2027 PR tree-optimization/79347
2028 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
2029 counters during peeling.
2031 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
2033 * Makefile.in (site.exp): Remove "set ISLVER".
2035 2017-02-15 Jakub Jelinek <jakub@redhat.com>
2038 * real.c (real_from_integer): Call real_convert even for decimal.
2040 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
2043 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
2045 2017-02-14 Andrew Pinski <apinski@cavium.com>
2047 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
2048 cores and change the partno/implementer to be correct.
2049 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
2050 the 'B" as the implementer.
2051 * config/aarch64/aarch64-tune.md: Regenerate.
2053 2017-02-14 Carl Love <cel@us.ibm.com>
2055 * config/rs6000/rs6000.c: Add case statement entry to make the
2056 xvcvuxdsp built-in argument unsigned.
2057 * config/rs6000/vsx.md: Fix the source and return operand types so they
2058 match the instruction definitions from the ISA document. Fix typo
2059 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
2062 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
2065 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
2066 member early_clobber_alts.
2067 * lra-lives.c (reg_early_clobber_p): New.
2068 (process_bb_lives): Use it.
2069 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
2070 (debug_operand_data): Initialize early_clobber_alts.
2071 (setup_operand_alternative): Set up early_clobber_alts.
2072 (collect_non_operand_hard_regs): Ditto. Pass early clobber
2073 alternatives to new_insn_reg.
2074 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
2076 (lra_update_insn_regno_info): Pass the new arg.
2078 2017-02-14 Jakub Jelinek <jakub@redhat.com>
2081 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
2082 (new_oacc_loop_raw): Don't clear already cleared fields.
2085 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
2086 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
2087 _mm512_prefetch_i64gather_ps): New inline functions and macros.
2089 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
2092 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
2094 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
2097 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
2098 the extra instruction to the right place to store 128-bit constant
2101 2017-02-14 Martin Sebor <msebor@redhat.com>
2104 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
2105 warning for strings of unknown length.
2107 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
2109 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
2111 2017-02-14 Jeff Law <law@redhat.com>
2114 * ira-costs.c (scan_one_insn): Initialize register move costs
2115 for pseudos seen in USE/CLOBBER insns.
2117 PR tree-optimization/79095
2118 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
2119 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
2120 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
2121 if the operands are known to be not equal, then the resulting range
2123 (intersect_ranges): If the new range is ~[0,0] and the old range is
2124 wide, then prefer ~[0,0].
2125 * tree-vrp.c (overflow_comparison_p_1): New function.
2126 (overflow_comparison_p): New function.
2127 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
2128 if NAME is used in an overflow test.
2129 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
2130 overflow check that can be expressed as an equality test, then adjust
2131 ops to be that equality test.
2133 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2135 * config/s390/s390-builtin-types.def: Remove flags argument.
2136 * config/s390/s390.c (s390_init_builtins): Likewise.
2138 2017-02-14 Martin Liska <mliska@suse.cz>
2140 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
2141 vector. Fix trailing white spaces.
2143 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
2145 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
2148 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2150 PR rtl-optimization/68664
2151 * config/arm/arm.c (arm_sched_can_speculate_insn):
2152 New function. Declare prototype.
2153 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
2155 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2157 PR rtl-optimization/68664
2158 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
2160 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
2162 2017-02-14 Amit Pawar <amit.pawar@amd.com>
2164 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
2165 max skip bytes for function, loop and jump.
2167 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2169 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
2170 ABS_EXPR for gimple dump.
2172 2017-02-14 Jakub Jelinek <jakub@redhat.com>
2175 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
2177 PR tree-optimization/79408
2178 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
2179 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
2180 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
2181 also if rhs1 is INTEGER_CST.
2183 2017-02-14 Richard Biener <rguenther@suse.de>
2186 * tree-into-ssa.c (insert_phi_nodes): When the function can
2187 have abnormal edges rewrite SSA names with broken use-def
2188 dominance out of SSA and register them for PHI insertion.
2190 2017-02-13 Martin Sebor <msebor@redhat.com>
2193 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
2194 clearing info.nowrite flag when snprintf size argument is a range.
2196 2017-02-13 Jakub Jelinek <jakub@redhat.com>
2198 * cprop.c (cprop_jump): Add missing space in string literal.
2199 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
2200 (get_constraint_for_component_ref): Likewise.
2201 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
2202 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
2203 * lra-constraints.c (process_alt_operands): Likewise.
2204 * ipa-inline.c (inline_small_functions): Likewise.
2205 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
2206 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
2207 * trans-mem.c (diagnose_tm_1_op): Likewise.
2208 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
2209 (grid_parallel_clauses_gridifiable): Likewise.
2211 * config/nvptx/mkoffload.c (process): Add space in between
2214 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
2215 "MOD4_SSE_REGS" and "ALL_REGS".
2217 * spellcheck.c (test_data): Add , in between "foo" and "food".
2219 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2222 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
2223 boundary crossing check and subsequent code generation agree.
2225 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2227 * config/aarch64/aarch64.c (has_memory_op): Delete.
2228 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
2231 2017-02-13 Jakub Jelinek <jakub@redhat.com>
2233 PR rtl-optimization/79388
2234 PR rtl-optimization/79450
2235 * combine.c (distribute_notes): When removing TEM_INSN for which
2236 corresponding dest has last value recorded, invalidate that last
2239 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2241 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
2242 of explicit '@'. Add missing assembly comment marker on branch costs
2245 2017-02-13 Nathan Sidwell <nathan@acm.org>
2247 * gengtype-lex.l (<in_struct>): Add '/'.
2249 2017-02-13 Martin Liska <mliska@suse.cz>
2252 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
2254 2017-02-13 Richard Biener <rguenther@suse.de>
2256 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
2258 * configure: Re-generate.
2259 * config.in: Likewise.
2260 * graphite-dependences.c: Simplify as if
2261 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
2262 * graphite-isl-ast-to-gimple.c: Likewise.
2263 * graphite-optimize-isl.c: Likewise.
2264 * graphite-poly.c: Likewise.
2265 * graphite-sese-to-poly.c: Likewise.
2266 * graphite.h: Likewise.
2267 * toplev.c: Include isl/version.h and use isl_version () for
2268 printing the ISL version.
2269 * doc/install.texi: Update ISL requirement.
2271 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2273 * doc/standards.texi (Standards): Update reference to
2276 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
2278 * doc/extend.texi (Named Address Spaces): sourceware.org now
2280 * doc/install.texi (Binaries): Ditto.
2283 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2285 * doc/cpp.texi: Replace "stringify"/"stringification" with C
2286 standard terminology "stringize"/"stringizing" throughout.
2287 * doc/cppinternals.texi: Likewise.
2289 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
2291 * doc/extend.texi: Fix some spelling mistakes and typos.
2292 * doc/invoke.texi: Likewise.
2294 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2297 * params.def (inline-min-speedup) Change from 10 to 8.
2299 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2301 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
2304 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
2307 * ipa-inline-analysis.c (get_minimal_bb): New function.
2308 (record_modified): Use it.
2309 (remap_edge_change_prob): Handle also ancestor functions.
2311 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
2313 * doc/contrib.texi (Contributors): Remove broken link into
2314 the Mauve CVS repository.
2316 2017-02-11 Jakub Jelinek <jakub@redhat.com>
2319 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
2320 result computation whenever lhs doesn't have vector mode, not
2321 just when it has BLKmode.
2323 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2325 * doc/makefile.texi (profiledbootstrap): Refer to the
2326 installation instructions only in textual form.
2328 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2331 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
2333 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
2335 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
2336 (Specific): Update mingw-w64 reference.
2338 (Specific): Remove broken link to Renesas RX processor.
2340 2017-02-10 Richard Biener <rguenther@suse.de>
2342 * toplev.c (process_options): Do not mention obsolete graphite
2343 options when printing sorry message about missing graphite support.
2344 Mention -floop-nest-optimize.
2346 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
2348 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
2349 (vtst_p16): Likewise.
2350 (vtstq_p8): Likewise.
2351 (vtstq_p16): Likewise.
2353 (vtstq_p64): Likewise.
2354 * config/arm/arm_neon.h (vgetq_lane_p64): New.
2355 (vset_lane_p64): New.
2356 (vsetq_lane_p64): New.
2358 2017-02-10 Jakub Jelinek <jakub@redhat.com>
2360 PR tree-optimization/79411
2361 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
2362 stmt operands are SSA_NAMEs used in abnormal phis.
2363 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
2366 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2369 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
2372 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
2374 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
2376 2017-02-09 Jakub Jelinek <jakub@redhat.com>
2378 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
2382 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
2383 not arbitrary TREE_CONSTANT.
2386 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
2387 "omp declare target link" attribute unless is_global_var.
2388 * omp-offload.c (find_link_var_op): Likewise.
2390 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
2391 Chung-Lin Tang <cltang@codesourcery.com>
2393 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
2395 (gimplify_adjust_omp_clauses): Don't delete TILE.
2396 (gimplify_omp_for): Deal with TILE.
2397 * internal-fn.c (expand_GOACC_TILE): New function.
2398 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
2400 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
2401 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
2403 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
2404 avoid DIV for outermost collapse var.
2405 (expand_oacc_for): Insert tile element loop as needed. Adjust.
2406 Remove out of date comments, fix whitespace.
2407 * omp-general.c (omp_extract_for_data): Deal with tiling.
2408 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
2409 adjust OLF_DIM_BASE value.
2410 (struct omp_for_data): Add tiling field.
2411 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
2412 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
2413 for auto loops. Remove default auto determining, moved to
2414 oacc_loop_fixed_partitions.
2415 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
2416 stmts, add e_mask field.
2417 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
2418 (oacc_thread_numbers): Use oacc_dim_call.
2419 (oacc_xform_tile): New.
2420 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
2421 (finish_oacc_loop): Adjust for ifns vector.
2422 (oacc_loop_discover_walk): Append loop abstraction sites to list,
2423 add case for GOACC_TILE fns.
2424 (oacc_loop_xform_loop): Delete.
2425 (oacc_loop_process): Iterate over call list directly, and add
2426 handling for GOACC_TILE fns.
2427 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
2429 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
2430 vector partitioning to outer loops. Assign 2 partitions to loops
2431 when available. Add TILE handling.
2432 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
2433 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
2434 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
2435 * tree.c (omp_clause_num_ops): Adjust TILE ops.
2436 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
2438 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
2440 * configure.ac (ACX_BUGURL): Update.
2441 * configure: Regenerate.
2443 2017-02-09 Richard Biener <rguenther@suse.de>
2445 PR tree-optimization/69823
2446 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
2447 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
2449 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2451 * config/arc/arc-c.def: Add __NPS400__ definition.
2452 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
2453 (TARGET_NPS400): Define.
2455 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
2457 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
2459 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
2461 (arc_cpu_types): Fill the arch_info field with a pointer into the
2462 arc_arch_types table.
2463 (arc_selected_cpu): Declare.
2464 * config/arc/arc.c (arc_selected_cpu): Make global.
2465 (arc_selected_arch): Delete.
2466 (arc_base_cpu): Delete.
2467 (arc_override_options): Remove references to deleted variables,
2468 update access to arch information.
2469 (ARC_OPT): Update access to arch information.
2470 (ARC_OPTX): Likewise.
2471 * config/arc/arc.h (arc_base_cpu): Remove declaration.
2472 (TARGET_ARC600): Update access to arch information.
2473 (TARGET_ARC601): Likewise.
2474 (TARGET_ARC700): Likewise.
2475 (TARGET_EM): Likewise.
2476 (TARGET_HS): Likewise.
2477 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
2480 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
2483 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
2484 condition/operands for integer GE/LE/GEU/LEU operations.
2486 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
2488 PR translation/79397
2489 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
2492 2017-02-08 Martin Jambor <mjambor@suse.cz>
2495 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
2496 whether allocation happened.
2497 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
2498 nothing was allocated.
2500 2017-02-08 Jakub Jelinek <jakub@redhat.com>
2502 PR tree-optimization/79408
2503 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
2504 constant, but SSA_NAME with a known integer range, use the minimum
2505 of that range instead of op1 to determine if modulo can be replaced
2506 with its first operand.
2508 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2510 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
2512 2017-02-08 Richard Biener <rguenther@suse.de>
2514 PR tree-optimization/71824
2515 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
2516 Check all loops contained in the merged region.
2518 2017-02-07 Andrew Pinski <apinski@cavium.com>
2520 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
2522 2017-02-07 Andrew Pinski <apinski@cavium.com>
2524 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
2525 (thunderxt88): Likewise.
2526 (thunderxt81): Disable LSE and change v8.1 to v8.
2527 (thunderxt83): Likewise.
2529 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2530 Richard Biener <rguenther@suse.de>
2533 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
2534 type from int to size_t.
2535 * ira-costs.c (struct_costs_size): Change type from int to size_t.
2537 2017-02-07 Jakub Jelinek <jakub@redhat.com>
2539 PR rtl-optimization/79386
2540 * cprop.c (bypass_conditional_jumps): Initialize
2541 bypass_last_basic_block already before splitting bbs after
2542 unconditional traps...
2543 (bypass_conditional_jumps): ... rather than here.
2546 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
2547 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
2548 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
2549 fix -masm=intel patterns.
2551 2017-02-07 Richard Biener <rguenther@suse.de>
2553 PR tree-optimization/79256
2555 * builtins.c (get_object_alignment_2): Use min_align_of_type
2556 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
2557 and ADJUST_FIELD_ALIGN.
2559 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
2561 * doc/tm.texi: Regenerate.
2562 * stor-layout.c (layout_decl): Adjust.
2563 (update_alignment_for_field): Likewise.
2564 (place_field): Likewise.
2565 (min_align_of_type): Likewise.
2566 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
2567 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
2568 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
2569 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
2570 * config/frv/frv.c (frv_adjust_field_align): Likewise.
2571 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
2572 * config/i386/i386.c (x86_field_alignment): Likewise.
2573 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
2574 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
2575 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
2576 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
2577 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
2578 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
2582 2017-01-30 Richard Biener <rguenther@suse.de>
2584 PR tree-optimization/79256
2585 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
2586 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
2589 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
2591 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
2592 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
2593 builtins to SImode and emit a zero-extend, if necessary.
2595 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2597 * docs/invoke.texi (RISC-V Options): Alphabetize.
2599 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2601 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
2604 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
2606 * config/riscv/riscv.c: New file.
2607 * gcc/common/config/riscv/riscv-common.c: Likewise.
2608 * config.gcc: Likewise.
2609 * config/riscv/constraints.md: Likewise.
2610 * config/riscv/elf.h: Likewise.
2611 * config/riscv/generic.md: Likewise.
2612 * config/riscv/linux.h: Likewise.
2613 * config/riscv/multilib-generator: Likewise.
2614 * config/riscv/peephole.md: Likewise.
2615 * config/riscv/pic.md: Likewise.
2616 * config/riscv/predicates.md: Likewise.
2617 * config/riscv/riscv-builtins.c: Likewise.
2618 * config/riscv/riscv-c.c: Likewise.
2619 * config/riscv/riscv-ftypes.def: Likewise.
2620 * config/riscv/riscv-modes.def: Likewise.
2621 * config/riscv/riscv-opts.h: Likewise.
2622 * config/riscv/riscv-protos.h: Likewise.
2623 * config/riscv/riscv.h: Likewise.
2624 * config/riscv/riscv.md: Likewise.
2625 * config/riscv/riscv.opt: Likewise.
2626 * config/riscv/sync.md: Likewise.
2627 * config/riscv/t-elf-multilib: Likewise.
2628 * config/riscv/t-linux: Likewise.
2629 * config/riscv/t-linux-multilib: Likewise.
2630 * config/riscv/t-riscv: Likewise.
2631 * configure.ac: Likewise.
2632 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
2633 Waterman as RISC-V maintainers.
2634 * doc/install.texi: Add RISC-V entries.
2635 * doc/invoke.texi: Add RISC-V options section.
2636 * doc/md.texi: Add RISC-V constraints section.
2637 * configure: Regenerated.
2639 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
2642 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
2643 false values to be constant vectors with all 0 or all 1 bits set.
2644 (vcondu<mode><mode>): Likewise.
2645 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
2647 (fpmask_comparison_operator): Update comment.
2648 (vecint_comparison_operator): New predicate.
2649 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
2650 vector conditionals when the true and false values are constant
2651 vectors with all 0 bits or all 1 bits set.
2653 2017-02-06 Martin Sebor <msebor@redhat.com>
2655 PR tree-optimization/79376
2656 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
2658 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
2660 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
2661 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
2662 to simplify split condition.
2664 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2666 * omp-expand.c (oxpand_omp_atomic_fetch_op,
2667 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
2670 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
2672 PR rtl-optimization/68664
2673 * target.def (can_speculate_insn): New hook.
2674 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
2675 * doc/tm.texi: Regenerate.
2676 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
2677 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
2678 (rs6000_sched_can_speculate_insn): New function.
2680 2017-02-06 Jakub Jelinek <jakub@redhat.com>
2682 PR tree-optimization/79284
2683 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
2684 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
2685 vectorizable_mask_load_store, vectorizable_operation,
2686 vect_is_simple_cond, get_same_sized_vectype): Use it instead
2687 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
2688 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
2689 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
2690 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
2691 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
2692 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
2693 is_gimple_assign (stmt). Replace another such test with
2694 is_gimple_assign (stmt).
2696 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2699 * config/avr/avr.c (rtl-iter.h): Include it.
2700 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
2701 (avr_legitimate_combined_insn): ...and implementation.
2703 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
2705 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
2706 * config/s390/s390.c (s390_const_operand_ok)
2707 (s390_canonicalize_comparison, s390_extract_part)
2708 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
2709 (s390_contiguous_bitmask_p, s390_rtx_costs)
2710 (legitimize_pic_address): Likewise.
2711 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
2712 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
2713 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
2714 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
2715 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
2717 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
2719 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
2720 REGNO($0) == REGNO($1).
2722 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2724 * config/s390/linux.h(SIZE_TYPE): Add comment.
2726 2017-02-06 Julian Brown <julian@codesourcery.com>
2727 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2728 Virendra Pathak <virendra.pathak@broadcom.com>
2730 * config/aarch64/aarch64-cores.def: Change the scheduler
2732 * config/aarch64/aarch64.md: Include thunderx2t99.md.
2733 * config/aarch64/thunderx2t99.md: New file.
2735 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2737 * doc/standards.texi (Go Language): Update link to language
2740 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2742 * tree-eh.c (lower_resx): Sanitize profile.
2743 (cleanup_empty_eh_move_lp): Likewise.
2745 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
2748 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
2750 * cfgloopmanip.h (loop_version): Update prototype.
2751 * modulo-sched.c (sms_schedule): Update call of loop_version.
2752 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
2753 * tree-parloops.c (gen_parallel_loop): Likewise.
2754 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
2755 * tree-ssa-loop-split.c (split_loop): Likewise.
2756 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
2757 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
2759 2017-02-05 Martin Liska <mliska@suse.cz>
2762 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
2764 (print_operand_address): Initialize a struct to zero.
2766 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2768 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
2769 garbage collector only in textual form.
2771 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
2773 * doc/extend.texi (x86 specific memory model extensions for
2774 transactional memory): Simplify a phrase.
2776 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
2779 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
2780 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
2781 (atomic_storedi_1): Likewise.
2783 2017-02-04 Jakub Jelinek <jakub@redhat.com>
2785 PR tree-optimization/79338
2786 * tree-parloops.c (gather_scalar_reductions): Don't call
2787 vect_analyze_loop_form for loop->inner before destroying loop's
2790 2017-02-03 Martin Sebor <msebor@redhat.com>
2792 PR tree-optimization/79327
2793 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
2794 when precision has resulted in leading zeros.
2795 (format_integer): Adjust the likely counter to assume an unknown
2796 argument that may be zero is non-zero.
2798 2017-02-03 Jason Merrill <jason@redhat.com>
2801 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
2802 avoid copying non-taken branch.
2804 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2806 PR tree-optimization/79340
2807 * tree-vect-loop.c (vectorizable_reduction): Release
2808 vec_defs elements after safe_splicing them into other vectors.
2811 PR tree-optimization/79327
2812 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
2813 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
2815 (format_integer): Use wide_int_to_tree instead of build_int_cst
2816 + to_?hwi. If argmin is NULL, just set argmin and argmax to
2817 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
2818 of shortest and longest sequence.
2820 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
2822 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
2823 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
2825 2017-02-03 Walter Lee <walt@tilera.com>
2828 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
2829 after initial stackframe link reg save.
2830 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
2832 2017-02-03 Jakub Jelinek <jakub@redhat.com>
2835 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
2836 wu for stxssp alternative.
2838 2017-02-03 Martin Sebor <msebor@redhat.com>
2840 PR tree-optimization/79352
2841 * gimple-fold.c (get_range_strlen): Add argument.
2842 (get_range_strlen): Change return type to bool.
2843 (get_maxval_strlen): Pass in a dummy argument.
2844 * gimple-fold.h (get_range_strlen): Change return type to bool.
2845 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
2846 * tree.h (array_at_struct_end_p): Add argument.
2847 * tree.c (array_at_struct_end_p): Handle it.
2849 2017-02-03 Martin Liska <mliska@suse.cz>
2852 * multiple_target.c (create_dispatcher_calls): Redirect edge
2853 from a caller of a dispatcher.
2854 (expand_target_clones): Make the clones local.
2855 (ipa_target_clone): Do both target clones and resolvers.
2856 (ipa_dispatcher_calls): Remove the pass.
2857 (pass_dispatcher_calls::gate): Likewise.
2858 (make_pass_dispatcher_calls): Likewise.
2859 * passes.def (pass_target_clone): Put as very first IPA early
2862 2017-02-03 Martin Liska <mliska@suse.cz>
2864 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
2865 in case of a function with ifunc attribute.
2867 2017-02-03 Martin Liska <mliska@suse.cz>
2869 * cgraph.c (cgraph_node::dump): Dump function version info.
2870 * symtab.c (symtab_node::dump_base): Add missing new line.
2872 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2874 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
2875 (ifcombine_ifandif): Use it.
2877 2017-02-03 Martin Liska <mliska@suse.cz>
2879 * doc/invoke.texi: Document default value for
2880 use-after-scope-direct-emission-threshold.
2882 2017-02-03 Martin Liska <mliska@suse.cz>
2884 PR tree-optimization/79339
2885 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
2886 (format_floating): Likewise.
2888 2017-02-03 Martin Liska <mliska@suse.cz>
2891 * ipa-prop.c (ipa_node_params_t::insert): Remove current
2893 (ipa_node_params_t::remove): Likewise.
2894 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
2895 initialization from removed ipa_node_params_t::insert.
2896 (ipa_node_params::~ipa_node_params): Move from removed
2897 ipa_node_params_t::release.
2898 * symbol-summary.h (symbol_summary::m_released): New member.
2899 Do not release a summary twice. Do not allow to call finalizer
2900 for types of a summary that live in GGC memory.
2902 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2904 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
2907 2017-02-02 Martin Sebor <msebor@redhat.com>
2910 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
2911 (format_string): Tighten up the range of output for non-constant
2912 strings and correct the expected range for wide non-constant strings.
2914 2017-02-02 Martin Sebor <msebor@redhat.com>
2916 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
2919 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
2921 (-fdump-tree-@var): Add to index and document how to come up
2922 with pass-specific option and dump file names.
2923 (-fdump-passes): Clarify where to look for output.
2925 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
2928 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
2929 statistics of the analyzed path; allow threading for speed when
2930 any of BBs along the path are optimized for speed.
2932 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
2935 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
2936 settings of the virtual registers.
2939 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
2941 * explow.c (get_dynamic_stack_size): Take known alignment of stack
2942 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
2945 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2947 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
2948 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
2950 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2952 * config/s390/s390.md: Add missing comments with the expanded
2954 * config/s390/vector.md: Likewise.
2955 * config/s390/vx-builtins.md: Likewise.
2957 2017-02-02 Jakub Jelinek <jakub@redhat.com>
2960 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
2961 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
2962 conditions on a single line.
2964 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2966 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
2967 __S390_VX__ to __VX__.
2969 2017-02-01 Andrew Pinski <apinski@cavium.com>
2971 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
2972 stmt_info to record_stmt_cost.
2973 (vect_get_known_peeling_cost): Pass stmt_info if known to
2975 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
2976 cpu_vector_cost field into
2977 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
2978 field into vec_int_stmt_cost and vec_fp_stmt_cost.
2979 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
2980 splitting of scalar_stmt_cost and vec_stmt_cost.
2981 (thunderx_vector_cost): Likewise.
2982 (cortexa57_vector_cost): LIkewise.
2983 (exynosm1_vector_cost): Likewise.
2984 (xgene1_vector_cost): Likewise.
2985 (thunderx2t99_vector_cost): Improve after the splitting of the two
2987 (aarch64_builtin_vectorization_cost): Update for the splitting of
2988 scalar_stmt_cost and vec_stmt_cost.
2990 2017-02-01 Torvald Riegel <triegel@redhat.com>
2991 Richard Henderson <rth@redhat.com>
2993 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
2994 conditional on existance of a fast atomic load.
2995 * optabs-query.c (can_atomic_load_p): New function.
2996 * optabs-query.h (can_atomic_load_p): Declare it.
2997 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
2998 no fast atomic load is available for the particular size of access.
2999 (expand_atomic_compare_and_swap): Likewise.
3000 (expand_atomic_load): Likewise.
3001 (expand_atomic_store): Likewise.
3002 (expand_atomic_fetch_op): Likewise.
3003 * testsuite/lib/target-supports.exp
3004 (check_effective_target_sync_int_128): Remove x86 because it provides
3005 no fast atomic load.
3006 (check_effective_target_sync_int_128_runtime): Likewise.
3008 2017-02-01 Richard Biener <rguenther@suse.de>
3010 * graphite.c: Include tree-vectorizer.h for find_loop_location.
3011 (graphite_transform_loops): Provide opt-info for optimized nests.
3012 * tree-parloop.c (parallelize_loops): Provide opt-info for
3015 2017-02-01 Richard Biener <rguenther@suse.de>
3018 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
3021 2017-02-01 Richard Biener <rguenther@suse.de>
3023 PR tree-optimization/71824
3024 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
3025 Verify the loops are valid in the merged SESE region.
3026 (scop_detection::can_represent_loop_1): Check analyzing the
3027 evolution of the number of iterations in the region succeeds.
3029 2017-01-31 Ian Lance Taylor <iant@golang.org>
3031 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
3032 REG_ARGS_SIZE note to 32-bit push insns and call insn.
3034 2017-01-31 David Malcolm <dmalcolm@redhat.com>
3036 PR preprocessor/79210
3037 * input.c (get_substring_ranges_for_loc): Replace line_width
3038 assertion with error-handling.
3040 2017-01-31 Richard Biener <rguenther@suse.de>
3042 PR tree-optimization/77318
3043 * graphite-sese-to-poly.c (extract_affine): Fix assert.
3044 (create_pw_aff_from_tree): Take loop parameter.
3045 (add_condition_to_pbb): Pass loop of the condition to
3046 create_pw_aff_from_tree.
3048 2017-01-31 Jakub Jelinek <jakub@redhat.com>
3050 * config/s390/s390.c (s390_asan_shadow_offset): New function.
3051 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
3053 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
3057 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
3059 (convert_int_to_float128): Likewise.
3060 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
3061 (convert_int_to_float128): Likewise.
3062 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
3063 (UNSPEC_IEEE128_CONVERT): Likewise.
3064 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
3065 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
3066 Use local variables for IBM extended format.
3067 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
3068 (fix_trunc<mode>si2_fprs): Likewise.
3069 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
3070 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
3071 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
3072 to know that we can now have integers of all sizes in vector
3074 (fix<uns>_<mode>di2_hw): Likewise.
3075 (float<uns>_<mode>si2_hw): Likewise.
3076 (fix_<mode>si2_hw): Likewise.
3077 (fixuns_<mode>si2_hw): Likewise.
3078 (float<uns>_<mode>di2_hw): Likewise.
3079 (float_<mode>di2_hw): Likewise.
3080 (float_<mode>si2_hw): Likewise.
3081 (floatuns_<mode>di2_hw): Likewise.
3082 (floatuns_<mode>si2_hw): Likewise.
3083 (xscvqp<su>wz_<mode>): Delete, no longer used.
3084 (xscvqp<su>dz_<mode>): Likewise.
3085 (xscv<su>dqp_<mode>): Likewise.
3086 (ieee128_mfvsrd_64bit): Likewise.
3087 (ieee128_mfvsrd_32bit): Likewise.
3088 (ieee128_mfvsrwz): Likewise.
3089 (ieee128_mtvsrw): Likewise.
3090 (ieee128_mtvsrd_64bit): Likewise.
3091 (ieee128_mtvsrd_32bit): Likewise.
3093 2017-01-31 Martin Liska <mliska@suse.cz>
3096 * ipa-prop.c (ipa_free_all_node_params): Call release method
3097 instead of ~sumbol_summary to not to trigger double times
3100 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
3102 PR tree-optimization/71691
3103 * bitmap.h (class auto_bitmap): New.
3104 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
3105 is_maybe_undefined instead of ssa_undefined_value_p.
3107 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3109 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
3110 __S390_ARCH_LEVEL__ to __ARCH__.
3112 2017-01-31 Jakub Jelinek <jakub@redhat.com>
3114 PR tree-optimization/79267
3115 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
3116 if should_remove_lhs_p is true.
3118 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
3121 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
3122 (add_alignment_attribute): New.
3123 (base_type_die): Add alignment attribute.
3124 (subrange_type_die): Likewise.
3125 (modified_type_die): Likewise.
3126 (gen_array_type_die): Likewise.
3127 (gen_descr_array_type_die: Likewise.
3128 (gen_enumeration_type_die): Likewise.
3129 (gen_subprogram_die): Likewise.
3130 (gen_variable_die): Likewise.
3131 (gen_field_die): Likewise.
3132 (gen_ptr_to_mbr_type_die): Likewise.
3133 (gen_struct_or_union_type_die): Likewise.
3134 (gen_subroutine_type_die): Likewise.
3135 (gen_typedef_die): Likewise.
3136 (base_type_cmp): Compare alignment attribute.
3138 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3141 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
3142 (setb_unsigned) New pattern for setb with CCUNS.
3143 * config/rs6000/rs6000.c (expand_block_compare): Use a different
3144 subfc./subfe sequence to avoid overflow problems. Generate a
3145 shorter sequence with cmpld/setb for power9.
3146 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
3147 for generating subfc. instruction.
3148 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
3149 now uses this instruction.
3151 2017-01-30 Ian Lance Taylor <iant@google.com>
3154 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
3155 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
3157 2017-01-30 Martin Sebor <msebor@redhat.com>
3159 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
3160 Move constant to the right of a relational operator.
3161 (get_mpfr_format_length, format_character, format_string): Ditto.
3162 (should_warn_p, maybe_warn): Same.
3164 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
3166 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3169 * asan.c (get_translation_unit_decl): Remove function.
3170 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
3172 2017-01-30 Martin Liska <mliska@suse.cz>
3174 PR gcov-profile/79259
3175 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
3178 2017-01-30 Martin Liska <mliska@suse.cz>
3181 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
3182 Initialize variables with NULL value.
3184 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
3187 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
3189 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
3191 2017-01-30 Richard Biener <rguenther@suse.de>
3193 PR tree-optimization/79276
3194 * tree-vrp.c (process_assert_insertions): Properly adjust common
3195 when removing a duplicate.
3197 * gcc.dg/torture/pr79276.c: New testcase.
3199 2017-01-30 Richard Biener <rguenther@suse.de>
3201 PR tree-optimization/79256
3202 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
3203 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
3205 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
3207 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
3210 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
3211 ("*r<noxa>sbg_<mode>_sll_bitmask")
3212 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
3213 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
3214 Use contiguous_bitmask_nowrap_operand.
3216 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3219 * config/rs6000/altivec.h (vec_xl): Revise #define.
3220 (vec_xst): Likewise.
3222 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
3224 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
3226 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
3228 PR rtl-optimization/79194
3229 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
3230 traps before call to bypass_conditional_jumps.
3232 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3234 PR tree-optimization/71374
3235 * lra-constraints.c (check_conflict_input_operands): New.
3236 (match_reload): Use it.
3238 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
3241 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
3242 account to calculate conflict_set.
3244 2017-01-27 Bin Cheng <bin.cheng@arm.com>
3246 PR rtl-optimization/78559
3247 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
3248 other_insn in combine.
3250 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
3252 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
3253 uint16_type_node for BT_UINT16.
3255 2017-01-27 David Malcolm <dmalcolm@redhat.com>
3257 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
3258 "RTL Tests" to menu.
3259 (GIMPLE Tests): New node.
3260 (RTL Tests): New node.
3262 2017-01-27 Richard Biener <rguenther@suse.de>
3264 PR tree-optimization/79245
3265 * tree-loop-distribution.c (distribute_loop): Apply cost
3266 modeling also to detected patterns.
3268 2017-01-27 Richard Biener <rguenther@suse.de>
3270 PR tree-optimization/71433
3271 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
3272 (compare_assert_loc): New function.
3273 (process_assert_insertions): Sort and optimize assert locations
3274 to remove duplicates and push down identical assertions on
3275 edges to their destination block.
3277 2017-01-27 Richard Biener <rguenther@suse.de>
3279 PR tree-optimization/79244
3280 * tree-vrp.c (remove_range_assertions): Forcefully propagate
3281 out SSA names even if abnormal.
3283 2017-01-27 Jakub Jelinek <jakub@redhat.com>
3285 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
3286 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
3287 instead of MPFR_RNDN.
3289 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
3292 * arm.c (arm_option_override): Don't call build_target_option_node
3293 until after doing all option overrides.
3294 (arm_valid_target_attribute_tree): Likewise.
3296 2017-01-27 Martin Liska <mliska@suse.cz>
3298 * doc/invoke.texi (-fprofile-arcs): Document profiling support
3299 for {cd}tors and C++ {cd}tors.
3301 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
3303 * config/s390/s390.md ("*setmem_long_and")
3304 ("*setmem_long_and_31z"): Use zero_extend instead of and.
3306 2017-01-26 Martin Sebor <msebor@redhat.com>
3308 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
3311 2017-01-26 Martin Sebor <msebor@redhat.com>
3313 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
3314 HAVE_DFmode before using XFmode or DFmode.
3315 (parse_directive): Avoid using the z length modifier to avoid
3316 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
3319 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
3320 to accept adjustment as an array.
3321 (get_int_range): New function.
3322 (struct directive): Make width and prec arrays.
3323 (directive::set_width, directive::set_precision): Call get_int_range.
3324 (format_integer, format_floating): Handle width and precision ranges.
3325 (format_string, parse_directive): Same.
3327 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3330 * dwarf2out.c (generate_skeleton_bottom_up): For children with
3331 comdat_type_p set, just clone them, but keep the children in the
3335 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
3336 which have direct callers with -fvar-tracking-assignments enabled
3338 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
3339 inside of type units.
3341 2017-01-26 Martin Sebor <msebor@redhat.com>
3344 * gimple-ssa-sprintf.c (struct result_range): Add likely and
3346 (struct format_result): Replace number_chars, number_chars_min,
3347 and number_chars_max with a single member of struct result_range.
3349 (format_result::operator+=): Adjust.
3350 (struct fmtresult): Remove bounded. Handle likely and unlikely
3352 (fmtresult::adjust_for_width_or_precision): New function.
3353 (fmtresult:type_max_digits): New function.
3354 (bytes_remaining): Handle likely and unlikely counters.
3355 (min_bytes_remaining): Remove.
3356 (format_percent): Simplify.
3357 (format_integer, format_floating): Set likely and unlikely counters.
3358 (get_string_length, format_character, format_string): Same.
3359 (format_plain, should_warn_p): New function.
3360 (maybe_warn): Call should_warn_p. Update diagnostic messages
3361 and handle those for all directives, including plain strings.
3362 (format_directive): Handle likely and unlikely counters.
3363 Remove unnecessary quoting from diagnostics. Add an informational
3365 (add_bytes): Remove.
3366 (pass_sprintf_length::compute_format_length): Simplify.
3367 (try_substitute_return_value): Handle likely and unlikely counters.
3369 2017-01-26 Carl Love <cel@us.ibm.com>
3371 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
3372 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
3374 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
3377 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
3378 endianess for subregs into account.
3379 * lra-constraints.c (lra_constraints): Do risky transformations
3380 always on the first iteration.
3381 * lra-lives.c (check_pseudos_live_through_calls): Add arg
3382 last_call_used_reg_set.
3383 (process_bb_lives): Define and use last_call_used_reg_set.
3384 * lra.c (lra): Always continue after lra_constraints on the first
3387 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
3389 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
3391 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
3393 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3395 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
3396 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
3397 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
3398 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
3399 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
3400 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
3401 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
3402 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
3403 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
3405 2017-01-26 Marek Polacek <polacek@redhat.com>
3408 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
3409 for the third operand.
3411 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3414 * omp-low.c (struct omp_context): Add simt_stmt field.
3415 (scan_omp_for): Return omp_context *.
3416 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
3417 context to the _simt_ SIMD stmt.
3418 (lower_omp_for): For combined SIMD with sibling _simt_
3419 SIMD, make sure to use the same decls in _looptemp_
3420 clauses as in the sibling.
3422 2017-01-26 David Sherwood <david.sherwood@arm.com>
3425 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
3428 2017-01-26 Jakub Jelinek <jakub@redhat.com>
3431 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
3432 emit fld b; fld a; if possible.
3434 * brig-builtins.def: Update copyright years.
3435 * config/arm/arm_acle_builtins.def: Update copyright years.
3437 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3440 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
3441 constraint instead of o for the stxsd instruction.
3443 2017-01-25 Carl Love <cel@us.ibm.com>
3445 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
3446 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
3448 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
3450 * doc/invoke.texi (C++ Dialect Options): Fix typo.
3452 2017-01-25 Richard Biener <rguenther@suse.de>
3454 PR tree-optimization/69264
3455 * target.def (vector_alignment_reachable): Improve documentation.
3456 * doc/tm.texi: Regenerate.
3457 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
3459 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
3460 earlier changes with respect to TYPE_USER_ALIGN.
3461 (vector_alignment_reachable_p): Likewise. Improve dumping.
3463 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3466 * config/arm/arm.md (xordi3): Force constant operand into a register
3469 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3471 * doc/invoke.texi (-fstore-merging): Correct default optimization
3472 levels at which it is enabled.
3473 (-O): Move -fstore-merging from list to...
3476 2017-01-25 Richard Biener <rguenther@suse.de>
3479 * omp-expand.c: Include debug.h.
3480 (expand_omp_taskreg): Make sure to generate early debug before
3481 outlining anything from a function.
3482 (expand_omp_target): Likewise.
3483 (grid_expand_target_grid_body): Likewise.
3485 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
3488 * asan.c (get_translation_unit_decl): New function.
3489 (asan_add_global): Extract modules file name from globals
3490 TRANSLATION_UNIT_DECL name.
3492 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
3495 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
3496 for long calls with APCS frame and VFP.
3498 2017-01-24 David Malcolm <dmalcolm@redhat.com>
3500 * cfg.c (original_copy_tables_initialized_p): New function.
3501 * cfg.h (original_copy_tables_initialized_p): New decl.
3502 * cfgrtl.c (relink_block_chain): Guard the call to
3503 free_original_copy_tables with a call to
3504 original_copy_tables_initialized_p.
3505 * cgraph.h (symtab_node::native_rtl_p): New decl.
3506 * cgraphunit.c (symtab_node::native_rtl_p): New function.
3507 (symtab_node::needed_p): Don't assert for early assembly output
3508 for __RTL functions.
3509 (cgraph_node::finalize_function): Set "force_output" for __RTL
3511 (cgraph_node::analyze): Bail out early for __RTL functions.
3512 (analyze_functions): Update assertion to support __RTL functions.
3513 (cgraph_node::expand): Bail out early for __RTL functions.
3514 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
3516 * function.h (struct function): Update comment for field
3518 * gimple-expr.c: Include "tree-pass.h".
3519 (gimple_has_body_p): Return false for __RTL functions.
3520 * Makefile.in (OBJS): Add run-rtl-passes.o.
3521 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
3523 (gcc::pass_manager::get_clean_slate): New accessor.
3524 * passes.c: Include "insn-addr.h".
3525 (should_skip_pass_p): Add logging. Update logic for running
3526 "expand" to be compatible with both __GIMPLE and __RTL. Guard
3527 property-provider override so it is only done for gimple passes.
3529 (skip_pass): New function.
3530 (execute_one_pass): Call skip_pass when skipping passes.
3531 * read-md.c (md_reader::read_char): Support filtering
3532 the input to a subset of line numbers.
3533 (md_reader::md_reader): Initialize fields
3534 m_first_line and m_last_line.
3535 (md_reader::read_file_fragment): New function.
3536 * read-md.h (md_reader::read_file_fragment): New decl.
3537 (md_reader::m_first_line): New field.
3538 (md_reader::m_last_line): New field.
3539 * read-rtl-function.c (function_reader::create_function): Only
3540 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
3541 curr_properties. Set DECL_INITIAL to a dummy block.
3542 (read_rtl_function_body_from_file_range): New function.
3543 * read-rtl-function.h (read_rtl_function_body_from_file_range):
3545 * run-rtl-passes.c: New file.
3546 * run-rtl-passes.h: New file.
3548 2017-01-24 Jeff Law <law@redhat.com>
3550 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
3553 2017-01-24 Bin Cheng <bin.cheng@arm.com>
3555 PR tree-optimization/79159
3556 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
3557 (record_nonwrapping_iv): Improve boundary using above function if no
3558 value range information.
3560 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
3561 Martin Jambor <mjambor@suse.cz>
3563 * brig-builtins.def: New file.
3564 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
3565 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
3566 (DEF_HSAIL_SAT_BUILTIN): Likewise.
3567 (DEF_HSAIL_INTR_BUILTIN): Likewise.
3568 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
3569 * builtin-types.def (BT_INT8): New.
3570 (BT_INT16): Likewise.
3571 (BT_UINT8): Likewise.
3572 (BT_UINT16): Likewise.
3573 (BT_FN_ULONG): Likewise.
3574 (BT_FN_UINT_INT): Likewise.
3575 (BT_FN_UINT_ULONG): Likewise.
3576 (BT_FN_UINT_LONG): Likewise.
3577 (BT_FN_UINT_PTR): Likewise.
3578 (BT_FN_ULONG_PTR): Likewise.
3579 (BT_FN_INT8_FLOAT): Likewise.
3580 (BT_FN_INT16_FLOAT): Likewise.
3581 (BT_FN_UINT32_FLOAT): Likewise.
3582 (BT_FN_UINT16_FLOAT): Likewise.
3583 (BT_FN_UINT8_FLOAT): Likewise.
3584 (BT_FN_UINT64_FLOAT): Likewise.
3585 (BT_FN_UINT16_UINT32): Likewise.
3586 (BT_FN_UINT32_UINT16): Likewise.
3587 (BT_FN_UINT16_UINT16_UINT16): Likewise.
3588 (BT_FN_INT_PTR_INT): Likewise.
3589 (BT_FN_UINT_PTR_UINT): Likewise.
3590 (BT_FN_LONG_PTR_LONG): Likewise.
3591 (BT_FN_ULONG_PTR_ULONG): Likewise.
3592 (BT_FN_VOID_UINT64_UINT64): Likewise.
3593 (BT_FN_UINT8_UINT8_UINT8): Likewise.
3594 (BT_FN_INT8_INT8_INT8): Likewise.
3595 (BT_FN_INT16_INT16_INT16): Likewise.
3596 (BT_FN_INT_INT_INT): Likewise.
3597 (BT_FN_UINT_FLOAT_UINT): Likewise.
3598 (BT_FN_FLOAT_UINT_UINT): Likewise.
3599 (BT_FN_ULONG_UINT_UINT): Likewise.
3600 (BT_FN_ULONG_UINT_PTR): Likewise.
3601 (BT_FN_ULONG_ULONG_ULONG): Likewise.
3602 (BT_FN_UINT_UINT_UINT): Likewise.
3603 (BT_FN_VOID_UINT_PTR): Likewise.
3604 (BT_FN_UINT_UINT_PTR: Likewise.
3605 (BT_FN_UINT32_UINT64_PTR): Likewise.
3606 (BT_FN_INT_INT_UINT_UINT): Likewise.
3607 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
3608 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
3609 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
3610 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
3611 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
3612 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
3613 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
3614 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
3615 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
3616 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
3617 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
3618 * doc/frontends.texi: List BRIG FE.
3619 * doc/install.texi (Testing): Add BRIG tesring requirements.
3620 * doc/invoke.texi (Overall Options): Mention BRIG.
3621 * doc/standards.texi (Standards): Doucment BRIG HSA version.
3623 2017-01-24 Richard Biener <rguenther@suse.de>
3625 PR translation/79208
3626 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
3628 2017-01-24 Martin Jambor <mjambor@suse.cz>
3631 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
3632 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
3635 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
3638 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
3639 casts from signed to unsigned really don't have a range.
3641 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
3643 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
3644 GMP_RNDx for compatiblity.
3646 2017-01-24 Martin Liska <mliska@suse.cz>
3649 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
3650 that would prevent us to call alloca with -1 as argument.
3652 2017-01-24 Jakub Jelinek <jakub@redhat.com>
3654 * dwarf2out.c (output_compilation_unit_header, output_file_names):
3655 Avoid -Wformat-security warning.
3657 2017-01-23 Andrew Pinski <apinski@cavium.com>
3659 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
3662 2017-01-23 Martin Sebor <msebor@redhat.com>
3665 * gimple-ssa-sprintf.c (warn_level): New global.
3666 (format_integer): Use it here and throughout the rest of the file.
3667 Use the same switch to compute sign as base.
3668 (maybe_warn): New function.
3669 (format_directive): Factor out warnings into maybe_warn.
3670 Add debugging output. Use warn_level.
3671 (add_bytes): Use warn_level.
3672 (pass_sprintf_length::compute_format_length): Add debugging output.
3673 (try_substitute_return_value): Same.
3674 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
3677 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
3678 (struct fmtresult, format_integer, format_floating): Adjust.
3679 (fmtresult::fmtresult): Set max correctly in two argument ctor.
3680 (get_string_length, format_string,format_directive): Same.
3681 (pass_sprintf_length::compute_format_length): Same.
3682 (try_substitute_return_value): Simplify slightly.
3685 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
3686 (fmtresult::operator+=): Outlined.
3687 (struct fmtresult): Add ctors.
3688 (struct conversion_spec): Rename...
3689 (struct directive): ...to this. Add and remove data members.
3690 (directive::set_width, directive::set_precision): New functions.
3691 (format_percent): Use fmtresult ctor.
3692 (get_width_and_precision): Remove.
3693 (format_integer): Make naming changes. Avoid computing width and
3695 (format_floating): Same. Adjust indentation.
3696 (format_character, format_none): New functions.
3697 (format_string): Moved character handling to format_character.
3698 (format_directive): Remove arguments, change return type.
3699 (parse_directive): New function.
3700 (pass_sprintf_length::compute_format_length): Move directive
3701 parsing to parse_directive.
3703 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3705 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
3706 (assign_assembler_name_if_needed): ... this.
3707 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
3708 (assign_assembler_name_if_needed): ... this.
3709 (free_lang_data_in_cgraph): Adjust callers.
3710 * cgraphunit.c (cgraph_node::analyze): Likewise.
3711 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
3714 2017-01-23 Richard Biener <rguenther@suse.de>
3716 PR tree-optimization/79088
3717 PR tree-optimization/79188
3718 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
3719 resetting loop bounds after last path deletion. Reset loop
3720 bounds of the target loop, make code match the comments.
3721 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
3722 Make sure loops need no fixups.
3724 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
3726 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
3727 exponent support with double type for first argument.
3728 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
3729 type returned by __builtin_vec_extract_sig,
3730 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
3731 functions from "vector int" to "vector unsigned int" or from
3732 "vector long long int" to "vector unsigned long long int".
3733 Changed type returned by __builtin_vec_extract_exp,
3734 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
3735 functions from "vector int" to "vector unsigned int" or from
3736 "vector long long int" to "vector unsigned long long int".
3737 Changed return type of __builtin_vec_test_data_class,
3738 __builtin_vec_test_data_class_sp, and
3739 __builtin_vec_test_data_class_dp from "vector int" to
3740 "vector bool int" or from "vector long long int" to "vector bool
3741 long long int" and changed second argument type from "unsigned
3742 int" to "int". Added new overloaded function forms "vector float
3743 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
3744 "vector float __builtin_vec_insert_exp_sp (vector float, vector
3745 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
3746 double, vector unsigned long long int)" and "vector double
3747 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
3748 long int)". Changed return type of
3749 __builtin_scalar_test_data_class and
3750 __builtin_scalar_test_data_class_sp and
3751 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
3752 int" and changed second argument from "unsigned int" to "int".
3753 Changed type returned by __builtin_scalar_test_neg,
3754 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
3755 from "int" to "bool int". Added new overloaded function form
3756 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
3757 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
3758 exponent double-precision with floating point first argument.
3759 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
3760 documentation of scalar_test_data_class, scalar_test_neg,
3761 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
3762 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
3763 vec_test_data_class built-in functions to reflect refinements in
3764 their type signatures.
3766 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
3768 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
3770 (aarch64_elf_asm_destructor): Likewise.
3772 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
3774 PR rtl-optimization/78634
3775 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
3776 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
3777 * ifcvt.c (noce_try_cmove): Add missing cost check.
3779 PR rtl-optimization/71724
3780 * combine.c (if_then_else_cond): Look for situations where it is
3781 beneficial to undo the work of one of the recursive calls.
3783 2017-01-23 Bin Cheng <bin.cheng@arm.com>
3785 PR tree-optimization/70754
3786 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
3787 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
3788 combined stmt before it if not NULL.
3789 (combine_chains): Process refs reversely and compute dominance point
3792 2017-01-23 Martin Liska <mliska@suse.cz>
3794 PR tree-optimization/79196
3795 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
3796 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
3798 (strlen_optimize_stmt): Call the renamed function.
3800 2017-01-23 Michael Matz <matz@suse.de>
3802 PR tree-optimization/78384
3803 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
3805 2017-01-23 Richard Biener <rguenther@suse.de>
3807 PR tree-optimization/79186
3808 * tree-vrp.c (register_new_assert_for): Make sure we've seen
3809 both incoming edges before moving an assert.
3811 2017-01-23 Martin Jambor <mjambor@suse.cz>
3813 * ipa-prop.c (load_from_param_1): Removed.
3814 (load_from_unmodified_param): Bits from load_from_param_1 put back
3816 (load_from_param): Removed.
3817 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
3818 with stmt. Reverted back to use of load_from_unmodified_param.
3820 2017-01-23 Martin Jambor <mjambor@suse.cz>
3823 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
3824 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
3825 field a pointer to garbage collected vector, mark lattices and
3826 ipcp_orig_node with GTY((skip)).
3827 (ipa_get_param_count): Adjust to descriptors being a pointer.
3828 (ipa_get_param): Likewise.
3829 (ipa_get_type): Likewise.
3830 (ipa_get_param_move_cost): Likewise.
3831 (ipa_set_param_used): Likewise.
3832 (ipa_get_controlled_uses): Likewise.
3833 (ipa_set_controlled_uses): Likewise.
3834 (ipa_is_param_used): Likewise.
3835 (ipa_node_params_t): Move into garbage collector. New methods insert
3837 (ipa_node_params_sum): Annotate wth GTY(()).
3838 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
3840 (ipa_load_from_parm_agg): Adjust declaration.
3841 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
3842 * ipa-profile.c (ipa_profile): Likewise.
3843 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
3844 (ipa_populate_param_decls): Make descriptors parameter garbage
3846 (ipa_dump_param): Adjust to descriptors being a pointer.
3847 (ipa_alloc_node_params): Likewise.
3848 (ipa_initialize_node_params): Likewise.
3849 (load_from_param_1): Make descriptors parameter garbage collected.
3850 (load_from_unmodified_param): Likewise.
3851 (load_from_param): Likewise.
3852 (ipa_load_from_parm_agg): Likewise.
3853 (ipa_node_params::~ipa_node_params): Removed.
3854 (ipa_free_all_node_params): Remove call to delete operator.
3855 (ipa_node_params_t::insert): New.
3856 (ipa_node_params_t::remove): Likewise.
3857 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
3858 copy known_csts and known_contexts vectors.
3859 (ipa_read_node_info): Adjust to descriptors being a pointer.
3860 (ipcp_modif_dom_walker): Make m_descriptors field garbage
3862 (ipcp_transform_function): Make descriptors variable garbage
3865 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
3867 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
3868 * config/i386/avx512dqintrin.h: Ditto.
3869 * config/i386/avx512fintrin.h: Ditto.
3870 * gcc/config/i386/i386.c: Handle new builtins.
3871 * config/i386/i386-builtin.def: Add new builtins.
3872 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
3873 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
3875 2017-01-23 Jakub Jelinek <jakub@redhat.com>
3876 Martin Liska <mliska@suse.cz>
3878 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
3879 * asan.c (asan_expand_poison_ifn): Support stores and use
3880 appropriate ASAN report function.
3881 * internal-fn.c (expand_ASAN_POISON_USE): New function.
3882 * internal-fn.def (ASAN_POISON_USE): Declare.
3883 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
3884 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
3885 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
3886 ASAN_POISON calls w/o LHS.
3887 * tree-ssa.c (execute_update_addresses_taken): Create clobber
3888 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
3889 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
3890 * gimplify.c (asan_poison_variables): Add attribute
3891 use_after_scope_memory to variables that really needs to live
3893 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
3894 having the attribute.
3896 2017-01-23 Martin Liska <mliska@suse.cz>
3898 * asan.c (create_asan_shadow_var): New function.
3899 (asan_expand_poison_ifn): Likewise.
3900 * asan.h (asan_expand_poison_ifn): New declaration.
3901 * internal-fn.c (expand_ASAN_POISON): Likewise.
3902 * internal-fn.def (ASAN_POISON): New builtin.
3903 * sanopt.c (pass_sanopt::execute): Expand
3904 asan_expand_poison_ifn.
3905 * tree-inline.c (copy_decl_for_dup_finish): Make function
3907 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
3908 * tree-ssa.c (is_asan_mark_p): New function.
3909 (execute_update_addresses_taken): Rewrite local variables
3910 (identified just by use-after-scope as addressable) into SSA.
3912 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3914 * doc/install.texi (Specific): opensource.apple.com uses https
3915 now. Remove trailing slash.
3917 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
3919 * README.Portability: Remove note on an Irix compatibility issue.
3921 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
3923 * gcov.c (INCLUDE_ALGORITHM): Define.
3924 (INCLUDE_VECTOR): Define.
3925 No longer include <vector> and <algorithm> directly.
3927 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
3929 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
3931 * doc/invoke.texi (Code Gen Options): Ditto.
3933 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
3936 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
3938 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
3940 rtl-optimization/79125
3941 * cprop.c (local_cprop_pass): Handle cases where we make an
3944 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
3948 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
3949 read from, for big endian.
3951 2017-01-20 Jiong Wang <jiong.wang@arm.com>
3953 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
3954 register pauth builtins for LP64 only.
3956 2017-01-20 Marek Polacek <polacek@redhat.com>
3959 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
3962 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3964 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
3966 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
3967 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
3968 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
3970 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3973 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
3974 in big-endian mode when they are not a single duplicated value.
3976 2017-01-20 Richard Biener <rguenther@suse.de>
3978 * BASE-VER: Bump to 7.0.1.
3980 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
3982 * omp-low.c (omplow_simd_context): New struct. Use it...
3983 (lower_rec_simd_input_clauses): ...here and...
3984 (lower_rec_input_clauses): ...here to hold common data. Adjust all
3985 references to idx, lane, max_vf, is_simt.
3987 2017-01-20 Graham Markall <graham.markall@embecosm.com>
3989 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
3992 2017-01-20 Martin Jambor <mjambor@suse.cz>
3994 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
3995 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
3997 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
3998 (GTFILES): Rename hsa.c to hsa-common.c.
3999 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
4000 * hsa-dump.c: Likewise.
4001 * hsa-gen.c: Likewise.
4002 * hsa-regalloc.c: Likewise.
4003 * ipa-hsa.c: Likewise.
4004 * omp-expand.c: Likewise.
4005 * omp-low.c: Likewise.
4006 * toplev.c: Likewise.
4008 2017-01-20 Marek Polacek <polacek@redhat.com>
4011 * doc/invoke.texi: Document -Wduplicated-branches.
4012 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
4013 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
4014 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
4015 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
4016 return 0 only when not OEP_LEXICOGRAPHIC.
4017 (fold_build_cleanup_point_expr): Use the expression
4018 location when building CLEANUP_POINT_EXPR.
4019 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
4020 * tree.c (add_expr): Handle error_mark_node.
4022 2017-01-20 Martin Liska <mliska@suse.cz>
4025 * tree-profile.c (init_ic_make_global_vars): Do not call
4027 (gimple_init_gcov_profiler): Likewise.
4029 2017-01-20 Martin Liska <mliska@suse.cz>
4032 * cgraph.h (maybe_create_reference): Remove argument and
4034 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
4036 * ipa-cp.c (create_specialized_node): Likewise.
4037 * symtab.c (symtab_node::maybe_create_reference): Handle
4038 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
4040 2017-01-20 Martin Liska <mliska@suse.cz>
4042 * read-rtl-function.c (function_reader::create_function): Use
4043 build_decl instread of build_decl_stat.
4045 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
4047 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
4048 * config/i386/avx512dqintrin.h: Ditto.
4049 * config/i386/avx512fintrin.h: Ditto.
4050 * config/i386/i386-builtin-types.def: Add new types.
4051 * gcc/config/i386/i386.c: Handle new types.
4052 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
4053 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
4054 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
4055 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
4056 (__builtin_ia32_kshiftridi): New.
4057 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
4059 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
4063 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
4064 define to rs6000_init_stack_protect_guard.
4065 (rs6000_init_stack_protect_guard): New function.
4067 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
4068 Yunqiang Su <yunqiang.su@imgtec.com>
4070 * config.gcc (supported_defaults): Add madd4.
4071 (with_madd4): Add validation.
4072 (all_defaults): Add madd4.
4073 * config/mips/mips.opt (mmadd4): New option.
4074 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
4076 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
4078 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
4079 (ISA_HAS_FUSED_MADD4): Likewise.
4080 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
4081 * gcc/doc/install.texi (--with-madd4): Document the new option.
4083 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4085 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
4086 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
4087 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
4088 (aarch64_init_pauth_hint_builtins): New.
4089 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
4090 (aarch64_expand_builtin): Expand new builtins.
4092 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4094 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
4095 * combine-stack-adj.c (no_unhandled_cfa): Handle
4096 REG_CFA_TOGGLE_RA_MANGLE.
4097 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
4098 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
4099 info for return address signing.
4100 (aarch64_expand_epilogue): Likewise.
4102 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4104 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
4105 * config/aarch64/aarch64-protos.h
4106 (aarch64_return_address_signing_enabled): New declaration.
4107 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
4109 (aarch64_expand_prologue): Sign return address before it's pushed onto
4111 (aarch64_expand_epilogue): Authenticate return address fetched from
4113 (aarch64_override_options): Sanity check for ILP32 and ISA level.
4114 (aarch64_attributes): New function attributes for "sign-return-address".
4115 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
4116 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
4117 ("*do_return"): Generate combined instructions according to key index.
4118 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
4119 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
4121 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
4122 * config/aarch64/aarch64.opt (msign-return-address=): New.
4123 * doc/extend.texi (AArch64 Function Attributes): Documents
4124 "sign-return-address=".
4125 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
4127 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
4129 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
4130 overall option summary.
4132 2017-01-19 Jiong Wang <jiong.wang@arm.com>
4134 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
4135 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
4136 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
4137 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
4139 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
4141 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
4142 -mpower9-minmax by default for -mcpu=power9.
4143 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
4144 128-bit floating point.
4146 2017-01-20 Alan Modra <amodra@gmail.com>
4148 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
4149 optimizing for size.
4151 2017-01-20 Alan Modra <amodra@gmail.com>
4154 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
4155 for strcmp and strncmp from corresponding builtin decl.
4157 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
4159 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
4160 instead of i386/rtems-64.h.
4161 * config/i386/rtems-64.h: Remove.
4163 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
4167 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
4169 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
4171 2017-01-19 Tamar Christina <tamar.christina@arm.com>
4173 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
4174 Change int to HOST_WIDE_INT.
4175 * config/aarch64/aarch64-protos.h
4176 (aarch64_simd_gen_const_vector_dup): Likewise.
4177 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
4179 2017-01-19 David Malcolm <dmalcolm@redhat.com>
4181 * langhooks-def.h (lhd_type_for_size): New decl.
4182 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
4183 * langhooks.c (lhd_type_for_size): New function, taken from
4186 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
4188 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
4189 define_bypass for CR latency.
4190 (power9-cracked-alu): Update bypass latency and remove power9-branch.
4191 (power9-alu2): Add define_bypass for CR latency.
4193 (power9-mul): Update insn latency.
4194 (power9-mul-compare): Update insn latency, bypass latency and remove
4197 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4199 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
4201 * config/aarch64/aarch64.md
4202 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
4203 aarch64_nopcrelative_literal_loads.
4204 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
4206 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
4208 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
4210 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
4212 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
4215 * config.gcc (supported_defaults): Add lxc1-sxc1.
4216 (with_lxc1_sxc1): Add validation.
4217 (all_defaults): Add lxc1-sxc1.
4218 * config/mips/mips.opt (mlxc1-sxc1): New option.
4219 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
4221 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
4222 __mips_no_lxc1_sxc1.
4223 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
4224 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
4225 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
4227 2017-01-19 Richard Biener <rguenther@suse.de>
4229 PR tree-optimization/72488
4230 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
4231 sure to restore SSA info.
4232 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
4234 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
4236 PR rtl-optimization/79121
4237 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
4238 of the inner type when shifting an extended value.
4240 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4243 * symtab.c (symtab_node::equal_address_to): Fix comparing of
4244 interposable aliases.
4246 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
4249 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
4250 Use the evmergelohi instruction.
4251 (mov_si<mode>_e500_subreg4_2_le): Likewise.
4252 (mov_sitf_e500_subreg8_2_be): Likewise.
4253 (mov_sitf_e500_subreg12_2_le): Likewise.
4254 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
4255 (mov_si<mode>_e500_subreg4_2_be): Likewise.
4256 (mov_sitf_e500_subreg8_2_le): Likewise.
4257 (mov_sitf_e500_subreg12_2_be): Likewise.
4259 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4261 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
4262 attribute from vecsimple to vecperm.
4263 (altivec_vbpermq2): Likewise.
4265 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4268 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
4270 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4271 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
4272 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
4273 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
4274 case where N arg is SIZE_MAX.
4275 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
4276 (cmpstrsi): Add pattern.
4278 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
4280 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4281 __builtin_vec_revb builtins.
4282 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
4283 built-in functions to support generation of the ISA 3.0 XXBR<x>
4284 vector byte reverse instructions.
4285 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
4286 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
4287 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
4288 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
4289 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
4290 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
4291 (P9V_BUILTIN_VEC_REVB): Likewise.
4292 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
4293 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
4294 (p9_xxbrq_v16qi): Likewise.
4295 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
4296 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
4297 (p9_xxbrh_v8hi): Likewise.
4298 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
4299 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
4300 vec_revb built-in functions.
4302 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
4304 PR rtl-optimization/78952
4305 * config/i386/i386.md (any_extract): New code iterator.
4306 (*insvqi_2): Use any_extract for source operand.
4307 (*insvqi_3): Use any_shiftrt for source operand.
4309 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
4311 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
4313 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
4315 2017-01-18 Matthias Klose <doko@ubuntu.com>
4317 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
4319 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4321 * config/rs6000/altivec.h (vec_bperm): Change #define.
4322 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
4323 (altivec_vbpermq2): New define_insn.
4324 (altivec_vbpermd): Likewise.
4325 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
4327 (VBPERMD): Likewise.
4328 (VBPERM): New polymorphic function interface.
4329 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
4330 Add entries for P9V_BUILTIN_VEC_VBPERM.
4331 * doc/extend.texi: Add interfaces for vec_bperm.
4333 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4335 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
4336 first letter of error messages.
4337 (s390_resolve_overloaded_builtin): Likewise.
4338 * config/s390/s390.c (s390_expand_builtin): Likewise.
4339 (s390_invalid_arg_for_unprototyped_fn): Likewise.
4340 (s390_valid_target_attribute_inner_p): Likewise.
4341 * config/s390/s390.md ("tabort"): Likewise.
4343 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
4345 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
4346 (ISA_AVOID_DIV_HILO): New macro.
4347 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
4348 (ISA_HAS_DDIV): Likewise.
4350 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4352 * doc/invoke.texi (fabi-version): Correct number of occurrences.
4354 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4356 * doc/invoke.texi (fabi-version): Spelling fix.
4358 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4361 * doc/invoke.texi (fabi-version): Mention mangling fix for
4364 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
4367 * doc/invoke.texi (fabi-version): Document discriminator mangling.
4369 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
4372 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
4373 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
4375 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
4377 (stack_protect_test): Ditto.
4378 * config/rs6000/rs6000.opt (mstack-protector-guard=,
4379 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
4381 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
4382 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
4383 -mstack-protector-guard-offset=.
4384 (RS/6000 and PowerPC Options): Ditto.
4386 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4388 * config/i386/i386.h (MASK_CLASS_P): New define.
4389 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
4390 there are no registers from different register sets also when
4391 mask registers are used. Update function comment.
4392 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
4393 to (*k/*r) and (*k/*km) alternatives.
4395 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
4397 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
4398 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
4399 (EH_RETURN_HANDLER_RTX): New define.
4400 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
4401 Force frame pointer in EH return functions.
4402 (aarch64_expand_epilogue): Add barrier for eh_return.
4403 (aarch64_final_eh_return_addr): Remove.
4404 (aarch64_eh_return_handler_rtx): New function.
4405 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
4407 (aarch64_eh_return_handler_rtx): New prototype.
4409 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4411 * config/rs6000/altivec.h (vec_rlmi): New #define.
4412 (vec_vrlnm): Likewise.
4413 (vec_rlnm): Likewise.
4414 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
4415 (UNSPEC_VRLNM): Likewise.
4416 (VIlong): New mode iterator.
4417 (altivec_vrl<VI_char>mi): New define_insn.
4418 (altivec_vrl<VI_char>nm): Likewise.
4419 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
4422 (RLNM): New polymorphic function entry.
4423 (VRLWMI): New monomorphic function entry.
4425 (RLMI): New polymorphic function entry.
4426 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
4427 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
4428 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
4431 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4434 * dwarf2out.c (field_byte_offset): Restore the
4435 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
4436 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
4437 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
4440 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4443 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
4445 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4448 * dwarf2out.c (add_data_member_location_attribute): For constant
4449 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
4450 instead of DW_AT_data_member_location, DW_AT_bit_offset and
4451 DW_AT_byte_size attributes.
4453 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
4455 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
4456 after forcing to constant memory when the code model is medium.
4458 2017-01-17 Julia Koval <julia.koval@intel.com>
4461 * config/i386/avx512fintrin.h
4462 (_mm512_i32gather_ps): Change __addr type to void const*.
4463 (_mm512_mask_i32gather_ps): Ditto.
4464 (_mm512_i32gather_pd): Ditto.
4465 (_mm512_mask_i32gather_pd): Ditto.
4466 (_mm512_i64gather_ps): Ditto.
4467 (_mm512_mask_i64gather_ps): Ditto.
4468 (_mm512_i64gather_pd): Ditto.
4469 (_mm512_mask_i64gather_pd): Ditto.
4470 (_mm512_i32gather_epi32): Ditto.
4471 (_mm512_mask_i32gather_epi32): Ditto.
4472 (_mm512_i32gather_epi64): Ditto.
4473 (_mm512_mask_i32gather_epi64): Ditto.
4474 (_mm512_i64gather_epi32): Ditto.
4475 (_mm512_mask_i64gather_epi32): Ditto.
4476 (_mm512_i64gather_epi64): Ditto.
4477 (_mm512_mask_i64gather_epi64): Ditto.
4478 (_mm512_i32scatter_ps): Change __addr type to void*.
4479 (_mm512_mask_i32scatter_ps): Ditto.
4480 (_mm512_i32scatter_pd): Ditto.
4481 (_mm512_mask_i32scatter_pd): Ditto.
4482 (_mm512_i64scatter_ps): Ditto.
4483 (_mm512_mask_i64scatter_ps): Ditto.
4484 (_mm512_i64scatter_pd): Ditto.
4485 (_mm512_mask_i64scatter_pd): Ditto.
4486 (_mm512_i32scatter_epi32): Ditto.
4487 (_mm512_mask_i32scatter_epi32): Ditto.
4488 (_mm512_i32scatter_epi64): Ditto.
4489 (_mm512_mask_i32scatter_epi64): Ditto.
4490 (_mm512_i64scatter_epi32): Ditto.
4491 (_mm512_mask_i64scatter_epi32): Ditto.
4492 (_mm512_i64scatter_epi64): Ditto.
4493 (_mm512_mask_i64scatter_epi64): Ditto.
4494 * config/i386/avx512pfintrin.h
4495 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
4496 (_mm512_mask_prefetch_i32gather_ps): Ditto.
4497 (_mm512_mask_prefetch_i64gather_pd): Ditto.
4498 (_mm512_mask_prefetch_i64gather_ps): Ditto.
4499 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
4500 (_mm512_prefetch_i32scatter_ps): Ditto.
4501 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
4502 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
4503 (_mm512_prefetch_i64scatter_pd): Ditto.
4504 (_mm512_prefetch_i64scatter_ps): Ditto.
4505 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
4506 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
4507 * config/i386/avx512vlintrin.h
4508 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
4509 (_mm_mmask_i32gather_ps): Ditto.
4510 (_mm256_mmask_i32gather_pd): Ditto.
4511 (_mm_mmask_i32gather_pd): Ditto.
4512 (_mm256_mmask_i64gather_ps): Ditto.
4513 (_mm_mmask_i64gather_ps): Ditto.
4514 (_mm256_mmask_i64gather_pd): Ditto.
4515 (_mm_mmask_i64gather_pd): Ditto.
4516 (_mm256_mmask_i32gather_epi32): Ditto.
4517 (_mm_mmask_i32gather_epi32): Ditto.
4518 (_mm256_mmask_i32gather_epi64): Ditto.
4519 (_mm_mmask_i32gather_epi64): Ditto.
4520 (_mm256_mmask_i64gather_epi32): Ditto.
4521 (_mm_mmask_i64gather_epi32): Ditto.
4522 (_mm256_mmask_i64gather_epi64): Ditto.
4523 (_mm_mmask_i64gather_epi64): Ditto.
4524 (_mm256_i32scatter_ps): Change __addr type to void*.
4525 (_mm256_mask_i32scatter_ps): Ditto.
4526 (_mm_i32scatter_ps): Ditto.
4527 (_mm_mask_i32scatter_ps): Ditto.
4528 (_mm256_i32scatter_pd): Ditto.
4529 (_mm256_mask_i32scatter_pd): Ditto.
4530 (_mm_i32scatter_pd): Ditto.
4531 (_mm_mask_i32scatter_pd): Ditto.
4532 (_mm256_i64scatter_ps): Ditto.
4533 (_mm256_mask_i64scatter_ps): Ditto.
4534 (_mm_i64scatter_ps): Ditto.
4535 (_mm_mask_i64scatter_ps): Ditto.
4536 (_mm256_i64scatter_pd): Ditto.
4537 (_mm256_mask_i64scatter_pd): Ditto.
4538 (_mm_i64scatter_pd): Ditto.
4539 (_mm_mask_i64scatter_pd): Ditto.
4540 (_mm256_i32scatter_epi32): Ditto.
4541 (_mm256_mask_i32scatter_epi32): Ditto.
4542 (_mm_i32scatter_epi32): Ditto.
4543 (_mm_mask_i32scatter_epi32): Ditto.
4544 (_mm256_i32scatter_epi64): Ditto.
4545 (_mm256_mask_i32scatter_epi64): Ditto.
4546 (_mm_i32scatter_epi64): Ditto.
4547 (_mm_mask_i32scatter_epi64): Ditto.
4548 (_mm256_i64scatter_epi32): Ditto.
4549 (_mm256_mask_i64scatter_epi32): Ditto.
4550 (_mm_i64scatter_epi32): Ditto.
4551 (_mm_mask_i64scatter_epi32): Ditto.
4552 (_mm256_i64scatter_epi64): Ditto.
4553 (_mm256_mask_i64scatter_epi64): Ditto.
4554 (_mm_i64scatter_epi64): Ditto.
4555 (_mm_mask_i64scatter_epi64): Ditto.
4556 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
4557 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
4558 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
4559 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
4560 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
4561 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
4562 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
4563 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
4564 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
4565 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
4566 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
4567 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
4568 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
4569 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
4570 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
4571 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
4572 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
4573 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
4574 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
4575 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
4576 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
4577 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
4578 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
4579 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
4580 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
4581 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
4582 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
4583 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
4584 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
4585 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
4586 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
4587 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
4588 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
4589 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
4590 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
4591 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
4592 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
4593 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
4594 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
4595 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
4596 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
4597 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
4598 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
4599 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
4600 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
4601 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
4602 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
4603 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
4604 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
4605 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
4606 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
4607 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
4608 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
4609 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
4610 definitions accordingly.
4612 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
4613 Kuan-Lin Chen <kuanlinchentw@gmail.com>
4616 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
4619 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
4622 * ira-conflicts.c (ira_build_conflicts): Update total conflict
4623 hard regs for inner regno.
4625 2017-01-17 Martin Liska <mliska@suse.cz>
4628 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
4629 assumption and add comment.
4631 2017-01-17 Nathan Sidwell <nathan@acm.org>
4633 * ipa-visibility.c (localize_node): New function, broken out of ...
4634 (function_and_variable_visibility): ... here. Call it.
4636 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
4639 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
4640 correctly set frequency of oudgoing edge.
4641 (duplicate_thread_path): Fix profile updating.
4643 2017-01-17 Jakub Jelinek <jakub@redhat.com>
4646 * configure.ac: Add GCC_BASE_VER.
4647 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
4648 version from BASE-VER file.
4649 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
4650 (gcc.o): Depend on $(BASEVER).
4651 * common.opt (dumpfullversion): New option.
4652 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
4653 * doc/invoke.texi: Document -dumpfullversion.
4654 * doc/install.texi: Document --with-gcc-major-version-only.
4655 * configure: Regenerated.
4657 2017-01-17 Richard Biener <rguenther@suse.de>
4659 PR tree-optimization/71433
4660 * tree-vrp.c (register_new_assert_for): Merge same asserts
4661 on all incoming edges.
4662 (process_assert_insertions_for): Handle insertions at the
4665 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
4667 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
4668 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
4670 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
4673 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
4676 2017-01-17 Alan Modra <amodra@gmail.com>
4679 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
4680 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
4681 symbolic stack limit when pic.
4683 2017-01-16 Martin Sebor <msebor@redhat.com>
4685 PR tree-optimization/78608
4686 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
4688 2017-01-16 Jeff Law <law@redhat.com>
4691 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
4692 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
4693 for several include directories that may be relative to sysroot.
4694 * config/i386/x-mingw32 (gplus_includedir): Define.
4695 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
4696 (native_system_includedir): Likewise.
4697 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
4698 override if TARGET_SYSTEM_ROOT is defined.
4699 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
4701 PR tree-optimization/79090
4702 PR tree-optimization/33562
4703 PR tree-optimization/61912
4704 PR tree-optimization/77485
4705 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
4706 and computed trims into the dump file.
4708 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
4710 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
4712 2017-01-16 Jakub Jelinek <jakub@redhat.com>
4715 * gimplify.c (gimplify_init_constructor): If want_value and
4716 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
4720 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
4721 sequence. Formatting fixes.
4722 (doloop_optimize): Formatting fixes.
4725 * gcc.c (debug_level_greater_than_spec_func): New function.
4726 (static_spec_functions): Add debug-level-gt spec function.
4727 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
4729 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4730 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
4731 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
4732 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
4733 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
4734 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
4736 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
4738 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
4739 QImode fixups to general and mask registers only.
4741 2017-01-16 Carl Love <cel@us.ibm.com>
4743 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
4744 for built-in functions
4745 vector signed char vec_nabs (vector signed char)
4746 vector signed short vec_nabs (vector signed short)
4747 vector signed int vec_nabs (vector signed int)
4748 vector signed long long vec_nabs (vector signed long long)
4749 vector float vec_nabs (vector float)
4750 vector double vec_nabs (vector double)
4751 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
4753 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
4754 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
4755 * doc/extend.texi: Update the documentation file for the new built-in
4758 2017-01-16 Martin Sebor <msebor@redhat.com>
4760 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
4763 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4765 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
4766 UNSPEC_VSX__XXSPLTD to require special splat handling.
4768 2017-01-16 David Malcolm <dmalcolm@redhat.com>
4771 * system.h: Poison strndup.
4773 2017-01-16 Alan Modra <amodra@gmail.com>
4776 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
4779 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
4781 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
4783 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
4785 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
4786 call recog here. Assert that INSN_CODE (insn) is non-negative.
4788 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4791 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
4793 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
4794 in the currently scheduled RTL fragment.
4796 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
4798 PR rtl-optimization/78751
4799 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
4802 2017-01-14 Jeff Law <law@redhat.com>
4804 PR tree-optimization/79090
4805 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
4806 variable length stores.
4807 (compute_trims): Delete dead assignment to *trim_tail.
4808 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
4811 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
4813 PR rtl-optimization/78626
4814 PR rtl-optimization/78727
4815 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
4816 of a block, and split such blocks after everything else is finished.
4818 2017-01-14 Alan Modra <amodra@gmail.com>
4821 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
4822 target legitimate_combined_insn.
4823 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
4824 (rs6000_legitimate_combined_insn): New function.
4825 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
4827 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
4828 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
4829 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
4831 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
4833 * doc/frontends.texi (G++ and GCC): Remove references to Java.
4835 2017-01-13 Jeff Law <law@redhat.com>
4837 PR tree-optimization/33562
4838 PR tree-optimization/61912
4839 PR tree-optimization/77485
4840 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
4842 (delete_dead_assignment): Likewise.
4843 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
4844 statement to delete_dead_call and delete_dead_assignment.
4846 2017-01-13 David Malcolm <dmalcolm@redhat.com>
4849 * substring-locations.c (format_warning_va): Strengthen case 1 so
4850 that both endpoints of the substring must be within the format
4851 range for just the substring to be printed.
4853 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
4855 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
4856 * config/i386/i386.c (ix86_target_string): Add missing options
4857 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
4858 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
4859 flags_other and ix86_target_other to flags2_other. Display unknown
4861 (ix86_valid_target_attribute_inner_p): Add missing options and
4862 reorder options by implied ISAs, as in ix86_target_string.
4864 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4866 * hash-table.h (hash_table::too_empty_p): New function.
4867 (hash_table::expand): Use it.
4868 (hash_table::traverse): Likewise.
4869 (hash_table::empty_slot): Use sizeof (value_type) instead of
4870 sizeof (PTR) to convert bytes to elements. Shrink the table
4871 if the current size is excessive for the current number of
4874 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
4876 * ira-costs.c (record_reg_classes): Break from the inner loop
4877 early once alt_fail is known to be true. Update outer loop
4878 handling accordingly.
4880 2017-01-13 Jeff Law <law@redhat.com>
4882 * tree-ssa-dse.c (decrement_count): New function.
4883 (increment_start_addr, maybe_trim_memstar_call): Likewise.
4884 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
4885 when we know the partially dead statement is a mem* function.
4887 PR tree-optimization/61912
4888 PR tree-optimization/77485
4889 * tree-ssa-dse.c: Include expr.h.
4890 (maybe_trim_constructor_store): New function.
4891 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
4893 PR tree-optimization/33562
4894 PR tree-optimization/61912
4895 PR tree-optimization/77485
4896 * doc/invoke.texi: Document new dse-max-object-size param.
4897 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
4898 * tree-ssa-dse.c: Include params.h.
4899 (dse_store_status): New enum.
4900 (initialize_ao_ref_for_dse): New, partially extracted from
4902 (valid_ao_ref_for_dse, normalize_ref): New.
4903 (setup_live_bytes_from_ref, compute_trims): Likewise.
4904 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
4905 (maybe_trim_partially_dead_store): Likewise.
4906 (maybe_trim_complex_store): Likewise.
4907 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
4908 Track what bytes live from the original store. Return tri-state
4909 for dead, partially dead or live.
4910 (dse_dom_walker): Add constructor, destructor and new private members.
4911 (delete_dead_call, delete_dead_assignment): New extracted from
4913 (dse_optimize_stmt): Make a member of dse_dom_walker.
4914 Use initialize_ao_ref_for_dse.
4916 PR tree-optimization/33562
4917 PR tree-optimization/61912
4918 PR tree-optimization/77485
4919 * sbitmap.h (bitmap_count_bits): Prototype.
4920 (bitmap_clear_range, bitmap_set_range): Likewise.
4921 * sbitmap.c (bitmap_clear_range): New function.
4922 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
4924 2017-01-13 Martin Liska <mliska@suse.cz>
4927 * function.c (set_cfun): Add new argument force.
4928 * function.h (set_cfun): Likewise.
4929 * ipa-inline-transform.c (inline_call): Use the function when
4930 strict alising from is dropped for function we inline to.
4932 2017-01-13 Richard Biener <rguenther@suse.de>
4934 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
4935 for dumping GIMPLE INTEGER_CSTs.
4937 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4939 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
4940 to 201112L since C++17.
4942 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
4945 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
4946 if -fsanitize=kernel-address is present.
4948 2017-01-13 Richard Biener <rguenther@suse.de>
4950 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
4951 as _Literal ( type ) number in case usual suffixes do not
4952 preserve all information.
4954 2017-01-13 Richard Biener <rguenther@suse.de>
4956 PR tree-optimization/77283
4957 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
4958 and ssa-iterators.h.
4959 (is_feasible_trace): Implement a cost model based on joiner
4962 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4965 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
4966 char or short to __float128/_Float128 directly.
4968 2017-01-12 Martin Sebor <msebor@redhat.com>
4970 to -Wformat-overflow.
4971 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
4972 (min_bytes_remaining): Same.
4973 (get_string_length): Same.
4974 (format_string): Same.
4975 (format_directive): Same.
4977 (pass_sprintf_length::handle_gimple_call): Same.
4979 2017-01-12 Jakub Jelinek <jakub@redhat.com>
4981 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
4982 info.nowrite calls with no lhs that can't throw. Return bool
4983 whether gsi_remove has been called or not.
4984 (pass_sprintf_length::handle_gimple_call): Return bool whether
4985 try_substitute_return_value called gsi_remove. Formatting fix.
4986 (pass_sprintf_length::execute): Don't use gsi_remove if
4987 handle_gimple_call returned true.
4990 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
4991 be removed due to side-effects, don't remove following barrier nor
4992 turn the successor edge into fallthru edge.
4994 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4997 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
4998 element-reversing loads and stores as not swappable.
5000 2017-01-12 Nathan Sidwell <nathan@acm.org>
5001 Nicolai Stange <nicstange@gmail.com>
5003 * combine.c (try_combine): Don't ignore result of overlap checking
5004 loop. Combine overlap & asm check into single loop.
5006 2017-01-12 Richard Biener <rguenther@suse.de>
5008 * tree-pretty-print.c (dump_generic_node): Provide -gimple
5009 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
5011 2017-01-12 Richard Biener <rguenther@suse.de>
5013 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
5014 and TS_TARGET_OPTION directly derive from TS_BASE.
5015 * tree-core.h (tree_optimization_option): Derive from tree_base.
5016 (tree_target_option): Likewise.
5018 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
5020 * config/i386/i386.c (memory_address_length): Increase len
5021 only when rip_relative_addr_p returns false.
5023 2017-01-11 Julia Koval <julia.koval@intel.com>
5025 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
5026 (OPTION_MASK_ISA_SGX_SET): New.
5027 (ix86_handle_option): Handle OPT_msgx.
5028 * config.gcc: Added sgxintrin.h.
5029 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
5030 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
5031 * config/i386/i386.c (ix86_target_string): Add -msgx.
5033 (ix86_option_override_internal): Handle new options.
5034 (ix86_valid_target_attribute_inner_p): Add sgx.
5035 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
5036 * config/i386/i386.opt: Add msgx.
5037 * config/i386/sgxintrin.h: New file.
5038 * config/i386/x86intrin.h: Add sgxintrin.h.
5040 2017-01-11 Jakub Jelinek <jakub@redhat.com>
5043 * fold-const.c (maybe_nonzero_address): Return 1 for function
5045 (tree_single_nonzero_warnv_p): Don't handle function local objects
5049 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
5052 2017-01-11 David Malcolm <dmalcolm@redhat.com>
5055 * opts.c: Include "spellcheck.h"
5056 (struct string_fragment): New struct.
5057 (struct edit_distance_traits<const string_fragment &>): New
5059 (get_closest_sanitizer_option): New function.
5060 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
5062 2017-01-11 Jakub Jelinek <jakub@redhat.com>
5064 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
5066 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
5067 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
5068 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
5069 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
5070 for initial die_offset if dwarf_split_debug_info.
5071 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
5072 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
5074 (output_skeleton_debug_sections): Formatting fix. Use
5075 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
5076 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
5078 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
5080 * config/arm/cortex-a53.md: Add bypasses for
5082 (cortex_a53_r2f): Only use for transfers.
5083 (cortex_a53_f2r): Likewise.
5084 (cortex_a53_r2f_cvt): Add reservation for conversions.
5085 (cortex_a53_f2r_cvt): Likewise.
5087 2017-01-11 Tamar Christina <tamar.christina@arm.com>
5089 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
5090 to all inlined functions, change static to extern.
5092 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
5095 * config/arm/arm.c (legitimize_pic_address): Handle reference to
5097 (arm_assemble_integer): Likewise.
5099 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
5101 * config.gcc: Use new awk script to check CPU, FPU and architecture
5102 parameters for --with-... options.
5103 * config/arm/parsecpu.awk: New file
5104 * config/arm/arm-cpus.in: New file.
5105 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
5107 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
5109 * config/arm/t-arm: Update dependency rules.
5110 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
5111 of processing .def files.
5112 * config/arm/genopt.sh: Deleted.
5113 * config/arm/gentune.sh: Deleted.
5114 * config/arm/arm-cores.def: Deleted.
5115 * config/arm/arm-arches.def: Deleted.
5116 * config/arm/arm-fpus.def: Deleted.
5117 * config/arm/arm-tune.md: Regenerated.
5118 * config/arm/arm-tables.opt: Regenerated.
5119 * config/arm/arm-cpu.h: New generated file.
5120 * config/arm/arm-cpu-data.h: New generated file.
5121 * config/arm/arm-cpu-cdata.h: New generated file.
5123 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
5126 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
5128 (input_varpool_node): Unpack dynamically_initialized bit.
5130 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
5132 PR rtl-optimization/79032
5133 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
5134 the alignment of the adjusted memory reference against that of MODE,
5135 instead of the alignment of the original memory reference.
5137 2017-01-11 Martin Jambor <mjambor@suse.cz>
5139 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
5141 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
5142 decorated functions.
5144 2017-01-11 Richard Biener <rguenther@suse.de>
5146 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
5147 set range/nonnull info for PHI results. Do not set it on
5148 stmts marked for removal.
5150 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
5152 * expr.c (store_field): In the bitfield case, fetch the return value
5153 from the registers before applying a single big-endian adjustment.
5154 Always do a final load for a BLKmode value not larger than a word.
5156 2017-01-10 David Malcolm <dmalcolm@redhat.com>
5159 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5160 that we correctly handle column numbers greater than
5161 LINE_MAP_MAX_COLUMN_NUMBER.
5163 2017-01-10 Martin Sebor <msebor@redhat.com>
5166 * gimple-ssa-sprintf.c (get_destination_size): Call
5167 {init,fini}object_sizes.
5168 * tree-object-size.c (addr_object_size): Adjust.
5169 (pass_through_call): Adjust.
5170 (pass_object_sizes::execute): Adjust.
5171 * tree-object-size.h (fini_object_sizes): Declare.
5173 2017-01-10 Martin Sebor <msebor@redhat.com>
5175 PR tree-optimization/78775
5176 * builtins.c (get_size_range): Move...
5177 * calls.c: ...to here.
5178 (alloc_max_size): Accept zero argument.
5179 (operand_signed_p): Remove.
5180 (maybe_warn_alloc_args_overflow): Call get_size_range.
5181 * calls.h (get_size_range): Declare.
5183 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
5185 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
5186 from TI's devices.csv file as of September 2016.
5187 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
5189 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
5191 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
5192 * doc/invoke.texi: Likewise.
5193 * doc/md.texi: Likewise.
5194 * doc/objc.texi: Likewise.
5196 2017-01-10 Joshua Conner <joshconner@google.com>
5198 * config/arm/fuchsia-elf.h: New file.
5199 * config/fuchsia.h: New file.
5200 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
5201 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
5203 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
5205 2016-01-10 Richard Biener <rguenther@suse.de>
5207 PR tree-optimization/79034
5208 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
5209 Propagate out degenerate PHIs in the joiner.
5211 2017-01-10 Martin Liska <mliska@suse.cz>
5213 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
5214 (sort_congruence_classes_by_decl_uid): Likewise.
5215 (sort_congruence_class_groups_by_decl_uid): Likewise.
5216 (sem_item_optimizer::merge_classes): Sort class, groups in these
5217 classes and members in the groups by DECL_UID of declarations.
5218 This would make merge operations stable.
5220 2017-01-10 Martin Liska <mliska@suse.cz>
5222 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
5223 usage of m_classes_vec.
5224 (sem_item_optimizer::~sem_item_optimizer): Likewise.
5225 (sem_item_optimizer::get_group_by_hash): Likewise.
5226 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
5227 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
5228 (sem_item_optimizer::verify_classes): Likewise.
5229 (sem_item_optimizer::process_cong_reduction): Likewise.
5230 (sem_item_optimizer::dump_cong_classes): Likewise.
5231 (sem_item_optimizer::merge_classes): Likewise.
5232 * ipa-icf.h (congruence_class_hash): Rename from
5233 congruence_class_group_hash. Remove declaration of m_classes_vec.
5235 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
5237 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
5238 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
5239 * config.gcc: Add avx512vpopcntdqintrin.h.
5240 * config/i386/avx512vpopcntdqintrin.h: New.
5241 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
5242 * config/i386/i386-builtin-types.def: Add new types.
5243 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
5244 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
5245 __builtin_ia32_vpopcountq_v8di_mask): New.
5246 * config/i386/i386-c.c (ix86_target_macros_internal): Define
5247 __AVX512VPOPCNTDQ__.
5248 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
5249 (PTA_AVX512VPOPCNTDQ): Define.
5250 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
5251 TARGET_AVX512VPOPCNTDQ_P): Define.
5252 * config/i386/i386.opt: Add mavx512vpopcntdq.
5253 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
5254 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
5256 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5259 * predict.def (PRED_CALL): Set to 67.
5261 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
5263 * expr.c (store_field): In the bitfield case, if the value comes from
5264 a function call and is of an aggregate type returned in registers, do
5265 not modify the field mode; extract the value in all cases if the mode
5266 is BLKmode and the size is not larger than a word.
5268 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
5271 * config/i386/cpuid.h: Fix undefined behavior.
5273 2017-01-04 Jeff Law <law@redhat.com>
5275 PR tree-optimization/79007
5276 PR tree-optimization/67955
5277 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
5278 conservative for pt.null when flag_non_call_exceptions is on.
5280 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5282 PR translation/79019
5283 PR translation/79020
5284 * params.def (PARAM_INLINE_MIN_SPEEDUP,
5285 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
5286 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
5288 * config/avr/avr.opt (maccumulate-args): Likewise.
5289 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
5290 * common.opt (freport-bug): Likewise.
5291 * cif-code.def (CIF_FINAL_ERROR): Likewise.
5292 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
5293 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
5294 translatable string.
5295 * config/i386/i386.c (function_value_32): Likewise.
5296 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
5297 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
5299 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
5300 * common/config/msp430/msp430-common.c (msp430_handle_option):
5302 * symtab.c (symtab_node::verify_base): Likewise.
5303 * opts.c (set_debug_level): Likewise.
5304 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
5305 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
5306 missing whitespace to translatable strings.
5307 * config/avr/avr.md (bswapsi2): Fix typo in comment.
5308 * config/sh/superh.h: Likewise.
5309 * config/i386/xopintrin.h: Likewise.
5310 * config/i386/znver1.md: Likewise.
5311 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
5312 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
5313 * double-int.h (struct double_int): Likewise.
5314 * double-int.c (div_and_round_double): Likewise.
5315 * wide-int.cc: Likewise.
5316 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
5317 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
5318 * cfgcleanup.c (crossjumps_occured): Renamed to ...
5319 (crossjumps_occurred): ... this.
5320 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
5323 PR tree-optimization/78899
5324 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
5325 returning bool return struct loop *, NULL for failure and the new
5327 (versionable_outer_loop_p): Don't version outer loop if it has
5328 dont_vectorized bit set.
5329 (tree_if_conversion): When versioning outer loop, ensure
5330 tree_if_conversion is performed also on the inner loop of the
5331 non-vectorizable outer loop copy.
5332 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
5333 LOOP_VECTORIZED in inner loop of the scalar outer loop and
5334 prevent vectorization of it.
5335 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
5336 the outer loop vectorization of the non-scalar version is attempted
5337 before vectorization of the inner loop in scalar version. If
5338 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
5339 vectorization of its inner loop.
5340 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
5341 has 2 inner loops, rename also on edges from bb whose single pred
5342 is outer_loop->header. Fix typo in function comment.
5344 2017-01-09 Martin Sebor <msebor@redhat.com>
5347 * asan.c (asan_emit_stack_protection): Increase local buffer size
5348 to avoid snprintf truncation warning.
5350 2017-01-09 Andrew Pinski <apinski@cavium.com>
5352 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
5353 to reference thunderx2t99 for the tuning structure
5354 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
5356 (thunderx2t99_extra_costs): This.
5357 * config/aarch64/aarch64-tune.md: Regenerate.
5358 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
5359 (vulcan_addrcost_table): This.
5360 (vulcan_regmove_cost): Rename to ...
5361 (thunderx2t99_regmove_cost): This.
5362 (vulcan_vector_cost): Rename to ...
5363 (thunderx2t99_vector_cost): this.
5364 (vulcan_branch_cost): Rename to ...
5365 (thunderx2t99_branch_cost): This.
5366 (vulcan_tunings): Rename to ...
5367 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
5368 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
5370 2017-01-09 Martin Jambor <mjambor@suse.cz>
5374 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
5375 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
5376 (propagate_vr_accross_jump_function): Use the above function for all
5377 value range computations for pass-through jump functions and type
5378 converasion from explicit value range values.
5379 (ipcp_propagate_stage): Do not attempt to deduce types of formal
5380 parameters from TYPE_ARG_TYPES.
5381 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
5382 (ipa_write_node_info): Stream type of the actual argument.
5383 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
5385 2017-01-09 Martin Liska <mliska@suse.cz>
5388 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
5389 (lookup_compiler): Do not show error message with have_E.
5391 2017-01-09 Jakub Jelinek <jakub@redhat.com>
5393 PR tree-optimization/78938
5394 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
5395 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
5396 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
5397 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
5400 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5402 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
5405 2017-01-09 Richard Biener <rguenther@suse.de>
5407 PR tree-optimization/78997
5408 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
5409 name condition properly.
5411 2017-01-09 Richard Biener <rguenther@suse.de>
5414 * dwarf2out.c (is_cxx): New overload with context.
5415 (is_naming_typedef_decl): Use it.
5417 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5419 * invoke.texi (Option Summary): Correct spacing in option lists
5420 and add line breaks to fix over-long lines.
5422 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5426 * extend.texi (Common Variable Attributes): Add xref to GCC
5427 Internals manual to explain mode attribute keywords.
5429 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
5432 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
5433 and Preprocessor Options.
5434 (Options for Linking): Document -pthread here....
5435 (RS/6000 and PowerPC Options): ...not here.
5436 (Solaris 2 Options): ...or here.
5437 * doc/cppopts.texi: Document -pthread.
5439 2017-01-08 Martin Sebor <msebor@redhat.com>
5442 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
5443 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
5444 New member functions.
5445 (format_directive): Used them.
5447 (pass_sprintf_length::handle_gimple_call): Same.
5448 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
5449 to avoid truncation for any argument.
5450 (extract_affine_mul): Same.
5451 * tree.c (get_file_function_name): Same.
5453 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5456 * predict.def (PRED_INDIR_CALL): Set to 86.
5458 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5460 PR preprocessor/54124
5461 * doc/cppopts.texi: Reformat -d subtable to list the full name
5462 of the options. Add cross-reference to the docs for the general
5463 compiler -d options.
5464 * doc/invoke.texi (Developer Options): Add cross-reference to the
5465 preprocessor-specific -d option documentation.
5467 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5469 PR preprocessor/13498
5470 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
5471 redudant material, and reflect new command-line options.
5472 (System Headers): Likewise.
5474 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
5476 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
5477 -isystem, and -idirafter. Copy-edit.
5478 * doc/cppopts.texi: Copy-edit. Remove contradiction about
5479 default for -ftrack-macro-expansion. Delete obsolete and
5480 badly-formatted implementation details about -fdebug-cpp output.
5481 * doc/cppwarnopts.texi: Copy-edit.
5483 2017-01-07 David Malcolm <dmalcolm@redhat.com>
5486 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
5487 that the transition from a max line width >= 1<<10 to narrower
5488 lines works correctly.
5490 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
5492 * doc/options.texi (PerFunction): New.
5493 * opt-functions.awk (switch_flags): Map both Optimization and
5494 PerFunction to CL_OPTIMIZATION.
5495 * opth-gen.awk: Test for PerFunction flag along with
5497 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
5498 it only when the latter is present. Skip those that don't in
5499 the hash function generator.
5500 * common.opt (fvar-tracking): Mark as PerFunction instead of
5502 (fvar-tracking-assignments): Likewise.
5503 (fvar-tracking-assignments-toggle): Likewise.
5504 (fvar-tracking-uninit): Likewise.
5506 2017-01-07 Jakub Jelinek <jakub@redhat.com>
5508 PR translation/79018
5509 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
5512 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
5515 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
5516 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
5517 TARGET_LONG_JUMP_TABLE_OFFSETS.
5518 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
5519 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
5520 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
5521 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
5522 * config/m68k/m68k.md (tablejump expander): Likewise.
5523 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
5524 TARGET_LONG_JUMP_TABLE_OFFSETS.
5525 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
5526 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
5528 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
5529 David Holsgrove <david.holsgrove@xilinx.com>
5531 * common/config/microblaze/microblaze-common.c
5532 (TARGET_EXCEPT_UNWIND_INFO): Remove.
5533 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
5535 * config/microblaze/microblaze.c (microblaze_must_save_register)
5536 (microblaze_expand_epilogue, microblaze_return_addr): Handle
5538 (microblaze_eh_return): New function.
5539 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
5540 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
5541 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
5542 * config/microblaze/microblaze.md (eh_return): New pattern.
5544 2017-01-06 Jakub Jelinek <jakub@redhat.com>
5546 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
5547 GCC_DIAGNOSTIC_STRINGIFY): Define.
5549 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
5551 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5553 * config/arm/arm.md (<mcrr>): New.
5555 * config/arm/arm.c (arm_arch5te): New.
5556 (arm_option_override): Set arm_arch5te.
5557 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
5559 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
5560 (arm_mcrr_qualifiers): ... this. New.
5561 (MRRC_QUALIFIERS): Define to...
5562 (arm_mrrc_qualifiers): ... this. New.
5563 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
5565 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
5566 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
5567 (MRRCI, mrrc, MRRC): New.
5568 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
5569 VUNSPEC_MRRC2): New.
5571 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5573 * config/arm/arm.md (<mcr>): New.
5575 * config/arm/arm.c (arm_coproc_builtin_available): Add
5576 support for mcr, mrc, mcr2 and mrc2.
5577 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
5578 (arm_mcr_qualifiers): ... this. New.
5579 (MRC_QUALIFIERS): Define to ...
5580 (arm_mrc_qualifiers): ... this. New.
5581 (MCR_QUALIFIERS): Define to ...
5582 (arm_mcr_qualifiers): ... this. New.
5583 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
5585 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
5586 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
5587 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
5590 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5592 * config/arm/arm.md (*ldc): New.
5596 * config/arm/arm.c (arm_coproc_builtin_available): Add
5597 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
5598 (arm_coproc_ldc_stc_legitimate_address): New.
5599 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
5600 'qualifier_const_pointer'.
5601 (LDC_QUALIFIERS): Define to...
5602 (arm_ldc_qualifiers): ... this. New.
5603 (STC_QUALIFIERS): Define to...
5604 (arm_stc_qualifiers): ... this. New.
5605 * config/arm/arm-protos.h
5606 (arm_coproc_ldc_stc_legitimate_address): New.
5607 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
5608 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
5609 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
5610 stc2, stcl, stc2l): New.
5611 * config/arm/constraints.md (Uz): New.
5612 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
5613 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
5614 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
5615 VUNSPEC_STC2L): New.
5617 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5619 * config/arm/arm.md (<cdp>): New.
5620 * config/arm/arm.c (neon_const_bounds): Rename this ...
5621 (arm_const_bounds): ... this.
5622 (arm_coproc_builtin_available): New.
5623 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
5624 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
5625 (CDP_QUALIFIERS): Define to...
5626 (arm_cdp_qualifiers): ... this. New.
5628 (arm_expand_builtin_args): Add case for 6 arguments.
5629 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
5630 (arm_const_bounds): ... this.
5631 (arm_coproc_builtin_available): New.
5632 * config/arm/arm_acle.h (__arm_cdp): New.
5634 * config/arm/arm_acle_builtins.def (cdp): New.
5636 * config/arm/iterators.md (CDPI,CDP,cdp): New.
5637 * config/arm/neon.md: Rename all 'neon_const_bounds' to
5639 * config/arm/types.md (coproc): New.
5640 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
5641 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
5642 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
5643 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
5645 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5647 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
5648 (UBINOP_QUALIFIERS): New.
5650 (acle_builtin_data): New. Change comment.
5651 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
5652 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
5653 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
5654 arm_acle_builtins.def.
5655 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
5656 (arm_init_acle_builtins): New.
5657 (CRC32_BUILTIN): Remove.
5658 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
5659 crc32cb, crc32ch and crc32cw.
5660 (arm_init_crc32_builtins): Remove.
5661 (arm_init_builtins): Use arm_init_acle_builtins rather
5662 than arm_init_crc32_builtins.
5663 (arm_expand_acle_builtin): New.
5664 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
5665 * config/arm/arm_acle_builtins.def: New.
5667 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
5669 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
5670 (arm_builtin_datum): ... this.
5671 (arm_init_neon_builtin): Rename to ...
5672 (arm_init_builtin): ... this. Add a new parameters PREFIX
5673 and USE_SIG_IN_NAME.
5674 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
5675 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
5676 'arm_builtin_datum'.
5677 (arm_init_vfp_builtins): Likewise.
5678 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
5679 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
5680 (arm_expand_neon_args): Rename to ...
5681 (arm_expand_builtin_args): ... this. Rename builtin_arg
5682 enum values and differentiate between ARG_BUILTIN_MEMORY
5683 and ARG_BUILTIN_NEON_MEMORY.
5684 (arm_expand_neon_builtin_1): Rename to ...
5685 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
5686 values, arm_expand_builtin_args and add bool parameter NEON.
5687 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
5688 (arm_expand_vfp_builtin): Likewise.
5689 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
5691 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
5694 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
5695 * predict.c (tree_estimate_probability_bb): Reverse direction of
5696 polymorphic call predictor.
5698 2017-01-06 David Malcolm <dmalcolm@redhat.com>
5700 * passes.c (execute_one_pass): Split out pass-skipping logic into...
5701 (determine_pass_name_match): ...this new function and...
5702 (should_skip_pass_p): ...this new function.
5704 2017-01-06 Nathan Sidwell <nathan@acm.org>
5706 * ipa-visibility.c (function_and_variable_visibility): Reformat
5707 comments and long lines. Remove extrneous if.
5708 * symtab.c (symtab_node::make_decl_local): Fix code format.
5709 (symtab_node::set_section_for_node): Fix comment typo.
5711 2017-01-06 Martin Liska <mliska@suse.cz>
5714 * lra-constraints.c: Rename invariant to lra_invariant.
5715 * predict.c (set_even_probabilities): Initialize e to NULL.
5717 2017-01-05 Martin Sebor <msebor@redhat.com>
5719 PR tree-optimization/78910
5720 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
5721 (format_integer): Correct off-by-one error in the handling
5722 of precision with negative numbers in signed conversions..
5724 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
5726 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
5728 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5730 PR tree-optimization/71016
5731 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
5732 factor_out_conditional_conversion. Formatting fix.
5733 (factor_out_conditional_conversion): Add cond_stmt argument.
5734 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
5735 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
5738 2017-01-05 David Malcolm <dmalcolm@redhat.com>
5740 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
5741 read-rtl-function.o, and selftest-rtl.o.
5742 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
5743 (selftest::aarch64_test_loading_full_dump): New function.
5744 (selftest::aarch64_run_selftests): New function.
5745 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
5746 selftest::aarch64_run_selftests.
5747 * config/i386/i386.c
5748 (selftest::ix86_test_loading_dump_fragment_1): New function.
5749 (selftest::ix86_test_loading_call_insn): New function.
5750 (selftest::ix86_test_loading_full_dump): New function.
5751 (selftest::ix86_test_loading_unspec): New function.
5752 (selftest::ix86_run_selftests): Call the new functions.
5753 * emit-rtl.c (maybe_set_max_label_num): New function.
5754 * emit-rtl.h (maybe_set_max_label_num): New decl.
5755 * function.c (instantiate_decls): Guard call to
5756 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
5757 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
5759 * gensupport.c (gen_reader::gen_reader): Pass "false"
5760 for new "compact" param of rtx_reader.
5761 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
5762 rather than an empty string for NULL strings.
5763 * read-md.c: Potentially include config.h rather than bconfig.h.
5764 Wrap include of errors.h with #ifdef GENERATOR_FILE.
5765 (have_error): New global, copied from errors.c.
5766 (md_reader::read_name): Rename to...
5767 (md_reader::read_name_1): ...this, adding "out_loc" param,
5768 and converting "missing name or number" to returning false, rather
5770 (md_reader::read_name): Reimplement in terms of read_name_1.
5771 (md_reader::read_name_or_nil): New function.
5772 (md_reader::read_string): Handle "(nil)" by returning NULL.
5773 (md_reader::md_reader): Add new param "compact".
5774 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
5775 (md_reader::read_file): New method.
5776 * read-md.h (md_reader::md_reader): Add new param "compact".
5777 (md_reader::read_file): New method.
5778 (md_reader::is_compact): New accessor.
5779 (md_reader::read_name): Convert return type from void to file_location.
5780 (md_reader::read_name_or_nil): New decl.
5781 (md_reader::read_name_1): New decl.
5782 (md_reader::m_compact): New field.
5783 (noop_reader::noop_reader): Pass "false" for new "compact" param
5785 (rtx_reader::rtx_reader): Add new "compact" param.
5786 (rtx_reader::read_rtx_operand): Make virtual and convert return
5787 type from void to rtx.
5788 (rtx_reader::read_until): New decl.
5789 (rtx_reader::handle_any_trailing_information): New virtual function.
5790 (rtx_reader::postprocess): New virtual function.
5791 (rtx_reader::finalize_string): New virtual function.
5792 (rtx_reader::m_in_call_function_usage): New field.
5793 (rtx_reader::m_reuse_rtx_by_id): New field.
5794 * read-rtl-function.c: New file.
5795 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
5796 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
5797 (selftest::verify_three_block_rtl_cfg): New decl.
5798 * read-rtl-function.h: New file.
5799 * read-rtl.c: Potentially include config.h rather than bconfig.h.
5800 For host, include function.h, memmodel.h, and emit-rtl.h.
5801 (one_time_initialization): New function.
5802 (struct compact_insn_name): New struct.
5803 (compact_insn_names): New array.
5804 (find_code): Handle insn codes in compact dumps.
5805 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
5806 (bind_subst_iter_and_attr): Likewise.
5807 (add_condition_to_string): Likewise.
5808 (add_condition_to_rtx): Likewise.
5809 (apply_attribute_uses): Likewise.
5810 (add_current_iterators): Likewise.
5811 (apply_iterators): Likewise.
5812 (initialize_iterators): Guard usage of apply_subst_iterator with
5813 #ifdef GENERATOR_FILE.
5814 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
5815 (md_reader::read_mapping): Likewise.
5816 (add_define_attr_for_define_subst): Likewise.
5817 (add_define_subst_attr): Likewise.
5818 (read_subst_mapping): Likewise.
5819 (check_code_iterator): Likewise.
5820 (rtx_reader::read_rtx): Likewise. Move one-time initialization
5822 (one_time_initialization): New function.
5823 (rtx_reader::read_until): New method.
5824 (read_flags): New function.
5825 (parse_reg_note_name): New function.
5826 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
5827 Handle reuse_rtx ids.
5828 Wrap iterator lookup within #ifdef GENERATOR_FILE.
5829 Add parsing support for RTL dumps, mirroring the special-cases in
5830 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
5831 values, and calling handle_any_trailing_information.
5832 (rtx_reader::read_rtx_operand): Convert return type from void
5833 to rtx, returning return_rtx. Handle case 'e'. Call
5834 finalize_string on XSTR and XTMPL fields.
5835 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
5836 "(nil)" values were omitted. Call the postprocess vfunc on the
5838 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
5839 class ctor. Initialize m_in_call_function_usage. Call
5840 one_time_initialization.
5841 * rtl-tests.c (selftest::test_uncond_jump): Call
5842 set_new_first_and_last_insn.
5843 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
5844 * selftest-rtl.c: New file.
5845 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
5846 (selftest::get_insn_by_uid): New decl.
5847 * selftest-run-tests.c (selftest::run_tests): Call
5848 read_rtl_function_c_tests.
5849 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
5850 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
5853 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
5855 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
5856 operands in a special way. Assert that pos+len <= mode precision.
5858 2017-01-05 Jakub Jelinek <jakub@redhat.com>
5860 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
5861 3 argument Alias with unlimited for the negative form.
5862 (fno-vect-cost-model): Removed.
5864 2017-01-05 Martin Liska <mliska@suse.cz>
5866 * hsa-gen.c (gen_hsa_divmod): New function.
5867 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
5869 2017-01-05 Martin Liska <mliska@suse.cz>
5872 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
5875 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5877 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
5878 small constant length operands.
5880 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5882 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
5883 between loop iterations.
5885 2017-01-05 Martin Liska <mliska@suse.cz>
5888 * gimplify.c (gimplify_decl_expr): Compare to
5889 asan_poisoned_variables instread of checking flags.
5890 (gimplify_target_expr): Likewise.
5891 (gimplify_expr): Likewise.
5892 (gimplify_function_tree): Conditionally initialize
5893 asan_poisoned_variables.
5895 2017-01-04 Jeff Law <law@redhat.com>
5897 PR tree-optimizatin/78812
5898 * rtl.h (contains_mem_rtx_p): Prototype.
5899 * ifcvt.c (containts_mem_rtx_p): Move from here to...
5900 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
5901 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
5902 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
5903 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
5905 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5907 * input.c (assert_char_at_range): Default-initialize actual_range.
5909 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5911 * df-scan.c (df_ref_create_structure): Make regno unsigned,
5912 to match the caller.
5914 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5916 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
5917 insns after final jump in test to emit dummy move.
5919 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5921 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
5922 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
5924 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
5926 * multiple_target.c (create_dispatcher_calls): Init e_next.
5927 * tree-ssa-loop-split.c (split_loop): Init border.
5928 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
5931 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
5936 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
5937 (altivec_register_operand): Do not return true if the operand
5938 contains a SUBREG mixing SImode and SFmode.
5939 (vsx_register_operand): Likewise.
5940 (vsx_reg_sfsubreg_ok): New predicate.
5941 (vfloat_operand): Do not return true if the operand contains a
5942 SUBREG mixing SImode and SFmode.
5943 (vint_operand): Likewise.
5944 (vlogical_operand): Likewise.
5945 (gpc_reg_operand): Likewise.
5946 (int_reg_operand): Likewise.
5947 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
5948 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
5949 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
5951 (rs6000_emit_move_si_sf_subreg): New helper function.
5952 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
5953 fixup SUBREGs involving SImode and SFmode.
5954 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
5955 numbers for the new peephole2 optimization.
5956 (peephole2 for SFmode unions): New peephole2 to optimize cases in
5957 the GLIBC math library that do AND/IOR/XOR operations on single
5958 precision floating point.
5959 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
5960 target macros to say whether we need to avoid SUBREGs mixing
5962 (TARGET_ALLOW_SF_SUBREG): Likewise.
5963 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
5964 (UNSPEC_SI_FROM_SF): Likewise.
5965 (iorxor): Change spacing.
5966 (and_ior_xor): New iterator for AND, IOR, and XOR.
5967 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
5968 (movdi_from_sf_zero_ext): Likewise.
5969 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
5970 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
5971 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
5972 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
5973 (fms<mode>4): Likewise.
5974 (fnma<mode>4): Likewise.
5975 (fnms<mode>4): Likewise.
5976 (nfma<mode>4): Likewise.
5977 (nfms<mode>4): Likewise.
5979 2017-01-04 Marek Polacek <polacek@redhat.com>
5982 * doc/invoke.texi: Document -Wpointer-compare.
5984 2017-01-04 Jakub Jelinek <jakub@redhat.com>
5986 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
5989 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
5990 descriptions for -gdwarf-5 and emit them as uleb128 instead of
5993 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
5996 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
5997 documentation of the powerpc_popcntb_ok attribute.
5998 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
5999 code to issue warning messages if a requested CPU configuration is
6000 not supported by the binary (assembler and loader) toolchain.
6001 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
6002 made to define a built-in function that has been disabled.
6003 (paired_init_builtins): Add assertion to prevent ICE if attempt is
6004 made to define a built-in function that has been disabled.
6005 (altivec_init_builtins): Add comment explaining why definition
6006 of the DST built-in functions is not preceded by an assertion
6007 check. Add assertions to prevent ICE if attempts are made to
6008 define an altivec predicate or an abs* built-in function that has
6010 (htm_init_builtins): Add comment explaining why definition of the
6011 htm built-in functions is not preceded by an assertion check.
6013 2017-01-04 Jeff Law <law@redhat.com>
6015 PR tree-optimizatin/67955
6016 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
6017 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
6018 the points-to solution does not include pt_null. Use DECL_PT_UID
6021 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
6023 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
6024 Use gen_int_mode instead of gen_lopwart for const_int operands.
6026 2017-01-04 Jakub Jelinek <jakub@redhat.com>
6028 PR tree-optimization/71563
6029 * match.pd: Simplify X << Y into X if Y is known to be 0 or
6030 out of range value - has low bits known to be zero.
6032 2017-01-04 Alan Modra <amodra@gmail.com>
6034 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
6035 * configure: Regenerate.
6036 * config.in: Regenerate.
6038 2017-01-04 Jakub Jelinek <jakub@redhat.com>
6041 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
6042 a substring of the message, but strcmp with the whole message. Ifdef
6043 ENABLE_NLS, translate the message first using dgettext.
6045 2017-01-03 Jeff Law <law@redhat.com>
6047 PR tree-optimizatin/78856
6048 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
6049 (mark_threaded_blocks): Remove code to truncate thread paths that
6050 cross multiple loop headers. Instead invalidate the cached loop
6051 iteration information and handle case of a thread path walking
6052 into an irreducible region.
6054 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6057 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
6058 assertions. Add support for doing the signbit if the IEEE 128-bit
6059 floating point value is in a GPR.
6060 * config/rs6000/rs6000.md (Fsignbit): Delete.
6061 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
6062 Update the length attribute if the value is in a GPR.
6063 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
6064 the sign or zero extension instruction, since the value is always 0/1.
6065 (signbit<mode>2_dm2): Delete using <Fsignbit>.
6068 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
6069 extracting SImode to a GPR register so that we can generate a
6070 store, limit the vector to be in a traditional Altivec register
6071 for the vextuwrx instruction.
6073 2017-01-03 Ian Lance Taylor <iant@google.com>
6075 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
6077 2017-01-03 Martin Sebor <msebor@redhat.com>
6079 PR tree-optimization/78696
6080 * gimple-ssa-sprintf.c (format_floating): Correct handling of
6081 precision. Use MPFR for %f for greater fidelity. Correct handling
6083 (pass_sprintf_length::compute_format_length): Set width and precision
6084 specified by asrerisk to void_node for vararg functions.
6085 (try_substitute_return_value): Adjust dump output.
6087 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
6089 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
6091 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
6093 * doc/invoke.texi (SPARC options): Document -mlra as the default.
6094 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
6095 -mlra/-mno-lra was passed to the compiler.
6097 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
6099 PR rtl-optimization/65618
6100 * emit-rtl.c (try_split): Move initialization of "before" and
6101 "after" to just before the call to emit_insn_after_setloc.
6103 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
6105 * doc/md.texi (Standard Names): Remove reference to Java frontend.
6107 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
6109 * dwarf2out.c (gen_enumeration_type_die): When
6110 -gno-strict-dwarf, add a DW_AT_encoding attribute.
6112 2017-01-03 Jakub Jelinek <jakub@redhat.com>
6114 PR tree-optimization/78965
6115 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
6116 Change first argument from const call_info & to call_info &. For %n
6117 set info.nowrite to false.
6120 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
6121 possibly throwing calls.
6123 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
6124 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
6125 and fns handling, rather than in a separate case SSA_NAME.
6127 2017-01-02 Jeff Law <law@redhat.com>
6129 * config/darwin-driver.c (darwin_driver_init): Const-correctness
6130 fixes for first_period and second_period variables.
6132 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
6135 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
6136 (*insvqi_1): New insn pattern.
6137 (*insvqi_1_mem_rex64): Ditto.
6139 (*insvqi_3): Rename from *insvqi.
6141 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
6143 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
6145 * doc/cfg.texi (Edges): Remove reference to Java.
6146 (Maintaining the CFG): Ditto.
6148 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6151 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
6152 transparent aliases.
6154 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
6157 * predict.def (PRED_CALL): Update hitrate.
6158 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
6159 * predict.c (tree_estimate_probability_bb): Split CALL predictor
6160 into direct/indirect/polymorphic variants.
6162 2017-01-01 Jakub Jelinek <jakub@redhat.com>
6164 Update copyright years.
6166 * gcc.c (process_command): Update copyright notice dates.
6167 * gcov-dump.c (print_version): Ditto.
6168 * gcov.c (print_version): Ditto.
6169 * gcov-tool.c (print_version): Ditto.
6170 * gengtype.c (create_file): Ditto.
6171 * doc/cpp.texi: Bump @copying's copyright year.
6172 * doc/cppinternals.texi: Ditto.
6173 * doc/gcc.texi: Ditto.
6174 * doc/gccint.texi: Ditto.
6175 * doc/gcov.texi: Ditto.
6176 * doc/install.texi: Ditto.
6177 * doc/invoke.texi: Ditto.
6179 Copyright (C) 2017 Free Software Foundation, Inc.
6181 Copying and distribution of this file, with or without modification,
6182 are permitted in any medium without royalty provided the copyright
6183 notice and this notice are preserved.