1 2017-04-13 Jakub Jelinek <jakub@redhat.com>
4 * fold-const.c (fold_ternary_loc): Revert
5 use op0 instead of fold_convert_loc (loc, type, arg0) part of
8 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
10 PR rtl-optimization/80343
11 * lra-remat.c (update_scratch_ops): Assign original hard reg to
14 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
18 to ubsan_encode_value.
20 2017-04-13 Jeff Law <law@redhat.com>
22 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
23 appearing in DEBUG_INSNs.
25 2017-04-13 Martin Liska <mliska@suse.cz>
28 * gcov-io.c (gcov_write_string): Copy to buffer just when
29 allocated size is greater than zero.
31 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34 * dwarf2out.c (decls_for_scope): Ignore declarations of
35 current_function_decl in BLOCK_NONLOCALIZED_VARS.
37 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
40 * ipa-visibility.c (non_local_p): Fix typos.
41 (localize_node): When localizing symbol in same comdat group,
42 dissolve the group only when we know external symbols are going
44 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
46 2017-04-12 Jakub Jelinek <jakub@redhat.com>
48 PR tree-optimization/79390
49 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
50 order does not result in usable sequence, retry with reversed operand
56 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
57 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
58 op0 instead of fold_convert_loc (loc, type, arg0).
60 2017-04-12 Jeff Law <law@redhat.com>
62 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
63 has a delay slot in the generated code.
65 * config/cris/cris.md (cris_preferred_reload_class): Return
66 GENNONACR_REGS rather than GENERAL_REGS.
68 2017-04-12 Jakub Jelinek <jakub@redhat.com>
71 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
72 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
73 signedness of the result type.
75 2017-04-12 Richard Biener <rguenther@suse.de>
76 Jeff Law <law@redhat.com>
78 PR tree-optimization/80359
79 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
80 trim stores to TARGET_MEM_REFs.
82 2017-04-12 Richard Biener <rguenther@suse.de>
84 PR tree-optimization/79390
85 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
86 threading case even more.
88 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
91 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
92 for quad_address_p for TImode, instead of just not indexed_address.
94 2017-04-12 Richard Biener <rguenther@suse.de>
95 Bernd Edlinger <bernd.edlinger@hotmail.de>
98 * alias.c (component_uses_parent_alias_set_from): Handle
99 TYPE_TYPELESS_STORAGE.
100 (get_alias_set): Likewise.
101 * tree-core.h (tree_type_common): Add typeless_storage flag.
102 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
103 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
104 for types containing members with TYPE_TYPELESS_STORAGE.
105 (place_field): Likewise.
106 (layout_type): Likewise for ARRAY_TYPE.
107 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
108 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
109 TYPE_TYPELESS_STORAGE.
110 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
112 2017-04-12 Jakub Jelinek <jakub@redhat.com>
115 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
116 first argument to type.
118 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
122 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
123 CONST0_RTX (mode) rather than const0_rtx where appropriate.
124 (rs6000_expand_binop_builtin): Likewise.
125 (rs6000_expand_ternop_builtin): Likewise; also add missing
126 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
128 * doc/extend.texi: Document that vec_xxpermdi's third argument
131 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
133 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
134 Use shift_const cost parameter when calculating gain of STV shifts.
136 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
138 PR rtl-optimization/70478
139 * lra-constraints.c (process_alt_operands): Check memory for
140 disfavoring memory insn operand.
142 2017-04-11 Jakub Jelinek <jakub@redhat.com>
145 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
146 left shift in unsigned HOST_WIDE_INT type.
148 PR rtl-optimization/80385
149 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
150 (not (neg X)) into (plus X -1) for complex or non-integral modes.
153 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
154 if they have any depend clauses.
156 2017-04-11 Martin Liska <mliska@suse.cz>
159 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
160 * ipa-split.c (split_function): Create a local comdat symbol
161 if caller is in a comdat group.
163 2017-04-11 Martin Liska <mliska@suse.cz>
166 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
169 2017-04-11 Martin Sebor <msebor@redhat.com>
172 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
173 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
175 (directive::set_width, directive::set_precision, format_character):
177 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
180 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
183 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
184 conflict, set target->arch_name instead of target->cpu_name.
186 2017-04-11 Richard Biener <rguenther@suse.de>
188 PR tree-optimization/80374
189 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
190 build_zero_cst, remove fold_convertible_p check again.
192 2017-04-11 Martin Liska <mliska@suse.cz>
195 * ubsan.c (instrument_object_size): Do not instrument register
198 2017-04-11 Jakub Jelinek <jakub@redhat.com>
201 * config/i386/i386-builtin-types.def
202 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
203 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
204 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
205 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
206 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
207 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
208 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
209 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
210 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
211 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
212 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
213 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
214 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
215 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
216 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
217 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
218 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
219 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
220 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
221 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
222 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
223 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
224 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
225 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
226 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
227 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
228 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
229 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
230 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
231 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
232 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
233 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
234 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
235 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
236 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
237 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
238 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
239 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
240 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
241 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
242 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
243 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
244 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
245 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
246 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
247 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
249 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
250 flag to second_arg_count, handle 4 argument function type _COUNT
251 aliases, handle second_arg_count on second argument rather than last.
253 2017-04-10 Jeff Law <law@redhat.com>
255 PR tree-optimization/80374
256 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
257 record anything if we can not convert integer_zero_node to the
260 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
263 * config/rs6000/rs6000.c (rs6000_option_override_internal):
264 Enhance special handling given to the TARGET_P9_MINMAX option in
265 relation to certain other options.
267 2017-04-10 Bin Cheng <bin.cheng@arm.com>
269 PR tree-optimization/80153
270 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
271 remove POINTER_PLUS_EXPR's base part directly, rather than through
274 2017-04-10 Richard Biener <rguenther@suse.de>
275 Bin Cheng <bin.cheng@arm.com>
277 PR tree-optimization/80153
278 * tree-affine.c (aff_combination_to_tree): Get base pointer from
279 the first element of pointer type aff_tree. Build result expr in
281 (add_elt_to_tree): Convert to type unconditionally. Remove other
283 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
284 (rewrite_use_nonlinear_expr): Check invariant using iv information.
286 2017-04-10 Richard Biener <rguenther@suse.de>
288 * tree-ssa-structalias.c (find_func_aliases): Properly handle
291 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
293 PR rtl-optimization/70478
294 * lra-constraints.c (curr_small_class_check): New.
295 (update_and_check_small_class_inputs): New.
296 (process_alt_operands): Update curr_small_class_check. Disfavor
297 alternative insn memory operands. Check available regs for small
300 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
303 * config/mips/mips.opt (-mvirt): Update description.
304 * doc/invoke.texi (-mvirt): Likewise.
306 2017-04-10 Richard Biener <rguenther@suse.de>
309 * fold-const.c (fold_binary_loc): Look at unstripped ops when
310 looking for NEGATE_EXPR in -A / -B to A / B folding.
312 2017-04-10 Martin Liska <mliska@suse.cz>
314 PR gcov-profile/80224
315 * gcov.c (print_usage): Fix usage string.
316 (get_gcov_intermediate_filename): Remove.
317 (output_gcov_file): Use both for normal and intermediate format.
318 (generate_results): Do not initialize special file for
321 2017-04-10 Richard Biener <rguenther@suse.de>
323 PR tree-optimization/80304
324 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
327 2017-04-10 Nathan Sidwell <nathan@acm.org>
330 * config/rs6000/rs6000.c (rs6000_vector_type): New.
331 (rs6000_init_builtins): Use it.
333 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
335 * config/arm/arm.md (<mrc>): Add mode to SET source.
338 2017-04-10 Richard Biener <rguenther@suse.de>
341 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
343 2017-04-10 Jakub Jelinek <jakub@redhat.com>
346 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
347 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
348 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
349 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
350 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
351 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
352 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
353 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
354 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
355 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
356 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
357 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
358 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
359 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
360 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
361 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
362 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
363 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
364 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
365 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
366 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
367 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
368 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
370 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
372 PR rtl-optimization/70478
373 * lra-constraints.c: Reverse the last patch.
375 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
377 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
378 Add comment for WCHAR_T.
380 2017-04-08 Martin Liska <mliska@suse.cz>
383 2017-04-07 Martin Liska <mliska@suse.cz>
386 * ipa-split.c (split_function): Add function part to a same comdat
389 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
392 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
394 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
396 * rs6000/rs6000.c (vec_load_pendulum): Rename...
397 (vec_pairing): ...to this.
398 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
399 (rs6000_sched_init): Adjust for name change.
400 (struct rs6000_sched_context): Likewise.
401 (rs6000_init_sched_context): Likewise.
402 (rs6000_set_sched_context): Likewise.
404 2017-04-07 Jakub Jelinek <jakub@redhat.com>
410 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
412 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
413 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
414 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
416 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
418 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
420 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
422 PR rtl-optimization/70703
423 * ira-color.c (update_conflict_hard_regno_costs): Use
424 int64_t instead of HOST_WIDE_INT.
426 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
428 PR rtl-optimization/70478
429 * lra-constraints.c (process_alt_operands): Disfavor alternative
430 insn memory operands.
432 2017-04-07 Jeff Law <law@redhat.com>
434 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
435 CALL and NOTE_INSN_CALL_ARG_LOCATION.
437 2017-04-07 Martin Liska <mliska@suse.cz>
440 * config/aarch64/aarch64.c (aarch64_process_target_attr):
441 Show error message instead of an ICE.
443 2017-04-07 Martin Liska <mliska@suse.cz>
446 * ipa-split.c (split_function): Add function part to a same comdat
449 2017-04-07 Richard Biener <rguenther@suse.de>
452 * tree.c (get_unwidened): Also handle ! for_type case for
454 * convert.c (do_narrow): Split out from ...
455 (convert_to_integer_1): ... here. Do not pass final truncation
456 type to get_unwidened for TRUNC_DIV_EXPR.
458 2017-04-07 Richard Biener <rguenther@suse.de>
460 * tree-affine.c (wide_int_ext_for_comb): Take type rather
462 (aff_combination_const): Adjust.
463 (aff_combination_scale): Likewise.
464 (aff_combination_add_elt): Likewise.
465 (aff_combination_add_cst): Likewise.
466 (aff_combination_convert): Likewise.
467 (add_elt_to_tree): Likewise. Remove unused argument.
468 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
470 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
472 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
474 * config/arm/arm.c (arm_default_short_enums): Use
475 ARM_DEFAULT_SHORT_ENUMS.
476 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
478 2017-04-06 Jakub Jelinek <jakub@redhat.com>
481 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
482 members with redundant out-of-class redeclaration.
484 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
487 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
488 * config/i386/i386.md (*zero_extendsidi2):
489 Add (?*x,*x) and (?*v,*v) alternatives.
491 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
494 * config/i386/i386.c (ix86_expand_builtin)
495 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
496 mode from insn data. Convert operands to insn operand mode.
497 Copy operands that don't satisfy insn predicate to a register.
499 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
501 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
504 2017-04-06 Richard Biener <rguenther@suse.de>
506 PR tree-optimization/80334
507 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
508 preserve alignment of accesses.
510 2017-04-06 Richard Biener <rguenther@suse.de>
512 PR tree-optimization/80262
513 * tree-sra.c (build_ref_for_offset): Preserve address-space
515 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
516 Drop useless address-space information on MEM_REF offsets.
518 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
520 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
522 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
524 PR rtl-optimization/70703
525 * ira-color.c (update_conflict_hard_regno_costs): Use
526 HOST_WIDE_INT instead of long.
528 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
531 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
532 not defined for x86_64 target. Add -mmmx target option when __SSE2__
534 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
535 for x86_64 target. Handle -m3dnowa option.
537 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
539 PR rtl-optimization/70703
540 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
541 (update_conflict_hard_regno_costs): Use long instead of unsigned
542 arithmetic for cost calculation.
544 2017-04-05 Jakub Jelinek <jakub@redhat.com>
545 Bernd Edlinger <bernd.edlinger@hotmail.de>
548 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
551 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
554 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
555 ptr_mode with Pmode throughout.
556 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
557 into probe_stack_range and use DImode.
559 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
562 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
563 call_eh_return is true.
565 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
567 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
568 Initialize last_match_fntype_index.
570 2017-04-05 Jakub Jelinek <jakub@redhat.com>
573 * tree-nvr.c: Include internal-fn.h.
574 (pass_return_slot::execute): Ignore internal calls without
577 2017-04-04 Jakub Jelinek <jakub@redhat.com>
578 Richard Biener <rguenther@suse.de>
581 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
582 captures used multiple times, except for the last use.
583 * generic-match-head.c: Include gimplify.h.
585 2017-04-04 Jakub Jelinek <jakub@redhat.com>
587 PR tree-optimization/79390
588 * target.h (struct noce_if_info): Declare.
589 * targhooks.h (default_noce_conversion_profitable_p): Declare.
590 * target.def (noce_conversion_profitable_p): New target hook.
591 * ifcvt.h (struct noce_if_info): New type, moved from ...
592 * ifcvt.c (struct noce_if_info): ... here.
593 (noce_conversion_profitable_p): Renamed to ...
594 (default_noce_conversion_profitable_p): ... this. No longer
596 (noce_try_store_flag_constants, noce_try_addcc,
597 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
598 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
599 instead of noce_conversion_profitable_p.
600 * config/i386/i386.c: Include ifcvt.h.
601 (ix86_option_override_internal): Don't override
602 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
603 (ix86_noce_conversion_profitable_p): New function.
604 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
605 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
606 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
607 * doc/tm.texi: Regenerated.
609 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
611 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
614 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
617 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
618 instructions for small multiply cores.
620 2017-04-04 Jeff Law <law@redhat.com>
622 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
624 (mips_expand_vec_perm_const): Initialize elements in orig_perm
625 that are not set by the loop over the elements.
627 2017-04-04 Jakub Jelinek <jakub@redhat.com>
630 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
631 int mode, convert_modes it to mode as unsigned, otherwise use
632 lowpart_subreg to mode rather than SImode.
633 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
634 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
635 Use DImode instead of SImode for the shift count operand.
636 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
639 2017-04-04 Richard Biener <rguenther@suse.de>
642 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
643 arithmetic done for the negate or the plus. Simplify.
644 (A - (-B) -> A + B): Likewise.
645 * fold-const.c (split_tree): Make sure to not negate pointers.
647 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
649 PR rtl-optimization/60818
650 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
651 a compare of comparisons with the thing compared if this results
652 in a different machine mode.
654 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
656 * alias.c (base_alias_check): Fix typo in comment.
657 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
658 * cgraphunit.c (symbol_table::compile): Likewise.
659 * collect2.c (maybe_run_lto_and_relink): Likewise.
660 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
661 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
662 * config/avr/avr.c (avr_map_op_t): Likewise.
663 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
664 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
665 * config/epiphany/epiphany.md (movcc): Likewise.
666 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
667 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
669 * config/mips/mips.c (mips_save_restore_reg): Likewise.
670 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
671 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
672 * config/sh/sh.c (sh_rtx_costs): Likewise.
673 * fold-const.c (fold_truth_andor): Likewise.
674 * genautomata.c (collapse_flag): Likewise.
675 * gengtype.h (struct type::u::s): Likewise.
676 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
677 * input.c (FORMAT_AMOUNT): Likewise.
678 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
679 (known_aggs_to_agg_replacement_list): Likewise.
680 * ipa-inline-analysis.c: Likewise.
681 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
682 * ipa-polymorphic-call.c
683 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
684 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
685 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
687 * modulo-sched.c (apply_reg_moves): Likewise.
688 * omp-expand.c (build_omp_regions_1): Likewise.
689 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
690 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
691 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
692 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
693 * value-prof.c: Likewise.
694 * var-tracking.c (val_reset): Likewise.
696 2017-04-03 Richard Biener <rguenther@suse.de>
698 PR tree-optimization/80275
699 * fold-const.c (split_address_to_core_and_offset): Handle
702 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
704 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
705 descriptors is at least equal to that of functions.
707 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
709 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
711 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
714 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
715 (mov<IMOD4:mode>): New expander.
716 (*mov<IMOD4:mode>_internal): New insn and split pattern.
718 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
720 PR rtl-optimization/79405
721 * fwprop.c (propagations_left): New variable.
722 (forward_propagate_into): Decrement it.
723 (fwprop_init): Initialize it.
724 (fw_prop): If the variable has reached zero, stop propagating.
725 (fwprop_addr): Ditto.
727 2017-03-31 Jakub Jelinek <jakub@redhat.com>
730 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
731 a FUNCTION_DECL, pass it as decl instead of origin to
734 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
736 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
739 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
742 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
743 TARGET_VSX_SMALL_INTEGER.
745 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
747 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
748 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
750 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
752 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
753 extraction from odd-numbered MSA register.
755 2017-03-31 Jakub Jelinek <jakub@redhat.com>
758 * expmed.c (store_bit_field_1): Don't attempt to create
759 a word subreg out of hard registers wider than word if they
760 have HARD_REGNO_NREGS of 1 for their mode.
763 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
764 conversions to integer types wider than word and pointer.
767 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
768 (rtx_equal_for_cselib_p): Pass 0 to it.
769 * cselib.c (cselib_hasher::equal): Likewise.
770 (rtx_equal_for_cselib_1): Add depth argument. If depth
771 is 128, don't look up VALUE locs and punt. Increment
772 depth in recursive calls when walking VALUE locs.
774 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
776 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
777 (make_gcov_file_name): Use the canonical path name for generating
779 (read_line): Fix handling of files with ascii null bytes.
781 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
783 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
784 to initialise a vector register instead
785 of using a const_int.
787 2017-03-30 Jakub Jelinek <jakub@redhat.com>
790 * gimplify.c (omp_default_clause): Use %qs instead of %s in
793 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
796 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
797 (dfp_diex_<mode>): Update mode of operand 1.
798 * doc/extend.texi (dxex, dxexq): Document change to return type.
799 (diex, diexq): Document change to argument type.
801 2017-03-30 Martin Jambor <mjambor@suse.cz>
804 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
805 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
806 it reflects the signature changes performed at the callee side.
807 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
808 to cgraph_build_function_type_skip_args.
809 (build_function_decl_skip_args): Adjust call to the above function.
811 2017-03-30 Jakub Jelinek <jakub@redhat.com>
815 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
816 register as dest whenever it is a MEM not rtx_equal_p to the
817 corresponding dup operand, and when forcing into reg move the
818 reg into the memory afterwards.
819 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
820 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
821 for the force_reg mode.
822 (avx512vl_vextractf128<mode>): Use register as dest either
823 always when a MEM, or when it is a MEM not rtx_equal_p to the
824 corresponding dup operand, or even not when it is a CONST_VECTOR
825 depending on the mode and lo vs. hi.
826 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
828 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
829 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
830 Likewise. Require that operands[2] is even.
831 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
832 Remove extraneous parens. Require that operands[2] is a multiple
834 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
835 operands[0] is a MEM if <mask_applied>, the predicates/constraints
836 disallow memory then.
838 2017-03-30 Richard Biener <rguenther@suse.de>
840 PR tree-optimization/77498
841 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
842 to non-constants over backedges.
844 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
846 PR rtl-optimization/80233
847 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
848 as last_combined_insn. Do not test for BARRIER_P separately.
850 2017-03-29 Andreas Schwab <schwab@suse.de>
853 * calls.c (prepare_call_address): Convert funexp to Pmode before
856 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
858 PR tree-optimization/80158
859 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
860 Handle possible future case of more than one alternate
862 (replace_rhs_if_not_dup): Likewise.
863 (replace_one_candidate): Likewise.
865 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
867 PR rtl-optimization/80193
868 * ira.c (ira): Do not check allocation for LRA.
870 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
872 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
873 (nvptx_output_simt_exit): Declare.
874 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
875 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
876 (init_softstack_frame): Move initialization of crtl->is_leaf to...
877 (nvptx_declare_function_name): ...here. Emit declaration of local
878 memory space buffer for omp_simt_enter insn.
879 (nvptx_output_unisimt_switch): New.
880 (nvptx_output_softstack_switch): New.
881 (nvptx_output_simt_enter): New.
882 (nvptx_output_simt_exit): New.
883 * config/nvptx/nvptx.h (struct machine_function): New fields
884 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
885 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
886 (UNSPECV_SIMT_EXIT): Ditto.
887 (omp_simt_enter_insn): New insn.
888 (omp_simt_enter): New expansion.
889 (omp_simt_exit): New insn.
890 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
892 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
893 (expand_GOMP_SIMT_ENTER_ALLOC): New.
894 (expand_GOMP_SIMT_EXIT): New.
895 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
896 (GOMP_SIMT_ENTER_ALLOC): Ditto.
897 (GOMP_SIMT_EXIT): Ditto.
898 * target-insns.def (omp_simt_enter): New insn.
899 (omp_simt_exit): Ditto.
900 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
902 (lower_rec_simd_input_clauses): Implement SIMT privatization.
903 (lower_rec_input_clauses): Likewise.
904 (lower_lastprivate_clauses): Handle SIMT privatization.
906 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
907 (ompdevlow_adjust_simt_enter): New.
908 (find_simtpriv_var_op): New.
909 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
910 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
912 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
913 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
914 (copy_decl_for_dup_finish): Ditto.
916 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
918 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
921 * config/i386/i386.c (ix86_option_override_internal): Always
922 allow -mpreferred-stack-boundary=3 for 64-bit targets.
924 2017-03-28 Bin Cheng <bin.cheng@arm.com>
926 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
928 2017-03-28 Bin Cheng <bin.cheng@arm.com>
930 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
931 mark new edge's irreducible flag accordign to it.
932 (vect_do_peeling): Check loop preheader edge's irreducible flag
933 and pass it to function slpeel_add_loop_guard.
935 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
937 PR tree-optimization/80218
938 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
939 Update block frequencies and counts.
941 2017-03-28 Richard Biener <rguenther@suse.de>
943 PR tree-optimization/78644
944 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
945 of a simplification result we may not use it at all.
947 2017-03-28 Richard Biener <rguenther@suse.de>
950 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
951 without arguments, generate default definition of a SSA name.
953 2017-03-28 Richard Biener <rguenther@suse.de>
956 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
957 TYPE_REF_CAN_ALIAS_ALL references.
958 * fold-const.c (fold_indirect_ref_1): Likewise.
960 2017-03-28 Martin Liska <mliska@suse.cz>
963 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
964 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
966 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
967 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
969 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
970 (EXTRA_SPECS): Define.
971 (SUBTARGET_EXTRA_SPECS): Likewise.
972 (SUBTARGET_CPP_SPEC): Likewise.
973 * config/arc/elf.h (EXTRA_SPECS): Renamed to
974 SUBTARGET_EXTRA_SPECS.
975 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
977 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
979 * config/arc/simdext.md (vst64_insn): Update pattern.
980 (vld32wh_insn): Likewise.
981 (vld32wl_insn): Likewise.
982 (vld64_insn): Likewise.
983 (vld32_insn): Likewise.
985 2017-03-28 Marek Polacek <polacek@redhat.com>
988 * fold-const.c (fold_comparison): Use protected_set_expr_location
989 instead of SET_EXPR_LOCATION.
991 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
993 * tree.c (add_expr): Avoid name lookup warning.
995 2017-03-27 Jeff Law <law@redhat.com>
997 PR tree-optimization/80216
998 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
999 function name. Limit recursion depth.
1000 (record_temporary_equivalences): Corresponding changes.
1002 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
1004 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
1007 2017-03-27 Jakub Jelinek <jakub@redhat.com>
1010 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
1012 * cfgcleanup.c (reg_note_cfa_p): New array.
1013 (insns_have_identical_cfa_notes): New function.
1014 (old_insns_match_p): Don't cross-jump in between /f
1015 and non-/f instructions. If both i1 and i2 are frame related,
1016 verify all CFA notes, their order and content.
1018 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1021 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
1022 HImode and SImode with zero extend to DImode to one insn.
1023 (bswap<mode>2_extenddi): Likewise.
1024 (bswapsi2_extenddi): Likewise.
1025 (bswaphi2_extendsi): Likewise.
1026 (bswaphi2): Combine bswap HImode and SImode into one insn.
1027 Separate memory insns from swapping register.
1028 (bswapsi2): Likewise.
1029 (bswap<mode>2): Likewise.
1030 (bswaphi2_internal): Delete, no longer used.
1031 (bswapsi2_internal): Likewise.
1032 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
1033 store, and gpr<-gpr swap insns.
1034 (bswap<mode>2_store): Likewise.
1035 (bswaphi2_reg): Register only splitter, combine with the splitter.
1036 (bswaphi2 splitter): Likewise.
1037 (bswapsi2_reg): Likewise.
1038 (bswapsi2 splitter): Likewise.
1039 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
1040 the insns into load, store, and register/register insns.
1041 (bswapdi2_ldbrx): Likewise.
1042 (bswapdi2_load): Likewise.
1043 (bswapdi2_store): Likewise.
1044 (bswapdi2_reg): Likewise.
1046 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
1048 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
1049 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
1051 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1054 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
1056 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1057 special handling for target option conflicts between dform
1058 options (-mpower9-dform, -mpower9-dform-vector,
1059 -mpower9-dform-scalar) and -mno-direct-move.
1061 2017-03-27 Richard Biener <rguenther@suse.de>
1063 PR tree-optimization/80181
1064 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
1066 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1068 * config/arc/predicates.md (move_double_src_operand): Replace the
1069 call to move_double_src_operand with a call to address_operand.
1071 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1073 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
1074 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
1075 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
1077 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1079 * config/arc/predicates.md (long_immediate_loadstore_operand):
1080 Consider scaled addresses cases.
1082 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1084 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
1085 restored when in interrupt.
1086 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
1087 doesn't have delay slot.
1089 2017-03-27 Richard Biener <rguenther@suse.de>
1092 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
1093 inlined thunk clones.
1095 2017-03-27 Jakub Jelinek <jakub@redhat.com>
1098 * asan.c (instrument_derefs): Copy over last operand from
1099 original COMPONENT_REF to the new COMPONENT_REF with
1100 DECL_BIT_FIELD_REPRESENTATIVE.
1101 * ubsan.c (instrument_object_size): Likewise.
1103 2017-03-27 Richard Biener <rguenther@suse.de>
1105 PR tree-optimization/80170
1106 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
1107 sure DR/SCEV didnt fold in constants we do not see when looking
1108 at the reference base alignment.
1110 2017-03-27 Richard Biener <rguenther@suse.de>
1113 * gimple-fold.c (fold_ctor_reference): Properly guard against
1114 NULL return value from canonicalize_constructor_val.
1116 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
1119 * config/i386/i386.c (ix86_expand_builtin)
1120 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
1121 flags reg setting and flags reg using instructions.
1122 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
1123 clobbering instructions to zero extend op2.
1125 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
1127 * doc/install.texi (Configuration) <--with-aix-soname>:
1128 Update link to AIX ld.
1130 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
1132 PR rtl-optimization/80160
1133 PR rtl-optimization/80159
1134 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
1135 reg_alternate_class into account.
1137 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
1140 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
1141 to consider in curr_insn_transform.
1143 2017-03-24 Jakub Jelinek <jakub@redhat.com>
1145 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
1146 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
1147 and emit_mode_inner.
1149 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1151 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
1152 argument to the overloaded builtin variants. Use the new flag to
1153 deprecate certain builtin variants.
1154 * config/s390/s390-builtin-types.def: Add new builtin types.
1155 * config/s390/s390-builtins.h: Support new flags field for
1156 overloaded builtins.
1157 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
1158 (s390_macro_to_expand): Enable vector float data type.
1159 (s390_cpu_cpp_builtins_internal): Indicate support of the new
1160 builtins by incrementing the __VEC__ version number.
1161 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
1163 (s390_resolve_overloaded_builtin): Emit error messages depending
1164 on the builtin flags.
1165 * config/s390/s390.c (s390_expand_builtin): Support additional
1166 flags argument. Change error message to match the messages
1167 emitted in s390-c.c.
1168 * config/s390/s390.md: New UNSPEC_* constants.
1169 (op_type): Add new instruction types.
1170 * config/s390/vecintrin.h: Add new builtins and test data class
1172 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
1173 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
1174 (VEC_INEXACT, VEC_NOINEXACT): New constants.
1175 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
1176 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
1177 ("vec_mergel<mode>"): V_HW -> VEC_HW.
1179 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
1180 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
1181 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
1182 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
1184 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
1185 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
1186 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
1187 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
1189 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
1190 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
1191 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
1192 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
1193 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
1194 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
1195 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
1197 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
1198 ("vec_scatter_element<V_HW_4:mode>_DI")
1199 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
1200 ("vec_fpint<mode>", "vflls")
1201 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
1202 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
1203 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
1204 ("*vec_cmphe<mode>_cc"): ... these.
1206 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
1207 mode constant instead of magic value.
1209 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1211 * config/s390/s390.c (s390_expand_vec_compare): Support other
1212 vector floating point modes than just V2DF.
1213 (s390_expand_vcond): Likewise.
1214 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
1215 (s390_cannot_change_mode_class): Prevent mode changes between TF
1216 and V1TF in vector registers.
1217 * config/s390/s390.md (DF, SF): New mode attributes.
1218 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
1219 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
1220 SFmode support for VRs.
1221 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
1223 (VFT, VF_HW): New mode iterators.
1224 (vw, sdx): New mode attributes.
1225 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
1226 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
1227 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
1228 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
1229 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
1230 also the new vector floating point modes. Renaming to ...
1232 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
1233 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
1234 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
1235 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
1236 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
1237 ("vec_unordered<mode>"): ... these.
1239 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
1240 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
1241 ("*vec_extendv2df"): New insn definitions.
1243 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1245 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
1246 ("mulditi3_2", "*muldi3_sign"): New patterns.
1247 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
1248 rename the pattern definition.
1250 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1252 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
1254 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
1256 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1258 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
1259 instruction if possible.
1260 * config/s390/vector.md (vec_halfnumelts): New mode
1262 ("*vec_vllezlf<mode>"): New pattern.
1264 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1266 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
1267 ("popcountv4si2", "popcountv2di2"): Rename to ...
1268 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
1269 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
1271 ("popcount<mode>2_vxe"): New pattern.
1273 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1275 * common/config/s390/s390-common.c (processor_flags_table): Add
1277 * config.gcc: Add arch12.
1278 * config/s390/driver-native.c (s390_host_detect_local_cpu):
1279 Default to arch12 for unknown CPU model numbers.
1280 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
1281 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
1282 PROCESSOR_max sanity check.
1283 * config/s390/s390-opts.h (enum processor_type): Add
1285 * config/s390/s390.c (processor_table): Add arch12.
1286 (s390_expand_builtin): Add check for B_VXE flag.
1287 (s390_issue_rate): Add PROCESSOR_ARCH12.
1288 (s390_get_sched_attrmask): Likewise.
1289 (s390_get_unit_mask): Likewise.
1290 (s390_sched_score): Enable z13 scheduling for arch12.
1291 (s390_sched_reorder): Likewise.
1292 (s390_sched_variable_issue): Likewise.
1293 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
1295 (s390_tune_attr): Use z13 scheduling also for arch12.
1296 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
1297 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
1298 (TARGET_VXE_P): New macros.
1299 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
1300 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
1301 * config/s390/s390.opt: Add arch12 as processor_type.
1303 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1305 * config/s390/s390.md
1306 ("fixuns_truncdddi2", "fixuns_trunctddi2")
1307 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
1308 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
1310 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
1311 Rename expanders to ...
1313 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
1314 ("fixuns_truncdddi2_emu"): ... these.
1316 ("fixuns_trunc<mode>si2_emu"): New expander.
1318 ("*fixuns_truncdfdi2_z13"): Rename to ...
1319 ("*fixuns_truncdfdi2_vx"): ... this.
1321 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1323 * config/s390/2964.md: Remove the single element vector compare
1324 instructions which are no longer used.
1325 * config/s390/s390.c (s390_select_ccmode): Remove handling of
1327 (s390_canonicalize_comparison): Remove handling of DFmode
1329 (s390_expand_vec_compare_scalar): Remove function.
1330 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
1331 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
1333 ("*cmp<mode>_ccs"): Add wfcdb instruction.
1335 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1337 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
1339 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
1340 will anyway by matched by mov<mode>_64dfp.
1342 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1344 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
1345 vlef/vstef. Add missing operand to vleif.
1347 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1349 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
1350 pair for all vector types with 64 bit elements.
1351 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
1352 * config/s390/vector.md (V_HW_64): ... here.
1353 (V_128_NOSINGLE): New mode iterator.
1354 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
1355 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
1356 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
1357 ("*vec_load_pairv2di"): Change to ...
1358 ("*vec_load_pair<mode>"): ... this one.
1360 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1362 * config/s390/constraints.md: Add comments.
1363 (jKK): Reject element sizes > 8 bytes.
1364 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
1366 * config/s390/s390.md: Add the s_operand checks formerly in
1367 s390_split_ok_p to various splitters where they are still
1369 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
1370 for 128 bit vectors. Plus two splitters.
1372 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1374 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
1377 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1380 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
1381 error if the boundary argument is not constant.
1383 2017-03-24 Jakub Jelinek <jakub@redhat.com>
1385 PR rtl-optimization/80112
1386 * loop-doloop.c (doloop_condition_get): Don't check condition
1387 if cmp isn't SET with IF_THEN_ELSE src.
1389 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1391 PR tree-optimization/80158
1392 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
1393 replacing a candidate statement, also replace it for the
1394 candidate's alternate interpretation.
1395 (replace_rhs_if_not_dup): Likewise.
1396 (replace_one_candidate): Likewise.
1398 2017-03-24 Richard Biener <rguenther@suse.de>
1400 PR tree-optimization/80167
1401 * graphite-isl-ast-to-gimple.c
1402 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
1404 (translate_isl_ast_to_gimple::get_rename): Likewise.
1406 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
1408 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
1409 handling of certain combinations of target options, including the
1410 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
1411 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
1413 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1416 * config/arm/arm.md (*load_multiple): Add reload_completed to
1419 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1420 Richard Biener <rguenth@suse.de>
1422 PR tree-optimization/79908
1423 PR tree-optimization/80136
1424 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
1425 been cast away, gimplify_and_add suffices.
1427 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
1429 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
1431 2017-03-23 Richard Biener <rguenther@suse.de>
1433 PR tree-optimization/80032
1434 * gimplify.c (gimple_push_cleanup): Forced unconditional
1435 cleanups still have to go to the conditional_cleanups
1438 2017-03-22 Jakub Jelinek <jakub@redhat.com>
1440 PR tree-optimization/80072
1441 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
1443 (next_operand_entry_id): Change type to unsigned int.
1444 (sort_by_operand_rank): Make sure to return the right return value
1445 even if unsigned fields are bigger than INT_MAX.
1446 (struct oecount): Change cnt and id type to unsigned int.
1447 (oecount_hasher::equal): Formatting fix.
1448 (oecount_cmp): Make sure to return the right return value
1449 even if unsigned fields are bigger than INT_MAX.
1450 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
1453 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
1454 TREE_READONLY on result if writing it more than once.
1457 * doc/invoke.texi (-fsanitize=thread): Document that with
1458 -fnon-call-exceptions atomics are not able to throw
1462 * tsan.c: Include tree-eh.h.
1463 (instrument_builtin_call): Call maybe_clean_eh_stmt or
1464 maybe_clean_or_replace_eh_stmt where needed.
1465 (instrument_memory_accesses): Add cfg_changed argument.
1466 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
1467 if it returned true.
1468 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
1470 PR rtl-optimization/63191
1471 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
1472 wrapper function, moved the whole old content into ...
1473 (ix86_delegitimize_address_1): ... this. New inline function.
1474 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
1475 true as last argument instead of ix86_delegitimize_address.
1477 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
1479 * config/aarch64/aarch64.c (generic_branch_cost): Copy
1480 cortexa57_branch_cost.
1482 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
1484 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
1486 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1489 * doc/md.texi (Constraints): Document wA constraint.
1490 * config/rs6000/constraints.md (wA): New.
1491 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
1492 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
1493 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
1494 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
1496 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
1499 * gimplify.c (is_oacc_declared): New function.
1500 (oacc_default_clause): Use it to set default flags for acc declared
1501 variables inside parallel regions.
1502 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
1504 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
1505 declare attribute to any decl as necessary.
1507 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1510 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
1511 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
1512 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
1513 (arm_arch_lpae): This.
1514 * config/arm/arm.c (arm_arch7ve): Rename into ...
1515 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
1516 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
1519 2017-03-22 Martin Liska <mliska@suse.cz>
1522 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
1523 error message instead of an ICE.
1525 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1527 * doc/extend.texi (6.11 Additional Floating Types): Revise.
1529 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
1531 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
1533 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1536 2017-03-21 Martin Sebor <msebor@redhat.com>
1538 * doc/extend.texi: Use "cannot" instead of "can't."
1539 * doc/hostconfig.texi: Same.
1540 * doc/install.texi: Same.
1541 * doc/invoke.texi: Same.
1542 * doc/loop.texi: Same.
1543 * doc/md.texi: Same.
1544 * doc/objc.texi: Same.
1545 * doc/rtl.texi: Same.
1546 * doc/tm.texi: Same.
1547 * doc/tm.texi.in: Same.
1548 * doc/trouble.texi: Same.
1550 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
1553 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
1554 (collect_checksum_attributes): Set it.
1555 (die_checksum_ordered): Use it.
1557 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1559 PR tree-optimization/79908
1560 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
1561 change: For a VA_ARG whose LHS has been cast away, use
1562 force_gimple_operand to construct the side effects.
1564 2017-03-21 David Malcolm <dmalcolm@redhat.com>
1566 PR translation/80001
1567 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
1568 more amenable to translation.
1569 (oacc_loop_auto_partitions): Likewise.
1571 2017-03-21 Marek Polacek <polacek@redhat.com>
1572 Martin Sebor <msebor@redhat.com>
1574 PR tree-optimization/80109
1575 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
1578 2017-03-21 Jakub Jelinek <jakub@redhat.com>
1579 Segher Boessenkool <segher@kernel.crashing.org>
1582 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
1583 check reg_used_between_p between insn and one of succ or succ2
1584 depending on if succ is artificial insn not inserted into insn
1587 2017-03-21 Martin Liska <mliska@suse.cz>
1589 PR gcov-profile/80081
1590 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
1591 * doc/gcc.texi: Include gcov-dump stuff.
1592 * doc/gcov-dump.texi: New file.
1594 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
1596 PR rtl-optimization/79150
1597 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
1598 conditional jump, if the jump is the last insn of the loop.
1600 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1601 Richard Biener <rguenth@suse.de>
1603 PR tree-optimization/79908
1604 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
1605 been cast away, use force_gimple_operand to construct the side
1608 2017-03-21 Martin Liska <mliska@suse.cz>
1611 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
1614 2017-03-21 Brad Spengler <spender@grsecurity.net>
1617 * plugin.c (htab_hash_plugin): New function.
1618 (add_new_plugin): Use it and adjust.
1619 (parse_plugin_arg_opt): Adjust.
1620 (init_one_plugin): Likewise.
1622 2017-03-21 Richard Biener <rguenther@suse.de>
1624 PR tree-optimization/80032
1625 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
1626 if set force the cleanup to happen unconditionally.
1627 (gimplify_target_expr): Push inserted clobbers with force_uncond
1628 to avoid them being removed by control-dependent DCE.
1630 2017-03-21 Richard Biener <rguenther@suse.de>
1632 PR tree-optimization/80122
1633 * tree-inline.c (copy_bb): Do not expans va-arg packs or
1634 va_arg_pack_len when the inlined call stmt requires pack
1636 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
1638 2017-03-21 Jakub Jelinek <jakub@redhat.com>
1641 * tsan.c (instrument_builtin_call): If the memory model argument
1642 is not a constant, assume it is valid.
1645 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
1648 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
1650 PR rtl-optimization/79910
1651 * combine.c (can_combine_p): Do not allow combining an I0 or I1
1652 if its dest is used by an insn before I2 (other than the combined
1653 insns themselves, which are properly handled already).
1655 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
1658 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
1660 * combine.c (record_used_regs): New static function.
1661 (try_combine): Handle situations where there is an additional
1662 instruction between I2 and I3 which needs to have a LOG_LINK
1666 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
1668 * combine.c (try_combine): Delete redundant i1 test. Call
1669 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
1671 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1674 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
1677 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1679 PR tree-optimization/80054
1680 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
1681 the optimization if a PHI or any of its arguments is not dominated
1682 by the candidate's basis. Use gphi* rather than gimple* as
1684 (replace_profitable_candidates): Clean up a gimple* variable that
1685 should be a gphi* variable.
1687 2017-03-20 Martin Sebor <msebor@redhat.com>
1690 * doc/extend.texi (attribute constructor): Document present limitation.
1692 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
1695 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
1696 __POWER9_VECTOR__ #ifdef control, change template definition to
1697 use Power9-specific built-in function.
1698 (vec_any_eq): Likewise.
1699 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
1700 to control outcomes from this test.
1701 (vector_ae_<mode>p): For VEC_F modes, likewise.
1703 2017-03-20 Ian Lance Taylor <iant@google.com>
1705 * config/i386/i386.c (ix86_function_regparm): Save an extra
1706 register for -fsplit-stack with DECL_STATIC_CHAIN.
1708 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
1711 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
1712 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
1714 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
1716 * config/riscv/riscv.c (riscv_print_operand): Use "fence
1718 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
1721 2017-03-20 Marek Polacek <polacek@redhat.com>
1724 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
1726 2017-03-20 Richard Biener <rguenther@suse.de>
1728 PR tree-optimization/80113
1729 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
1730 allocate extra SSA name for PHI def.
1731 (add_close_phis_to_outer_loops): Likewise.
1732 (add_close_phis_to_merge_points): Likewise.
1733 (copy_loop_close_phi_args): Likewise.
1734 (copy_cond_phi_nodes): Likewise.
1736 2017-03-20 Martin Liska <mliska@suse.cz>
1739 * tree-chkp.c (chkp_build_returned_bound): Do not build
1740 returned bounds for a LHS that's not a BOUNDED_P type.
1742 2017-03-20 Martin Liska <mliska@suse.cz>
1746 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
1747 COMPLEX_CST and VECTOR_CST.
1749 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1752 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
1753 target operand. A new splitter adds the clobber statement in case
1754 the target operand is dead anyway.
1756 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
1758 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
1759 to age-old versions of binutils and glibc.
1761 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
1763 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
1765 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
1767 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
1769 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
1771 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
1772 requirement for binutils 2.13.
1774 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
1776 * combine.c (try_combine): Delete redundant i1 test. Call
1777 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
1779 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
1781 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
1782 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
1784 <riscv64-*-elf>: Re-arrange section
1785 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
1786 <riscv32-*-linux>: Likewise.
1787 <riscv64-*-elf>: Likewise
1788 <riscv64-*-linux>: Likewise.
1790 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
1793 * aarch64.opt(verbose-cost-dump): Fix typo.
1795 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
1798 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
1799 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
1801 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
1803 * reload.c (find_reloads): When reloading a nonoffsettable address,
1804 use RELOAD_OTHER for it and its address reloads.
1806 PR rtl-optimization/79910
1807 * combine.c (record_used_regs): New static function.
1808 (try_combine): Handle situations where there is an additional
1809 instruction between I2 and I3 which needs to have a LOG_LINK
1812 2017-03-17 Jeff Law <law@redhat.com>
1814 PR tree-optimization/71437
1815 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
1816 conditional in the hash table first.
1817 (vrp_dom_walker::before_dom_children): Extract condition from
1818 ASSERT_EXPR. Record condition, its inverion and any implied
1821 2017-03-17 Marek Polacek <polacek@redhat.com>
1822 Markus Trippelsdorf <markus@trippelsdorf.de>
1824 PR tree-optimization/80079
1825 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
1828 2017-03-17 Richard Biener <rguenther@suse.de>
1831 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
1832 Properly verify the LHS before the RHS possibly claims to be
1834 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
1837 2017-03-17 Martin Jambor <mjambor@suse.cz>
1839 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
1840 (List of -O2 options): Likewise.
1841 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
1844 2017-03-17 Tom de Vries <tom@codesourcery.com>
1846 * gcov-dump.c (print_usage): Print bug_report_url.
1848 2017-03-17 Richard Biener <rguenther@suse.de>
1851 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
1852 (parser::peek): Likewise.
1854 2017-03-17 Richard Biener <rguenther@suse.de>
1856 PR tree-optimization/80048
1857 * sese.c (free_sese_info): Properly release rename_map and
1858 copied_bb_map elements.
1860 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
1862 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
1863 Add linked-list forward and backlinks. Insert on
1864 construction, remove on destruction.
1865 (class pass_store_merging): Add m_stores_head field.
1866 (pass_store_merging::terminate_and_process_all_chains):
1867 Iterate over m_stores_head list.
1868 (pass_store_merging::terminate_all_aliasing_chains):
1870 (pass_store_merging::execute): Check for debug stmts first.
1871 Push new chains onto the m_stores_head stack.
1873 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1876 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
1877 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
1878 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
1880 2017-03-16 Jeff Law <law@redhat.com>
1882 PR tree-optimization/71437
1883 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
1884 member function. Implementation moved into after_dom_children
1885 member function and into the threader's thread_outgoing_edges
1887 (dom_opt_dom_walker::after_dom_children): Simplify by moving
1888 some code into new thread_outgoing_edges.
1889 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
1890 definition. Simplify marker handling (do it here). Assume we always
1891 have the available expression and the const/copies tables.
1892 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
1894 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
1895 * tree-vrp.c (equiv_stack): No longer file scoped.
1896 (vrp_dom_walker): New class.
1897 (vrp_dom_walker::before_dom_children): New member function.
1898 (vrp_dom_walker::after_dom_children): Likewise.
1899 (identify_jump_threads): Setup domwalker. Use it rather than
1900 walking edges in a random order by hand. Simplify setup/finalization.
1901 (finalize_jump_threads): Remove.
1902 (vrp_finalize): Do not call identify_jump_threads here.
1903 (execute_vrp): Do it here instead and call thread_through_all_blocks
1906 PR tree-optimization/71437
1907 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
1909 (simplify_stmt_for_jump_threading): Add basic_block argument. All
1911 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
1912 (dom_opt_dom_walker::thread_across_edge): Remove
1913 handle_dominating_asserts argument. All callers changed.
1914 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
1915 changes. Remove calls to lhs_of_dominating_assert. Other
1916 uses of handle_dominating_asserts turn into unconditional code
1917 (simplify_control_stmt_condition_1): Likewise.
1918 (simplify_control_stmt_condition): Likewise.
1919 (thread_through_normal_block, thread_across_edge): Likewise.
1920 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
1921 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
1922 object if it is not an SSA_NAME.
1923 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
1924 before calling into the VRP specific simplifiers.
1925 (identify_jump_threads): Remove handle_dominating_asserts
1928 2017-03-16 Jakub Jelinek <jakub@redhat.com>
1931 * tree-diagnostic.c (default_tree_printer): No longer static.
1932 * tree-diagnostic.h (default_tree_printer): New prototype.
1934 2017-03-16 Tamar Christina <tamar.christina@arm.com>
1936 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
1937 Change ins into fmov.
1939 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1941 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
1942 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
1943 Use h_con constraint for operand 1.
1944 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
1945 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
1947 2017-03-15 Jeff Law <law@redhat.com>
1949 PR tree-optimization/71437
1950 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
1951 (record_temporary_equivalences): Use it.
1953 PR tree-optimization/71437
1954 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
1955 tree-ssa-scopedtables.
1956 (lookup_avail_expr, build_and_record_new_cond): Likewise.
1957 (record_conditions, record_cond, vuse_eq): Likewise.
1958 (record_edge_info): Adjust to API tweak of record_conditions.
1959 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
1960 (record_temporary_equivalences, optimize_stmt): Likewise.
1961 (eliminate_redundant_computations): Likewise.
1962 (record_equivalences_from_stmt): Likewise.
1963 * tree-ssa-scopedtables.c: Include options.h and params.h.
1964 (vuse_eq): New function, moved from tree-ssa-dom.c
1965 (build_and_record_new_cond): Likewise.
1966 (record_conditions): Likewise. Accept vector of conditions rather
1967 than edge_equivalence structure for first argument.
1968 for the first argument.
1969 (avail_exprs_stack::lookup_avail_expr): New member function, moved
1970 from tree-ssa-dom.c.
1971 (avail_exprs_stack::record_cond): Likewise.
1972 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
1973 from tree-ssa-dom.c.
1974 (avail_exprs_stack): Add new member functions lookup_avail_expr
1976 (record_conditions): Declare.
1978 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
1981 * lra-constraints.c (process_alt_operands): Increase reject for
1982 reloading an input/output operand.
1984 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1987 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
1988 insns to convert from signed/unsigned char/short to IEEE 128-bit
1990 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
1992 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
1995 * config/i386/i386.c (ix86_vector_duplicate_value): Create
1996 subreg of inner mode for values already in registers.
1998 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
2000 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
2001 iteration reg is used after the loop.
2003 2017-03-14 Martin Sebor <msebor@redhat.com>
2005 PR tree-optimization/79800
2006 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
2007 precision in negative-positive range.
2008 (format_floating): Call non-const overload with adjusted precision.
2010 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2013 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
2016 2017-03-14 Martin Sebor <msebor@redhat.com>
2019 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
2020 * builtins.def (aligned_alloc): Use it.
2023 * Makefile.in (GTFILES): Add calls.c.
2024 * calls.c: Include "gt-calls.h".
2026 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
2028 PR rtl-optimization/79728
2029 * regs.h (struct target_regs): New field
2030 x_contains_allocatable_regs_of_mode.
2031 (contains_allocatable_regs_of_mode): New macro.
2032 * reginfo.c (init_reg_sets_1): Initialize it, and change
2033 contains_reg_of_mode so it includes global regs as well.
2034 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
2035 rather than contains_regs_of_mode.
2037 2017-03-14 Martin Liska <mliska@suse.cz>
2039 * doc/invoke.texi: Document options that can't be combined with
2040 -fcheck-pointer-bounds.
2042 2017-03-14 Martin Liska <mliska@suse.cz>
2045 * doc/invoke.texi (-Wchkp): Document the option.
2047 2017-03-14 Martin Liska <mliska@suse.cz>
2049 * Makefile.in: Install gcov-dump.
2051 2017-03-14 Martin Liska <mliska@suse.cz>
2053 * multiple_target.c (expand_target_clones): Bail out for
2054 an invalid attribute.
2056 2017-03-14 Richard Biener <rguenther@suse.de>
2058 * alias.c (struct alias_set_entry): Pack properly.
2059 * cfgloop.h (struct loop): Likewise.
2060 * cse.c (struct set): Likewise.
2061 * ipa-utils.c (struct searchc_env): Likewise.
2062 * loop-invariant.c (struct invariant): Likewise.
2063 * lra-remat.c (struct cand): Likewise.
2064 * recog.c (struct change_t): Likewise.
2065 * rtl.h (struct address_info): Likewise.
2066 * symbol-summary.h (function_summary): Likewise.
2067 * tree-loop-distribution.c (struct partition): Likewise.
2068 * tree-object-size.c (struct object_size_info): Likewise.
2069 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
2070 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
2071 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
2072 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
2073 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
2074 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
2075 (struct _stmt_vec_info): Likewise.
2077 2017-03-14 Martin Liska <mliska@suse.cz>
2080 * multiple_target.c (create_dispatcher_calls): Check that
2081 a target can create a function dispatcher.
2083 2017-03-14 Martin Liska <mliska@suse.cz>
2086 * multiple_target.c (expand_target_clones): Drop local.local
2087 flag for default implementation.
2089 2017-03-14 Richard Biener <rguenther@suse.de>
2091 PR tree-optimization/80030
2092 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
2094 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
2096 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
2097 gcc_fallthrough() instead of __attribute__((fallthrough));
2099 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
2101 * doc/gcc.texi: Remove "up" link to (DIR).
2102 * doc/gccint.texi: Ditto.
2104 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
2106 * doc/install.texi (Specific) <avr>: Remove reference to
2109 2017-03-13 Jeff Law <law@redhat.com>
2111 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
2112 attribute rather than comments.
2114 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
2115 match_scratch operand is highest.
2117 2017-03-13 Martin Liska <mliska@suse.cz>
2120 * ipa-pure-const.c (warn_function_noreturn): If the declarations
2121 is a CHKP clone, use original declaration.
2123 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2125 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
2126 (arc_conditional_register_usage): Use a different allocation order
2127 when optimizing for size.
2128 * common/config/arc/arc-common.c (arc_option_optimization_table):
2129 Section anchors default on when optimizing for size.
2131 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2133 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
2135 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2137 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
2138 * config/arc/arc.md (cpu_facility): Add cd variant.
2139 (*movqi_insn): Add code density variant.
2140 (*movhi_insn): Likewise.
2141 (*movqi_insn): Likewise.
2142 (*addsi3_mixed): Likewise.
2143 (subsi3_insn): Likewise.
2145 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2147 * config/arc/arc.md (movsi_cond_exec): Update constraint.
2149 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2151 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
2152 expressions with MINUS and UNARY ops.
2154 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2157 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
2159 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
2160 between vec_select and vector argument.
2161 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
2162 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
2163 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
2164 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
2165 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
2166 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
2168 2017-03-13 Richard Biener <rguenther@suse.de>
2171 * params.def (vect-max-peeling-for-alignment): Fix typo.
2173 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2175 * doc/install.texi (Specific) <mips-*-*>: Remove description of
2176 issue that only occurred with binutils below 2.18.
2178 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2180 * doc/install.texi (Specific) <cris-axis-elf>: No longer
2181 refer to binutils 2.11/2.12 minimum.
2183 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2185 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
2186 ftp.kernel.org and simplify binutils requirement.
2188 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
2190 * doc/invoke.texi (Warning Options): Fix spelling of link-time
2192 (Optimize Options): Ditto. Also remove redundancy.
2194 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2196 PR translation/79848
2197 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
2199 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
2200 to G_ to avoid double translation.
2202 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2204 PR translation/79923
2205 * auto-profile.c (get_combined_location): Convert leading
2206 character of diagnostics to lower case and remove trailing period.
2207 (read_profile): Likewise for various diagnostics.
2208 * config/arm/arm.c (arm_option_override): Remove trailing period
2209 from various diagnostics.
2210 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
2211 (msp430_expand_delay_cycles): Likewise.
2213 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2216 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
2217 full command-line argument, rather than just "str".
2218 (aarch64_validate_march): Likewise.
2219 (aarch64_validate_mtune): Likewise.
2221 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
2223 PR rtl-optimization/78911
2224 * lra-assigns.c (must_not_spill_p): New function.
2225 (spill_for): Use it.
2227 2017-03-10 Jakub Jelinek <jakub@redhat.com>
2229 PR tree-optimization/79981
2230 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
2231 ATOMIC_COMPARE_EXCHANGE ifn result.
2232 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
2233 IFN_ATOMIC_COMPARE_EXCHANGE.
2235 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2238 * opts.c (parse_sanitizer_options): Add missing question mark to
2239 "did you mean" message.
2241 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2243 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
2245 (VMULEUH_UNS): Likewise.
2246 (VMULOUB_UNS): Likewise.
2247 (VMULOUH_UNS): Likewise.
2248 * config/rs6000/rs6000.c (builtin_function_type): Remove
2249 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
2251 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2254 * read-rtl-function.c (function_reader::read_rtx_operand): Update
2255 x with result of extra_parsing_for_operand_code_0.
2256 (function_reader::extra_parsing_for_operand_code_0): Convert
2257 return type from void to rtx, returning x. When reading
2258 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
2259 larger size containing struct block_symbol.
2261 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
2263 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
2264 -mfloat128-hardware without -m64.
2266 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
2269 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
2270 entries to the case statement that marks unsigned arguments to
2271 overloaded functions.
2273 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
2275 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
2276 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
2278 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
2281 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
2282 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
2284 2017-03-10 Martin Liska <mliska@suse.cz>
2288 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
2289 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
2292 2017-03-10 Olivier Hainque <hainque@adacore.com>
2294 * tree-switch-conversion (array_value_type): Start by resetting
2295 candidate type to it's main variant.
2297 2017-03-10 Jakub Jelinek <jakub@redhat.com>
2299 PR rtl-optimization/79909
2300 * combine.c (try_combine): Use simplify_replace_rtx on individual
2301 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
2302 of the whole CALL_INSN_FUNCTION_USAGE.
2304 PR tree-optimization/79972
2305 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
2306 get_range_info on SSA_NAMEs. Formatting fixes.
2308 2017-03-10 Richard Biener <rguenther@suse.de>
2309 Jakub Jelinek <jakub@redhat.com>
2311 PR tree-optimization/77975
2312 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
2313 edge to be constant.
2314 (get_val_for): For constant x return it. Formatting fix.
2315 (loop_niter_by_eval): Avoid pointless looping if the next iteration
2316 would use the same bases as the current one.
2318 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2320 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
2321 instead of vec_select for V1TImode.
2322 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
2324 (VSX_LE_128): Add V1TI to this mode iterator.
2325 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
2326 (*vsx_le_perm_store_<mode>): Likewise.
2327 (pre-reload splitter for VSX stores): Likewise.
2328 (post-reload splitter for VSX stores): Likewise.
2329 (*vsx_xxpermdi2_le_<mode>): Likewise.
2330 (*vsx_lxvd2x2_le_<mode>): Likewise.
2331 (*vsx_stxvd2x2_le_<mode>): Likewise.
2333 2017-03-09 Michael Eager <eager@eagercon.com>
2335 Correct failures with --enable-checking=yes,rtl.
2337 * config/microblaze/microblaze.c (microblaze_expand_shift):
2338 Replace GET_CODE test with CONST_INT_P and INTVAL test with
2339 test for const0_rtx.
2340 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
2341 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
2343 2017-03-09 Richard Biener <rguenther@suse.de>
2345 PR tree-optimization/79977
2346 * graphite-scop-detection.c (scop_detection::merge_sese):
2347 Handle the case of extra exits to blocks dominating the entry.
2349 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
2351 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2354 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
2356 PR rtl-optimization/79949
2357 * lra-constraints.c (process_alt_operands): Check memory when
2358 trying to predict a cycle. Print about the overall increase.
2360 2017-03-09 Richard Biener <rguenther@suse.de>
2363 * gimple-expr.c (useless_type_conversion_p): Preserve
2364 TYPE_SATURATING for fixed-point types.
2366 2017-03-09 Richard Biener <rguenther@suse.de>
2369 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
2370 alignment of BLKmode params.
2372 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2375 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
2376 (VALL_NO_V2Q): Likewise.
2378 * config/aarch64/aarch64-simd.md
2379 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
2381 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
2382 VALL_NO_V2Q mode iterator.
2383 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
2385 2017-03-09 Martin Liska <mliska@suse.cz>
2387 PR tree-optimization/79631
2388 * tree-chkp-opt.c (chkp_is_constant_addr): Call
2389 tree_int_cst_sign_bit just for INTEGER constants.
2391 2017-03-09 Martin Liska <mliska@suse.cz>
2395 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
2398 2017-03-09 Marek Polacek <polacek@redhat.com>
2401 * tree.c (inchash::add_expr): Handle TREE_VEC.
2403 2017-03-09 Martin Liska <mliska@suse.cz>
2406 (chkp_narrow_size_and_offset): New function.
2407 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
2408 (void chkp_parse_bit_field_ref): New function.
2409 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
2410 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
2412 2017-03-09 Martin Liska <mliska@suse.cz>
2415 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
2416 (chkp_find_bounds_1): Remove gcc_unreachable.
2418 2017-03-09 Jakub Jelinek <jakub@redhat.com>
2421 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
2422 BUILT_IN_SYNC*, determine the access type from the size suffix and
2423 always build a MEM_REF with that type. Handle forgotten
2424 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
2427 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
2428 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
2429 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
2430 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
2431 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
2432 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
2433 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
2434 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
2435 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
2436 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
2437 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
2438 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
2439 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
2440 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
2441 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
2442 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
2443 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
2444 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
2445 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
2446 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
2447 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
2448 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
2449 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
2450 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
2451 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
2452 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
2453 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
2454 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
2455 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
2456 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
2457 definitions outside of __OPTIMIZE__ guarded section.
2460 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
2461 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
2462 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
2463 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
2466 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2468 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
2469 ("vfenez<mode>"): Add missing constraints.
2471 2017-03-08 Martin Sebor <msebor@redhat.com>
2474 * config/nds32/nds32.c (nds32_option_override):
2475 Fix misspelled diagnostic.
2477 2017-03-08 Jakub Jelinek <jakub@redhat.com>
2480 * gimplify.c (gimplify_omp_for): Replace index var in outer
2481 taskloop statement with an artificial variable and add
2482 OMP_CLAUSE_PRIVATE clause for it.
2484 2017-03-08 Richard Biener <rguenther@suse.de>
2486 PR tree-optimization/79955
2487 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
2488 for accesses that are completely outside of the variable.
2490 2017-03-08 Andrew Haley <aph@redhat.com>
2492 PR tree-optimization/79943
2493 * tree-ssa-loop-split.c (compute_new_first_bound): When
2494 calculating the new upper bound, (END-BEG) should be added, not
2497 2017-03-08 Jakub Jelinek <jakub@redhat.com>
2499 * config/avr/avr.md (setmemhi): Make sure match_dup
2500 operand number comes before match_scratch.
2502 2017-03-08 Richard Biener <rguenther@suse.de>
2504 PR tree-optimization/79920
2505 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
2506 with ncopies == 1 to ...
2507 (vect_transform_slp_perm_load): ... here. Properly compute
2508 all element loads by iterating VF times over the group. Do
2509 not handle ncopies (computed in a broken way) in
2510 vect_create_mask_and_perm.
2512 2017-03-08 Jakub Jelinek <jakub@redhat.com>
2515 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
2516 is a uniform vector, use uniform_vector_p return value instead of
2517 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
2519 2017-03-07 Marek Polacek <polacek@redhat.com>
2522 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
2523 (alloca_call_type): Likewise.
2525 2017-03-07 Martin Liska <mliska@suse.cz>
2527 * gcov.c (process_args): Put comment to correct location.
2529 2017-03-07 Martin Liska <mliska@suse.cz>
2532 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
2533 Use array_at_struct_end_p instead of DECL_CHAIN (field).
2534 (chkp_narrow_bounds_for_field): Likewise.
2535 (chkp_parse_array_and_component_ref): Pass one more argument to
2538 2017-03-07 Richard Biener <rguenther@suse.de>
2540 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
2543 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
2545 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
2546 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
2548 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2551 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
2552 to end of description.
2553 (PARAM_MAX_STORES_TO_MERGE): Likewise.
2555 2017-03-07 Jakub Jelinek <jakub@redhat.com>
2557 PR rtl-optimization/79901
2558 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
2560 (*avx512f_<code><mode>3<mask_name>): ... this.
2561 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
2562 iterator instead of VI8_AVX2_AVX512BW.
2564 PR rtl-optimization/79901
2565 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
2566 min/max expander, expand it using expand_vec_cond_expr.
2569 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
2572 2017-03-06 Jakub Jelinek <jakub@redhat.com>
2575 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
2576 to void * for PCH reasons.
2577 * dwarf2out.c (output_loc_operands, output_die): Cast
2578 v.val_vec.array to unsigned char *.
2580 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
2583 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
2586 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
2588 PR rtl-optimization/79571
2589 * lra-constraints.c (process_alt_operands): Calculate static
2590 reject and subtract it from overall when only addresses will be
2593 2017-03-06 Julia Koval <julia.koval@intel.com>
2596 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
2597 incoming stack boundary to 128 for 64-bit targets.
2599 2017-03-06 Richard Biener <rguenther@suse.de>
2601 PR tree-optimization/79894
2602 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
2603 to NULL after folding it.
2605 2017-03-06 Richard Biener <rguenther@suse.de>
2607 PR tree-optimization/79824
2608 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
2609 check disabling peeling for gaps.
2611 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
2613 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
2614 attributes): Document gettimeofday.
2616 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
2618 * config/s390/s390.c (s390_option_override_internal): Set
2619 PARAM_MIN_VECT_LOOP_BOUND
2621 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
2623 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
2624 * config/s390/s390.md: Likewise.
2626 2017-03-06 Jakub Jelinek <jakub@redhat.com>
2629 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
2630 (<avx2_avx512>_perm<mode>): Rename to ...
2631 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
2633 (<avx512>_perm<mode>_mask): Rename to ...
2634 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
2636 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
2637 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
2638 instead of VI8F_256_512.
2639 (avx512f_perm<mode>): New define_expand.
2640 (avx512f_perm<mode>_mask): Likewise.
2641 (avx512f_perm<mode>_1<mask_name>): New define_insn.
2642 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
2644 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
2646 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
2647 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
2649 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
2651 2017-03-06 Martin Liska <mliska@suse.cz>
2654 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
2655 when having a SSA NAME w/o VAR_DECL assigned to it.
2657 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
2659 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
2660 msa_dpsub_<su>_d): Fix MODE for vec_select.
2662 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
2664 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
2666 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
2668 2017-03-06 Richard Biener <rguenther@suse.de>
2670 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
2671 * plugin.c (register_plugin_info): Likewise.
2672 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
2674 2017-03-05 Jakub Jelinek <jakub@redhat.com>
2676 * config/i386/sse.md (sse_storehps, sse_storelps,
2677 avx_<castmode><avxsizesuffix>_<castmode>,
2678 avx512f_<castmode><avxsizesuffix>_<castmode>,
2679 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
2680 in condition that at least one operand is not a MEM.
2682 2017-03-03 Jakub Jelinek <jakub@redhat.com>
2685 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
2686 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
2688 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
2689 gimple_call_nothrow_p flag based on whether original builtin can throw.
2690 If it can, emit following stmts on the fallthrough edge.
2691 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
2692 don't create new bb if inserting just debug stmts on the edge, try to
2693 insert them on the fallthru bb or just reset debug stmts.
2695 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
2698 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
2699 restore recog_data (including the operand rtxes inside it) around
2700 the call to get_insn_template.
2702 2017-03-03 Martin Sebor <msebor@redhat.com>
2704 PR tree-optimization/79699
2705 * context.c (context::~context): Free MPFR caches to avoid
2706 a memory leak on program exit.
2708 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2710 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
2711 Use wide_int::ulow () instead of .elt (0).
2713 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
2715 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
2716 (*pushxf): Limit oF constraint to 32bit targets and add oC
2717 constraint for 64bit targets.
2718 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
2719 (*pushdf): Change rmF constraint to rmC.
2721 2017-03-03 Martin Liska <mliska@suse.cz>
2723 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
2724 Remove unused variable.
2726 2017-03-03 Jakub Jelinek <jakub@redhat.com>
2729 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
2730 is a memory operand, increase num_memory.
2731 (ix86_expand_args_builtin): Likewise.
2733 2017-03-03 Jan Hubicka <jh@suse.cz>
2736 * ipa-devirt.c (maybe_record_node): Properly handle
2737 __cxa_pure_virtual visibility.
2739 2017-03-03 Martin Liska <mliska@suse.cz>
2741 PR tree-optimization/79803
2742 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
2744 (pass_loop_prefetch::execute): Disabled optimization if an
2745 assumption about L1 cache size is not met.
2747 2017-03-03 Martin Liska <mliska@suse.cz>
2749 PR rtl-optimization/79574
2750 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
2751 (hash_scan_set): Likewise.
2752 (dump_hash_table): Likewise.
2753 (hoist_code): Likewise.
2755 2017-03-03 Richard Biener <rguenther@suse.de>
2757 * fixed-value.c (fixed_from_string): Restore use of elt (1)
2758 in place of uhigh ().
2759 (fixed_convert_from_real): Likewise.
2761 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
2764 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
2766 2017-03-03 Richard Biener <rguenther@suse.de>
2769 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
2770 TYPE_OVERFLOW_UNDEFINED check.
2772 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2774 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
2776 (vector_ae_<mode>_p): Likewise.
2777 (vector_nez_<mode>_p): Likewise.
2778 (vector_ne_v2di_p): Likewise.
2779 (vector_ae_v2di_p): Likewise.
2780 (vector_ne_<mode>_p): Likewise.
2781 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
2783 (vsx_tsqrt<mode>2_fe): Likewise.
2785 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
2788 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
2790 2017-03-02 Jakub Jelinek <jakub@redhat.com>
2792 PR rtl-optimization/79780
2793 * cprop.c (one_cprop_pass): When second and further conditional trap
2794 in a single basic block is turned into an unconditional trap, turn it
2795 into a deleted note to avoid RTL verification failures.
2797 2017-03-02 Richard Biener <rguenther@suse.de>
2799 * fold-const.c (const_binop): Use ulow () instead of elt (0).
2801 2017-03-02 Richard Biener <rguenther@suse.de>
2803 PR tree-optimization/79345
2805 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
2806 param and abort the walk, returning -1 if it is hit.
2807 (walk_aliased_vdefs): Take a limit param and pass it on.
2808 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
2809 defaulting to 0 and return a signed int.
2810 * tree-ssa-uninit.c (struct check_defs_data): New struct.
2811 (check_defs): New helper.
2812 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
2813 about uninitialized memory.
2814 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
2815 bogus uninitialized warning.
2816 (fixed_convert_from_real): Likewise.
2818 2017-03-02 Bin Cheng <bin.cheng@arm.com>
2820 PR tree-optimization/66768
2821 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
2822 iv_use if base object can't be determined.
2824 2017-03-02 Jakub Jelinek <jakub@redhat.com>
2826 PR tree-optimization/79345
2827 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
2828 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
2829 (get_pattern_stats): Initialize it.
2830 * genemit.c (gen_expand): Verify match_scratch numbers come after
2831 match_operand/match_dup numbers.
2832 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
2833 match_scratch numbers.
2834 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
2836 * config/s390/s390.md (trunctdsd2): Likewise.
2838 2017-03-02 Richard Biener <rguenther@suse.de>
2840 * wide-int.h (wide_int_storage::operator=): Implement in terms
2843 2017-03-02 Richard Biener <rguenther@suse.de>
2845 PR tree-optimization/79777
2846 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
2847 the to insert expression to sth existing.
2849 2017-03-01 Martin Sebor <msebor@redhat.com>
2852 * gimple-ssa-sprintf.c
2853 (directive::known_width_and_precision): New function.
2854 (format_integer): Use it.
2855 (get_mpfr_format_length): Consider the full range of precision
2856 when computing %g output with the # flag. Set the likely byte
2857 count to 3 rather than 1 when precision is indeterminate.
2858 (format_floating): Correct the lower bound of precision.
2860 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2862 * doc/invoke.texi: Document default code model for 64-bit Linux.
2864 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2867 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
2868 udiv rather than div since input pattern is unsigned.
2870 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
2872 * config/i386/i386.c (print_reg): Warn for values of
2873 unsupported size in integer register.
2875 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2878 * config/rs6000/predicates.md (current_file_function_operand): Do
2879 not allow self calls to be local if the function is replaceable.
2881 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
2884 * config/rs6000/altivec.h (vec_ctz and others): Change the
2885 preprocessor macro that controls conditional compilation from
2886 _ARCH_PWR9 to __POWER9_VECTOR__.
2887 (vec_all_ne): Change parameterization of __altivec_scalar_pred
2888 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
2889 control (instead of _ARCH_PWR9 control) so that template
2890 definition uses power9-specific function.
2891 (vec_any_eq): Likewise.
2892 (vec_all_ne): Change macro definition to use a power9-specific
2893 expansion under #ifdef __POWER9_VECTOR__ control (instead of
2894 _ARCH_PWR9 control).
2895 (vec_any_eq) Likewise.
2896 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
2897 expansion for CMPNEF to remove support for xvcmpnesp instruction.
2898 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
2899 support for xvcmpnedp instruction.
2900 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
2901 macro expansion so that Power9 implementation of vec_all_ne does
2902 not use the AltiVec predicate framework.
2903 (VCMPNEH_P): Likewise.
2904 (VCMPNEW_P): Likewise.
2905 (VCMPNED_P): Likewise.
2906 (VCMPNEFP_P): Likewise.
2907 (VCMPNEDP_P): Likewise.
2908 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
2909 implementation of vec_any_eq to not use AltiVec predicate
2911 (VCMPAEH_P): Likewise.
2912 (VCMPAEW_P): Likewise.
2913 (VCMPAED_P): Likewise.
2914 (VCMPAEFP_P): Likewise.
2915 (VCMPAEDP_P): Likewise.
2916 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
2917 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
2918 not use the AltiVec predicate framework.
2919 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
2920 of vec_any_eq to not use AltiVec predicate framework.
2921 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
2922 support for predefined __POWER9_VECTOR__ macro to indicate that
2923 Power9 instruction selection is enabled.
2924 (altivec_overloaded_builtins): Remove extraneous
2925 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
2926 function argument types RS6000_BTI_bool_V16QI and
2927 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
2928 entry for overloaded function argument types RS6000_BTI_bool_V4SI
2929 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
2930 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
2931 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
2932 Power9 for implementations of vec_cmpne. Change the signature for
2933 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
2934 (representing vec_all_ne) to remove the previously described first
2935 argument of type RS6000_BTI_INTSI, as this was an artifact of
2936 reliance on the AltiVec predicate framework, which is no longer
2937 used in the implementation of these functions. Add
2938 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
2939 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
2940 since, unlike the AltiVec predicate framework implementation, we
2941 do not share function descriptors between vec_alle and vec_anyeq.
2942 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
2943 set of modes that receive special treatment even when
2944 TARGET_P9_VECTOR is true. The special treatment emits code that
2945 does not depend on Power9 instructions.
2946 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
2947 define_expand to not rely on AltiVec predicate framework.
2948 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
2950 (vector_ne_v2di_p): Change this define_expand to not rely on
2951 AltiVec predicate framework.
2952 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
2954 (vector_ne_<mode>_p): Change this define_expand to not rely on
2955 AltiVec predicate framework.
2956 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
2958 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
2959 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
2960 define_insn pattern.
2961 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
2962 define_insn pattern because the xvcmpne<VSs>. instruction is not
2964 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
2965 instruction is not supported.
2967 2017-03-01 Jakub Jelinek <jakub@redhat.com>
2969 * config/nvptx/nvptx.c: Include intl.h.
2971 2017-03-01 Martin Jambor <mjambor@suse.cz>
2974 * ipa-prop.h (ipa_bits): Removed field known.
2975 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
2976 to pointers. Adjusted their comments to warn about their sharing.
2977 (ipcp_transformation_summary): Change bits to a vector of pointers.
2978 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
2979 (ipa_get_ipa_bits_for_value): Declare.
2980 * tree-vrp.h (value_range): Mark as GTY((for_user)).
2981 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
2982 (ipa_bits_hash_table): Likewise.
2983 (ipa_vr_ggc_hash_traits): Likewise.
2984 (ipa_vr_hash_table): Likewise.
2985 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
2986 being pointers and vr_known being removed.
2987 (ipa_set_jf_unknown): Likewise.
2988 (ipa_get_ipa_bits_for_value): New function.
2989 (ipa_set_jfunc_bits): Likewise.
2990 (ipa_get_value_range): New overloaded functions.
2991 (ipa_set_jfunc_vr): Likewise.
2992 (ipa_compute_jump_functions_for_edge): Use the above functions to
2993 construct bits and vr parts of jump functions.
2994 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
2995 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
2997 (ipcp_grow_transformations_if_necessary): Also allocate
2998 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
3000 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
3001 them. Fix too long lines.
3002 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
3003 vr_known being removed.
3004 (ipa_read_jump_function): Use new setter functions to construct bits
3005 and vr parts of jump functions or set them to NULL.
3006 (write_ipcp_transformation_info): Adjust for bits being pointers.
3007 (read_ipcp_transformation_info): Likewise.
3008 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
3010 Include gt-ipa-prop.h.
3011 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
3013 (ipcp_store_bits_results): Likewise.
3014 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
3015 Do not write to existing jump functions but use a temporary instead.
3017 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3020 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
3021 attempt to use its first operand as BIT_FIELD_REF base.
3023 2017-03-01 Richard Biener <rguenther@suse.de>
3026 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
3027 interpolating formula in wrapping arithmetic.
3028 (chrec_apply): Convert chrec_evaluate return value to wanted type.
3030 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3032 PR tree-optimization/79734
3033 * tree-vect-generic.c (expand_vector_condition): Optimize
3034 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
3035 Handle VEC_COND_EXPR where comparison has different inner width from
3038 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
3040 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
3041 markup, and similar issues. Remove @opindex entries for things
3042 that aren't options. Add missing -mmpy-option entries.
3044 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3046 PR tree-optimization/79737
3047 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
3048 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
3049 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
3050 instead of byte_size. Formatting fix.
3051 (shift_bytes_in_array_right): Formatting fix.
3053 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
3056 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
3057 condition on optimize for the leaf function test.
3059 2017-02-28 Martin Liska <mliska@suse.cz>
3062 * read-rtl-function.c (function_reader::handle_unknown_directive):
3063 Bail out when one uses -flto.
3065 2017-02-28 Martin Liska <mliska@suse.cz>
3067 * common.opt: Replace space with tabular for options of <number>
3069 * config/i386/i386.opt: Show <number> value for
3070 -mlarge-data-threshold.
3071 * opts.c (print_filtered_help): Do not display number in hexadecimal
3074 2017-02-28 Martin Liska <mliska@suse.cz>
3076 * common.opt: Fix --help=option -Q for options which are of
3079 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
3081 * config/i386/i386.c (print_reg): Error out for values
3082 of 8-bit size in invalid integer register.
3084 2017-02-28 Martin Sebor <msebor@redhat.com>
3086 PR tree-optimization/79691
3087 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
3089 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3092 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
3093 gcc_unreachable with output_operand_lossage.
3095 2017-02-28 Richard Biener <rguenther@suse.de>
3097 PR tree-optimization/79740
3098 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
3100 (visit_nary_op): Insert the nary into the hashtable if we
3101 pattern-matched sth.
3102 * tree-ssa-pre.c (eliminate_insert): Robustify.
3104 2017-02-28 Richard Biener <rguenther@suse.de>
3107 * fold-const.c (decode_field_reference): Reject out-of-bound
3110 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3112 * config/i386/i386.c: Include intl.h.
3113 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
3114 instead of just cond ? "..." : "...".
3115 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
3116 * coverage.c (read_counts_file): Likewise.
3117 * omp-offload.c: Include intl.h.
3118 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
3119 of just cond ? "..." : "...".
3120 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
3121 of just cond ? "..." : "...".
3123 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
3126 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
3128 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
3129 'tune for' CPU name.
3130 * config/arm/arm-cpu-data.h: Regenerated.
3132 2017-02-28 Richard Biener <rguenther@suse.de>
3134 PR tree-optimization/79732
3135 * tree-inline.c (expand_call_inline): Do not shadow var.
3137 2017-02-28 Richard Biener <rguenther@suse.de>
3139 PR tree-optimization/79723
3140 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
3141 address-space properly.
3143 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
3145 * doc/optinfo.texi (Optimization groups): Fix option used for
3147 * doc/invoke.texi (-fopt-info): Document "omp".
3148 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
3149 (OPTGROUP_ALL): Add OPTGROUP_OMP.
3150 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
3151 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
3152 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
3154 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
3156 * dumpfile.c (optgroup_options): Instead of "openmp", associate
3157 OPTGROUP_OMP with "omp".
3159 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
3162 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
3163 for arithmetic shift of unsigned V2DI.
3165 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
3167 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
3168 arc/linux.h headers.
3169 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
3170 (LINK_SPEC): Likewise.
3171 (ARC_TLS_EXTRA_START_SPEC): Likewise.
3172 (EXTRA_SPECS): Likewise.
3173 (STARTFILE_SPEC): Likewise.
3174 (ENDFILE_SPEC): Likewise.
3175 (LIB_SPEC): Likewise.
3176 (TARGET_SDATA_DEFAULT): Likewise.
3177 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
3178 (MULTILIB_DEFAULTS): Likewise.
3179 (DWARF2_UNWIND_INFO): Likewise.
3180 * config/arc/big.h: New file.
3181 * config/arc/elf.h: Likewise.
3182 * config/arc/linux.h: Likewise.
3183 * config/arc/t-uClibc: Remove.
3185 2017-02-27 Bin Cheng <bin.cheng@arm.com>
3187 PR tree-optimization/77536
3188 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
3189 (tree_transform_and_unroll_loop): Use above function to compute the
3190 estimated niter of unrolled loop and use it when scaling profile.
3191 Also use count info rather than frequency if it's non-zero.
3192 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
3193 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
3194 (vect_transform_loop): Call above function.
3196 2017-02-27 Richard Biener <rguenther@suse.de>
3198 PR tree-optimization/45397
3199 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
3200 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
3201 (visit_nary_op): Add pattern matching for CSEing sign-changed
3202 or truncated operations with wider ones.
3204 2017-02-27 Richard Biener <rguenther@suse.de>
3206 PR tree-optimization/79690
3207 * tree-vect-stmts.c (vectorizable_store): Use vector type
3208 built from the DR with address-space.
3210 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
3212 * doc/invoke.texi (Optimize Options): Refine the description
3213 of asan-use-after-return.
3215 2017-02-25 Alan Modra <amodra@gmail.com>
3217 PR rtl-optimization/79584
3218 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
3219 base, not ad->base_term, the reg within base. Remove assertion
3220 that ad->base == ad->base_term. Replace gen_int_mode using
3221 bogus mode with const0_rtx.
3223 2017-02-25 Jakub Jelinek <jakub@redhat.com>
3226 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
3227 FMA_EXPR like tcc_binary or tcc_unary.
3229 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
3232 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
3234 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
3235 (output_loc_operands): Handle DW_OP_call_ref and
3236 DW_OP_GNU_variable_value.
3237 (struct variable_value_struct): New type.
3238 (struct variable_value_hasher): Likewise.
3239 (variable_value_hash): New variable.
3240 (string_types): Remove.
3241 (copy_loc_descr): New function.
3242 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
3243 (prepend_loc_descr_to_each): New function.
3244 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
3245 instead of add_loc_descr_to_each if the first argument is single
3246 location list and the second has multiple.
3247 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
3248 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
3249 when looking for variable value which doesn't have other location info.
3250 (loc_list_from_tree): Formatting fix.
3251 (gen_array_type_die): Simplify DW_AT_string_length handling.
3252 (adjust_string_types): Remove.
3253 (gen_subprogram_die): Don't call adjust_string_types nor test/set
3254 string_types. Call resolve_variable_values.
3255 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
3256 (resolve_addr_in_expr): Likewise. Add A argument.
3257 (copy_deref_exprloc): Remove deref argument. Adjust for the
3258 original expression being DW_OP_GNU_variable_value with optionally
3259 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
3260 optionally after it.
3261 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
3262 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
3263 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
3264 (variable_value_hasher::hash, variable_value_hasher::equal): New
3266 (resolve_variable_value_in_expr, resolve_variable_value,
3267 resolve_variable_values, note_variable_value_in_expr,
3268 note_variable_value): New functions.
3269 (dwarf2out_early_finish): Call note_variable_value on all toplevel
3272 2017-02-24 Jakub Jelinek <jakub@redhat.com>
3275 * opts.h (handle_generated_option): Add GENERATED_P argument.
3276 * opts-common.c (handle_option): Adjust function comment.
3277 (handle_generated_option): Add GENERATED_P argument, pass it to
3279 (control_warning_option): Pass false to handle_generated_option
3281 * opts.c (maybe_default_option): Pass true to handle_generated_option
3283 * optc-gen.awk: Likewise.
3285 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3287 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
3288 a REG, look at the REG it is a SUBREG of.
3289 (splitter for cmpeqsi_t): Ditto.
3291 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3293 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
3294 the special USEs with the pattern of the insn, not the insn itself.
3296 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
3299 * doc/invoke.texi: Document -mload-store-pairs.
3301 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3302 Sandra Loosemore <sandra@codesourcery.com>
3304 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
3305 argument isn't a CONST_INT.
3306 (nios2_alternate_compare_const): Assert op is a CONST_INT.
3307 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
3308 (nios2_validate_compare): Bypass alternate compare logic if *op2
3310 (ldstwm_operation_p): Return false if first_base is not a REG or
3311 if first_offset is not a CONST_INT.
3313 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3315 * config/cris/cris.md: Use correct operand in a define_peephole2.
3317 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3319 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
3321 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3323 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
3324 this_insn if it is an INSN or JUMP_INSN.
3325 (force_offsettable): Look at base, not at addr.
3326 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
3327 on things that aren't necessarily CONST_INTs.
3329 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
3331 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
3332 -mfpmath=sse is the default also for x86-32 targets with SSE2
3333 instruction set when @option{-ffast-math} is enabled
3335 2017-02-24 Jeff Law <law@redhat.com>
3337 PR rtl-optimizatoin/79286
3338 * ira.c (update_equiv_regs): Drop may_trap_p exception to
3341 2017-02-24 Richard Biener <rguenther@suse.de>
3343 PR tree-optimization/79389
3344 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
3347 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
3349 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
3350 function comment to reflect reality.
3351 (loop_exits_before_overflow): Fix typo in function description.
3353 2017-02-24 Richard Biener <rguenther@suse.de>
3355 PR tree-optimization/79389
3356 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
3357 properly that a threading opportunity exists. Detect conditional
3358 copy/constant propagation opportunities.
3360 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
3362 * config/visium/visium.md (type): Add trap.
3363 (b): New mode attribute.
3364 (*btst): Rename into...
3365 (*btst<mode>): ...this and adjust.
3366 (*cbranchsi4_btst_insn): Rename into...
3367 (*cbranch<mode>4_btst_insn): ...this and adjust.
3368 (trap): New define_insn.
3370 2017-02-23 Jakub Jelinek <jakub@redhat.com>
3372 PR tree-optimization/79389
3373 * ifcvt.c (struct noce_if_info): Add rev_cond field.
3374 (noce_reversed_cond_code): New function.
3375 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
3376 reversed_comparison_code. Formatting fix.
3377 (noce_try_store_flag): Test rev_cond != NULL in addition to
3378 reversed_comparison_code.
3379 (noce_try_store_flag_constants): Likewise.
3380 (noce_try_store_flag_mask): Likewise.
3381 (noce_try_addcc): Use rev_cond if non-NULL instead of
3382 reversed_comparison_code.
3383 (noce_try_cmove_arith): Likewise. Formatting fixes.
3384 (noce_try_minmax, noce_try_abs): Clear rev_cond.
3385 (noce_find_if_block): Initialize rev_cond.
3386 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
3387 instead of false as last argument never attempt to reverse it
3390 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3392 PR tree-optimization/79663
3393 * tree-predcom.c (combine_chains): Process refs in reverse order
3394 only for ZERO length chains, and add explaining comment.
3396 2017-02-23 Jeff Law <law@redhat.com>
3398 PR tree-optimization/79578
3399 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
3400 in call to operand_equal_p.
3402 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
3405 * config/i386/cpuid.h: Fix another undefined behavior.
3407 2017-02-23 Richard Biener <rguenther@suse.de>
3409 PR tree-optimization/79683
3410 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
3411 vector types for data-refs.
3413 2017-02-23 Martin Liska <mliska@suse.cz>
3415 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
3417 2017-02-23 Jakub Jelinek <jakub@redhat.com>
3420 * internal-fn.c (get_range_pos_neg): Moved to ...
3421 * tree.c (get_range_pos_neg): ... here. No longer static.
3422 * tree.h (get_range_pos_neg): New prototype.
3423 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
3424 are known to be in between 0 and signed maximum inclusive, try to
3425 expand both unsigned and signed divmod and use the cheaper one from
3428 2017-02-22 Jeff Law <law@redhat.com>
3430 PR tree-optimization/79578
3431 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
3432 to compare base operands.
3434 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
3437 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
3438 gpc_reg_operand instead of fpr_reg_operand.
3440 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
3442 * config/mips/mips.c (mips_return_in_memory): Force FP
3443 vector types to be returned in memory for o32 ABI.
3445 2017-02-22 Jakub Jelinek <jakub@redhat.com>
3447 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
3448 instead of DW_TAG_member for static data member declarations and don't
3449 set no_linkage_name for static inline data members.
3450 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
3453 2017-02-22 Martin Liska <mliska@suse.cz>
3455 * doc/invoke.texi: Replace inequality signs with square brackets
3458 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3460 PR tree-optimization/68644
3461 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
3463 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
3466 * lra-constraints.c (simplify_operand_subreg): Handle
3467 WORD_REGISTER_OPERATIONS targets.
3469 2017-02-22 Jakub Jelinek <jakub@redhat.com>
3472 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
3473 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
3474 elimination by swapping fld*.
3476 2017-02-22 Richard Biener <rguenther@suse.de>
3478 PR tree-optimization/79673
3479 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
3480 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
3481 irrelevant address-space qualifiers and avoiding a
3482 ADDR_SPACE_CONVERT_EXPR from fold_convert.
3484 2017-02-22 Richard Biener <rguenther@suse.de>
3486 PR tree-optimization/79666
3487 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
3488 to not symbolically negate if that may introduce undefined
3491 2017-02-22 Martin Liska <mliska@suse.cz>
3494 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
3495 * data-streamer-out.c (streamer_write_gcov_count_stream):
3497 * value-prof.c (stream_out_histogram_value): Make assert more
3498 precise based on type of counter.
3500 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
3503 * config/i386/i386.md (standard_x87sse_constant_load splitter):
3504 Use nonimmediate_operand instead of memory_operand for operand 1.
3505 (float-extend standard_x87sse_constant_load splitter): Ditto.
3507 2017-02-21 Jeff Law <law@redhat.com>
3509 PR tree-optimization/79621
3510 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
3511 blocks with edges to themselves.
3513 2017-02-21 Jakub Jelinek <jakub@redhat.com>
3516 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
3517 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
3518 Use gimple_call_builtin_p.
3521 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
3522 on temporarily removed DEBUG_INSNs.
3524 PR tree-optimization/79649
3525 * tree-loop-distribution.c (classify_partition): Give up on
3526 non-generic address space loads/stores.
3528 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
3530 * doc/loop.texi (Loop manipulation): Remove nonexistent
3531 tree_ssa_loop_version from the documentation.
3532 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
3534 2017-02-21 Jakub Jelinek <jakub@redhat.com>
3537 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
3538 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
3539 * config/rs6000/rs6000.c: Include except.h.
3540 (rs6000_expand_split_stack_prologue): Call
3541 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
3543 2017-02-21 Martin Jambor <mjambor@suse.cz>
3546 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
3549 2017-02-21 Martin Jambor <mjambor@suse.cz>
3551 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
3552 for backward compatibility only.
3553 * doc/invoke.texi (Option Summary): Remove all references to
3556 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
3560 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3562 * lra-constraints.c (curr_insn_transform): Handle
3563 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
3565 2017-02-21 Martin Liska <mliska@suse.cz>
3567 * config/i386/i386.opt: Replace -masm-dialect with -masm.
3569 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
3571 PR translation/79638
3572 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
3574 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
3577 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
3578 (arm_function_ok_for_sibcall): Return false for an indirect call by
3579 descriptor if all the argument registers are used.
3580 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
3581 alignment of the function.
3583 2017-02-21 Jakub Jelinek <jakub@redhat.com>
3585 PR tree-optimization/61441
3586 * simplify-rtx.c (simplify_const_unary_operation): For
3587 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
3588 the sNaN unmodified.
3590 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
3592 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
3593 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
3594 instead of SYSTEM_HEADER_DIR.
3596 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
3597 Martin Liška <mliska@suse.cz>
3599 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
3600 Fix typos and grammar, use active voice, and clarify.
3602 2017-02-20 Marek Polacek <polacek@redhat.com>
3605 * gimplify.c (gimplify_expr): Handle unused *&&L;.
3608 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
3610 2017-02-20 Jakub Jelinek <jakub@redhat.com>
3613 * config/i386/i386.c (ix86_expand_builtin): Handle
3614 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
3615 ix86_builtins_isa[fcode].isa as a requirement of those
3616 flags and any other flag in the bitmask.
3617 (ix86_init_mmx_sse_builtins): Use 0 instead of
3618 ~OPTION_MASK_ISA_64BIT as mask.
3619 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
3620 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
3621 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
3622 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
3624 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3627 * lra-constraints.c (split_reg): Check requested split mode
3628 is supported by the register.
3630 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3632 * lra-constraints.c (simplify_operand_subreg): Remove early
3635 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3638 * lra-constraints.c (curr_insn_transform): Tighten condition
3639 for converting SUBREG reloads from OP_OUT to OP_INOUT.
3641 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3644 * lra-constraints.c (curr_insn_transform): Handle
3645 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
3647 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
3650 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
3652 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
3654 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
3657 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
3660 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3662 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
3665 2017-02-18 Jakub Jelinek <jakub@redhat.com>
3667 * final.c (last_columnnum, override_columnnum): New variables.
3668 (final_start_function): Set last_columnnum, pass it to begin_prologue
3669 hook and pass 0 to dwarf2out_begin_prologue.
3670 (final_scan_insn): Update override_columnnum. Pass last_columnnum
3671 to source_line debug hook.
3672 (notice_source_line): Compute last_columnnum and for debug_column_info
3673 return true on column changes.
3674 * debug.h (struct gcc_debug_hooks): Add column argument to
3675 source_line and begin_prologue hooks.
3676 (debug_nothing_int_charstar_int_bool): Remove prototype.
3677 (debug_nothing_int_int_charstar,
3678 debug_nothing_int_int_charstar_int_bool): New prototypes.
3679 (dwarf2out_begin_prologue): Add column argument.
3680 * debug.c (do_nothing_debug_hooks): Adjust source_line and
3681 begin_prologue hooks.
3682 (debug_nothing_int_charstar_int_bool): Remove.
3683 (debug_nothing_int_int_charstar,
3684 debug_nothing_int_int_charstar_int_bool): New functions.
3685 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
3686 through to dwarf2out_source_line.
3687 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
3688 (dwarf2out_source_line): Add column argument, emit it if requested.
3689 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
3691 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
3692 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
3693 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
3694 through to dwarf2out_begin_prologue.
3695 (vmsdbgout_source_line): Add column argument, pass it through to
3696 dwarf2out_source_line.
3697 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
3698 dbxout_source_line caller.
3699 (dbxout_source_line): Add column argument.
3701 * common.opt (gno-column-info, gcolumn-info): New options.
3702 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
3703 (check_die): Also test for multiple DW_AT_decl_column attributes.
3704 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
3705 DW_AT_decl_column if requested.
3706 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
3708 (gen_variable_die): Likewise.
3709 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
3710 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
3713 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
3714 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
3715 (ix86_handle_option): Handle OPT_m3dnowa.
3716 * doc/invoke.texi (-m3dnowa): Document.
3717 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
3718 -m3dnowa instead of -m3dnow -march=athlon.
3721 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
3722 instead of gcc_assert for K, r and R code checks. Formatting fixes.
3724 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3727 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
3728 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
3729 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
3730 generator for vsx_xxpermdi_<mode>_be.
3731 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
3732 force big-endian semantics.
3733 (vsx_xxpermdi_<mode>_be): New define_expand with same
3734 implementation as previous version of vsx_xxpermdi_<mode>.
3736 2017-02-17 Jakub Jelinek <jakub@redhat.com>
3738 PR tree-optimization/79327
3739 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
3740 variable, its initialization and use.
3742 2017-02-17 Julia Koval <julia.koval@intel.com>
3744 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
3745 (OPTION_MASK_ISA_PKU_UNSET): New.
3746 (ix86_handle_option): Handle -mrdpid.
3747 * config/i386/cpuid.h (bit_RDPID): New.
3748 * config/i386/driver-i386.c (host_detect_local_cpu):
3749 Detect RDPID feature.
3750 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
3751 * config/i386/i386-c.c (ix86_target_macros_internal):
3753 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
3754 (ix86_valid_target_attribute_inner_p): Add "rdpid".
3755 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
3756 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
3757 * config/i386/i386.md (define_insn "rdpid"): New.
3758 * config/i386/i386.opt Add -mrdpid.
3759 * config/i386/immintrin.h (_rdpid_u32): New.
3761 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
3763 PR rtl-optimization/79541
3764 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
3765 instead of transforming it into USE.
3767 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
3769 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
3770 If HONOR_SNANS (SFmode) force the input to a register.
3771 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
3772 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
3773 an frsp or similar insn.
3775 2017-02-17 Martin Liska <mliska@suse.cz>
3777 PR rtl-optimization/79577
3778 * params.def (selsched-max-sched-times): Increase minimum to 1.
3780 2017-02-17 Martin Liska <mliska@suse.cz>
3782 PR rtl-optimization/79574
3783 * gcse.c (want_to_gcse_p): Prevent integer overflow.
3785 2017-02-17 Martin Liska <mliska@suse.cz>
3787 PR tree-optimization/79529
3788 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
3789 ssa_defined_default_def_p to handle cases which are implicitly
3791 * tree-ssa.c (ssa_defined_default_def_p): New function.
3792 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
3793 which are implicitly defined.
3794 * tree-ssa.h (ssa_defined_default_def_p): Declare.
3796 2017-02-17 Richard Biener <rguenther@suse.de>
3799 * params.def (max-ssa-name-query-depth): Limit to 10.
3801 2017-02-17 Richard Biener <rguenther@suse.de>
3803 PR tree-optimization/79552
3804 * tree-ssa-structalias.c (visit_loadstore): Properly verify
3807 2017-02-17 Richard Biener <rguenther@suse.de>
3810 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
3812 2017-02-17 Marek Polacek <polacek@redhat.com>
3815 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
3816 (fold_negate_expr): New wrapper.
3818 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
3820 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
3821 Correct terminology and de-emphasize pre-standard behavior.
3823 2017-02-16 Alan Modra <amodra@gmail.com>
3825 PR rtl-optimization/79286
3826 * ira.c (def_dominates_uses): New function.
3827 (update_equiv_regs): Don't create an equivalence for insns that
3828 may trap where the register def does not dominate the use.
3830 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
3832 PR rtl-optimization/78127
3833 * lra.c (lra): Call lra_eliminate before finish the loop after
3836 2017-02-16 Richard Biener <rguenther@suse.de>
3838 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
3840 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
3841 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
3842 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
3843 (isl_val_int_from_wi): New function.
3844 (extract_affine_gmp): Rename to ...
3845 (extract_affine_wi): ... this, take a widest_int.
3846 (extract_affine_int): Just wrap extract_affine_wi.
3847 (add_param_constraints): Use isl_val_int_from_wi.
3848 (add_loop_constraints): Likewise, and extract_affine_wi.
3850 2017-02-15 Jeff Law <law@redhat.com>
3853 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
3854 ira_init_register_move_cost_if_necessary.
3856 2017-02-15 Martin Sebor <msebor@redhat.com>
3859 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
3860 removed in a prior commit.
3862 2017-02-15 Bin Cheng <bin.cheng@arm.com>
3864 PR tree-optimization/79347
3865 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
3866 counters during peeling.
3868 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
3870 * Makefile.in (site.exp): Remove "set ISLVER".
3872 2017-02-15 Jakub Jelinek <jakub@redhat.com>
3875 * real.c (real_from_integer): Call real_convert even for decimal.
3877 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
3880 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
3882 2017-02-14 Andrew Pinski <apinski@cavium.com>
3884 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
3885 cores and change the partno/implementer to be correct.
3886 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
3887 the 'B" as the implementer.
3888 * config/aarch64/aarch64-tune.md: Regenerate.
3890 2017-02-14 Carl Love <cel@us.ibm.com>
3892 * config/rs6000/rs6000.c: Add case statement entry to make the
3893 xvcvuxdsp built-in argument unsigned.
3894 * config/rs6000/vsx.md: Fix the source and return operand types so they
3895 match the instruction definitions from the ISA document. Fix typo
3896 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
3899 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
3902 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
3903 member early_clobber_alts.
3904 * lra-lives.c (reg_early_clobber_p): New.
3905 (process_bb_lives): Use it.
3906 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
3907 (debug_operand_data): Initialize early_clobber_alts.
3908 (setup_operand_alternative): Set up early_clobber_alts.
3909 (collect_non_operand_hard_regs): Ditto. Pass early clobber
3910 alternatives to new_insn_reg.
3911 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
3913 (lra_update_insn_regno_info): Pass the new arg.
3915 2017-02-14 Jakub Jelinek <jakub@redhat.com>
3918 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
3919 (new_oacc_loop_raw): Don't clear already cleared fields.
3922 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
3923 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
3924 _mm512_prefetch_i64gather_ps): New inline functions and macros.
3926 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
3929 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
3931 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
3934 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
3935 the extra instruction to the right place to store 128-bit constant
3938 2017-02-14 Martin Sebor <msebor@redhat.com>
3941 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
3942 warning for strings of unknown length.
3944 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
3946 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
3948 2017-02-14 Jeff Law <law@redhat.com>
3951 * ira-costs.c (scan_one_insn): Initialize register move costs
3952 for pseudos seen in USE/CLOBBER insns.
3954 PR tree-optimization/79095
3955 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
3956 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
3957 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
3958 if the operands are known to be not equal, then the resulting range
3960 (intersect_ranges): If the new range is ~[0,0] and the old range is
3961 wide, then prefer ~[0,0].
3962 * tree-vrp.c (overflow_comparison_p_1): New function.
3963 (overflow_comparison_p): New function.
3964 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
3965 if NAME is used in an overflow test.
3966 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
3967 overflow check that can be expressed as an equality test, then adjust
3968 ops to be that equality test.
3970 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3972 * config/s390/s390-builtin-types.def: Remove flags argument.
3973 * config/s390/s390.c (s390_init_builtins): Likewise.
3975 2017-02-14 Martin Liska <mliska@suse.cz>
3977 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
3978 vector. Fix trailing white spaces.
3980 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
3982 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
3985 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3987 PR rtl-optimization/68664
3988 * config/arm/arm.c (arm_sched_can_speculate_insn):
3989 New function. Declare prototype.
3990 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
3992 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3994 PR rtl-optimization/68664
3995 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
3997 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
3999 2017-02-14 Amit Pawar <amit.pawar@amd.com>
4001 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
4002 max skip bytes for function, loop and jump.
4004 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4006 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
4007 ABS_EXPR for gimple dump.
4009 2017-02-14 Jakub Jelinek <jakub@redhat.com>
4012 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
4014 PR tree-optimization/79408
4015 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
4016 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
4017 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
4018 also if rhs1 is INTEGER_CST.
4020 2017-02-14 Richard Biener <rguenther@suse.de>
4023 * tree-into-ssa.c (insert_phi_nodes): When the function can
4024 have abnormal edges rewrite SSA names with broken use-def
4025 dominance out of SSA and register them for PHI insertion.
4027 2017-02-13 Martin Sebor <msebor@redhat.com>
4030 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
4031 clearing info.nowrite flag when snprintf size argument is a range.
4033 2017-02-13 Jakub Jelinek <jakub@redhat.com>
4035 * cprop.c (cprop_jump): Add missing space in string literal.
4036 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
4037 (get_constraint_for_component_ref): Likewise.
4038 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
4039 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
4040 * lra-constraints.c (process_alt_operands): Likewise.
4041 * ipa-inline.c (inline_small_functions): Likewise.
4042 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
4043 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
4044 * trans-mem.c (diagnose_tm_1_op): Likewise.
4045 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
4046 (grid_parallel_clauses_gridifiable): Likewise.
4048 * config/nvptx/mkoffload.c (process): Add space in between
4051 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
4052 "MOD4_SSE_REGS" and "ALL_REGS".
4054 * spellcheck.c (test_data): Add , in between "foo" and "food".
4056 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4059 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
4060 boundary crossing check and subsequent code generation agree.
4062 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4064 * config/aarch64/aarch64.c (has_memory_op): Delete.
4065 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
4068 2017-02-13 Jakub Jelinek <jakub@redhat.com>
4070 PR rtl-optimization/79388
4071 PR rtl-optimization/79450
4072 * combine.c (distribute_notes): When removing TEM_INSN for which
4073 corresponding dest has last value recorded, invalidate that last
4076 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4078 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
4079 of explicit '@'. Add missing assembly comment marker on branch costs
4082 2017-02-13 Nathan Sidwell <nathan@acm.org>
4084 * gengtype-lex.l (<in_struct>): Add '/'.
4086 2017-02-13 Martin Liska <mliska@suse.cz>
4089 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
4091 2017-02-13 Richard Biener <rguenther@suse.de>
4093 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
4095 * configure: Re-generate.
4096 * config.in: Likewise.
4097 * graphite-dependences.c: Simplify as if
4098 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
4099 * graphite-isl-ast-to-gimple.c: Likewise.
4100 * graphite-optimize-isl.c: Likewise.
4101 * graphite-poly.c: Likewise.
4102 * graphite-sese-to-poly.c: Likewise.
4103 * graphite.h: Likewise.
4104 * toplev.c: Include isl/version.h and use isl_version () for
4105 printing the ISL version.
4106 * doc/install.texi: Update ISL requirement.
4108 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
4110 * doc/standards.texi (Standards): Update reference to
4113 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
4115 * doc/extend.texi (Named Address Spaces): sourceware.org now
4117 * doc/install.texi (Binaries): Ditto.
4120 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
4122 * doc/cpp.texi: Replace "stringify"/"stringification" with C
4123 standard terminology "stringize"/"stringizing" throughout.
4124 * doc/cppinternals.texi: Likewise.
4126 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
4128 * doc/extend.texi: Fix some spelling mistakes and typos.
4129 * doc/invoke.texi: Likewise.
4131 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
4134 * params.def (inline-min-speedup) Change from 10 to 8.
4136 2017-02-11 Jakub Jelinek <jakub@redhat.com>
4138 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
4141 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
4144 * ipa-inline-analysis.c (get_minimal_bb): New function.
4145 (record_modified): Use it.
4146 (remap_edge_change_prob): Handle also ancestor functions.
4148 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
4150 * doc/contrib.texi (Contributors): Remove broken link into
4151 the Mauve CVS repository.
4153 2017-02-11 Jakub Jelinek <jakub@redhat.com>
4156 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
4157 result computation whenever lhs doesn't have vector mode, not
4158 just when it has BLKmode.
4160 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
4162 * doc/makefile.texi (profiledbootstrap): Refer to the
4163 installation instructions only in textual form.
4165 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4168 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
4170 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
4172 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
4173 (Specific): Update mingw-w64 reference.
4175 (Specific): Remove broken link to Renesas RX processor.
4177 2017-02-10 Richard Biener <rguenther@suse.de>
4179 * toplev.c (process_options): Do not mention obsolete graphite
4180 options when printing sorry message about missing graphite support.
4181 Mention -floop-nest-optimize.
4183 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
4185 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
4186 (vtst_p16): Likewise.
4187 (vtstq_p8): Likewise.
4188 (vtstq_p16): Likewise.
4190 (vtstq_p64): Likewise.
4191 * config/arm/arm_neon.h (vgetq_lane_p64): New.
4192 (vset_lane_p64): New.
4193 (vsetq_lane_p64): New.
4195 2017-02-10 Jakub Jelinek <jakub@redhat.com>
4197 PR tree-optimization/79411
4198 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
4199 stmt operands are SSA_NAMEs used in abnormal phis.
4200 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
4203 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
4206 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
4209 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
4211 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
4213 2017-02-09 Jakub Jelinek <jakub@redhat.com>
4215 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
4219 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
4220 not arbitrary TREE_CONSTANT.
4223 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
4224 "omp declare target link" attribute unless is_global_var.
4225 * omp-offload.c (find_link_var_op): Likewise.
4227 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
4228 Chung-Lin Tang <cltang@codesourcery.com>
4230 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
4232 (gimplify_adjust_omp_clauses): Don't delete TILE.
4233 (gimplify_omp_for): Deal with TILE.
4234 * internal-fn.c (expand_GOACC_TILE): New function.
4235 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
4237 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
4238 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
4240 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
4241 avoid DIV for outermost collapse var.
4242 (expand_oacc_for): Insert tile element loop as needed. Adjust.
4243 Remove out of date comments, fix whitespace.
4244 * omp-general.c (omp_extract_for_data): Deal with tiling.
4245 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
4246 adjust OLF_DIM_BASE value.
4247 (struct omp_for_data): Add tiling field.
4248 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
4249 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
4250 for auto loops. Remove default auto determining, moved to
4251 oacc_loop_fixed_partitions.
4252 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
4253 stmts, add e_mask field.
4254 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
4255 (oacc_thread_numbers): Use oacc_dim_call.
4256 (oacc_xform_tile): New.
4257 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
4258 (finish_oacc_loop): Adjust for ifns vector.
4259 (oacc_loop_discover_walk): Append loop abstraction sites to list,
4260 add case for GOACC_TILE fns.
4261 (oacc_loop_xform_loop): Delete.
4262 (oacc_loop_process): Iterate over call list directly, and add
4263 handling for GOACC_TILE fns.
4264 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
4266 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
4267 vector partitioning to outer loops. Assign 2 partitions to loops
4268 when available. Add TILE handling.
4269 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
4270 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
4271 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
4272 * tree.c (omp_clause_num_ops): Adjust TILE ops.
4273 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
4275 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
4277 * configure.ac (ACX_BUGURL): Update.
4278 * configure: Regenerate.
4280 2017-02-09 Richard Biener <rguenther@suse.de>
4282 PR tree-optimization/69823
4283 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
4284 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
4286 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
4288 * config/arc/arc-c.def: Add __NPS400__ definition.
4289 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
4290 (TARGET_NPS400): Define.
4292 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
4294 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
4296 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
4298 (arc_cpu_types): Fill the arch_info field with a pointer into the
4299 arc_arch_types table.
4300 (arc_selected_cpu): Declare.
4301 * config/arc/arc.c (arc_selected_cpu): Make global.
4302 (arc_selected_arch): Delete.
4303 (arc_base_cpu): Delete.
4304 (arc_override_options): Remove references to deleted variables,
4305 update access to arch information.
4306 (ARC_OPT): Update access to arch information.
4307 (ARC_OPTX): Likewise.
4308 * config/arc/arc.h (arc_base_cpu): Remove declaration.
4309 (TARGET_ARC600): Update access to arch information.
4310 (TARGET_ARC601): Likewise.
4311 (TARGET_ARC700): Likewise.
4312 (TARGET_EM): Likewise.
4313 (TARGET_HS): Likewise.
4314 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
4317 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
4320 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
4321 condition/operands for integer GE/LE/GEU/LEU operations.
4323 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
4325 PR translation/79397
4326 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
4329 2017-02-08 Martin Jambor <mjambor@suse.cz>
4332 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
4333 whether allocation happened.
4334 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
4335 nothing was allocated.
4337 2017-02-08 Jakub Jelinek <jakub@redhat.com>
4339 PR tree-optimization/79408
4340 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
4341 constant, but SSA_NAME with a known integer range, use the minimum
4342 of that range instead of op1 to determine if modulo can be replaced
4343 with its first operand.
4345 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4347 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
4349 2017-02-08 Richard Biener <rguenther@suse.de>
4351 PR tree-optimization/71824
4352 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
4353 Check all loops contained in the merged region.
4355 2017-02-07 Andrew Pinski <apinski@cavium.com>
4357 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
4359 2017-02-07 Andrew Pinski <apinski@cavium.com>
4361 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
4362 (thunderxt88): Likewise.
4363 (thunderxt81): Disable LSE and change v8.1 to v8.
4364 (thunderxt83): Likewise.
4366 2017-02-07 Jakub Jelinek <jakub@redhat.com>
4367 Richard Biener <rguenther@suse.de>
4370 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
4371 type from int to size_t.
4372 * ira-costs.c (struct_costs_size): Change type from int to size_t.
4374 2017-02-07 Jakub Jelinek <jakub@redhat.com>
4376 PR rtl-optimization/79386
4377 * cprop.c (bypass_conditional_jumps): Initialize
4378 bypass_last_basic_block already before splitting bbs after
4379 unconditional traps...
4380 (bypass_conditional_jumps): ... rather than here.
4383 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
4384 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
4385 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
4386 fix -masm=intel patterns.
4388 2017-02-07 Richard Biener <rguenther@suse.de>
4390 PR tree-optimization/79256
4392 * builtins.c (get_object_alignment_2): Use min_align_of_type
4393 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
4394 and ADJUST_FIELD_ALIGN.
4396 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
4398 * doc/tm.texi: Regenerate.
4399 * stor-layout.c (layout_decl): Adjust.
4400 (update_alignment_for_field): Likewise.
4401 (place_field): Likewise.
4402 (min_align_of_type): Likewise.
4403 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
4404 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
4405 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
4406 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
4407 * config/frv/frv.c (frv_adjust_field_align): Likewise.
4408 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
4409 * config/i386/i386.c (x86_field_alignment): Likewise.
4410 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
4411 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
4412 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
4413 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
4414 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
4415 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
4419 2017-01-30 Richard Biener <rguenther@suse.de>
4421 PR tree-optimization/79256
4422 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
4423 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
4426 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
4428 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
4429 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
4430 builtins to SImode and emit a zero-extend, if necessary.
4432 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
4434 * docs/invoke.texi (RISC-V Options): Alphabetize.
4436 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
4438 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
4441 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
4443 * config/riscv/riscv.c: New file.
4444 * gcc/common/config/riscv/riscv-common.c: Likewise.
4445 * config.gcc: Likewise.
4446 * config/riscv/constraints.md: Likewise.
4447 * config/riscv/elf.h: Likewise.
4448 * config/riscv/generic.md: Likewise.
4449 * config/riscv/linux.h: Likewise.
4450 * config/riscv/multilib-generator: Likewise.
4451 * config/riscv/peephole.md: Likewise.
4452 * config/riscv/pic.md: Likewise.
4453 * config/riscv/predicates.md: Likewise.
4454 * config/riscv/riscv-builtins.c: Likewise.
4455 * config/riscv/riscv-c.c: Likewise.
4456 * config/riscv/riscv-ftypes.def: Likewise.
4457 * config/riscv/riscv-modes.def: Likewise.
4458 * config/riscv/riscv-opts.h: Likewise.
4459 * config/riscv/riscv-protos.h: Likewise.
4460 * config/riscv/riscv.h: Likewise.
4461 * config/riscv/riscv.md: Likewise.
4462 * config/riscv/riscv.opt: Likewise.
4463 * config/riscv/sync.md: Likewise.
4464 * config/riscv/t-elf-multilib: Likewise.
4465 * config/riscv/t-linux: Likewise.
4466 * config/riscv/t-linux-multilib: Likewise.
4467 * config/riscv/t-riscv: Likewise.
4468 * configure.ac: Likewise.
4469 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
4470 Waterman as RISC-V maintainers.
4471 * doc/install.texi: Add RISC-V entries.
4472 * doc/invoke.texi: Add RISC-V options section.
4473 * doc/md.texi: Add RISC-V constraints section.
4474 * configure: Regenerated.
4476 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4479 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
4480 false values to be constant vectors with all 0 or all 1 bits set.
4481 (vcondu<mode><mode>): Likewise.
4482 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
4484 (fpmask_comparison_operator): Update comment.
4485 (vecint_comparison_operator): New predicate.
4486 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
4487 vector conditionals when the true and false values are constant
4488 vectors with all 0 bits or all 1 bits set.
4490 2017-02-06 Martin Sebor <msebor@redhat.com>
4492 PR tree-optimization/79376
4493 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
4495 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
4497 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
4498 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
4499 to simplify split condition.
4501 2017-02-06 Jakub Jelinek <jakub@redhat.com>
4503 * omp-expand.c (oxpand_omp_atomic_fetch_op,
4504 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
4507 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
4509 PR rtl-optimization/68664
4510 * target.def (can_speculate_insn): New hook.
4511 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
4512 * doc/tm.texi: Regenerate.
4513 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
4514 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
4515 (rs6000_sched_can_speculate_insn): New function.
4517 2017-02-06 Jakub Jelinek <jakub@redhat.com>
4519 PR tree-optimization/79284
4520 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
4521 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
4522 vectorizable_mask_load_store, vectorizable_operation,
4523 vect_is_simple_cond, get_same_sized_vectype): Use it instead
4524 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
4525 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
4526 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
4527 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4528 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
4529 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
4530 is_gimple_assign (stmt). Replace another such test with
4531 is_gimple_assign (stmt).
4533 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
4536 * config/avr/avr.c (rtl-iter.h): Include it.
4537 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
4538 (avr_legitimate_combined_insn): ...and implementation.
4540 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
4542 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
4543 * config/s390/s390.c (s390_const_operand_ok)
4544 (s390_canonicalize_comparison, s390_extract_part)
4545 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
4546 (s390_contiguous_bitmask_p, s390_rtx_costs)
4547 (legitimize_pic_address): Likewise.
4548 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
4549 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
4550 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
4551 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
4552 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
4554 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
4556 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
4557 REGNO($0) == REGNO($1).
4559 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4561 * config/s390/linux.h(SIZE_TYPE): Add comment.
4563 2017-02-06 Julian Brown <julian@codesourcery.com>
4564 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4565 Virendra Pathak <virendra.pathak@broadcom.com>
4567 * config/aarch64/aarch64-cores.def: Change the scheduler
4569 * config/aarch64/aarch64.md: Include thunderx2t99.md.
4570 * config/aarch64/thunderx2t99.md: New file.
4572 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
4574 * doc/standards.texi (Go Language): Update link to language
4577 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
4579 * tree-eh.c (lower_resx): Sanitize profile.
4580 (cleanup_empty_eh_move_lp): Likewise.
4582 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
4585 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
4587 * cfgloopmanip.h (loop_version): Update prototype.
4588 * modulo-sched.c (sms_schedule): Update call of loop_version.
4589 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
4590 * tree-parloops.c (gen_parallel_loop): Likewise.
4591 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
4592 * tree-ssa-loop-split.c (split_loop): Likewise.
4593 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
4594 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
4596 2017-02-05 Martin Liska <mliska@suse.cz>
4599 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
4601 (print_operand_address): Initialize a struct to zero.
4603 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
4605 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
4606 garbage collector only in textual form.
4608 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
4610 * doc/extend.texi (x86 specific memory model extensions for
4611 transactional memory): Simplify a phrase.
4613 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
4616 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
4617 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
4618 (atomic_storedi_1): Likewise.
4620 2017-02-04 Jakub Jelinek <jakub@redhat.com>
4622 PR tree-optimization/79338
4623 * tree-parloops.c (gather_scalar_reductions): Don't call
4624 vect_analyze_loop_form for loop->inner before destroying loop's
4627 2017-02-03 Martin Sebor <msebor@redhat.com>
4629 PR tree-optimization/79327
4630 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
4631 when precision has resulted in leading zeros.
4632 (format_integer): Adjust the likely counter to assume an unknown
4633 argument that may be zero is non-zero.
4635 2017-02-03 Jason Merrill <jason@redhat.com>
4638 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
4639 avoid copying non-taken branch.
4641 2017-02-03 Jakub Jelinek <jakub@redhat.com>
4643 PR tree-optimization/79340
4644 * tree-vect-loop.c (vectorizable_reduction): Release
4645 vec_defs elements after safe_splicing them into other vectors.
4648 PR tree-optimization/79327
4649 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
4650 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
4652 (format_integer): Use wide_int_to_tree instead of build_int_cst
4653 + to_?hwi. If argmin is NULL, just set argmin and argmax to
4654 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
4655 of shortest and longest sequence.
4657 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
4659 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
4660 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
4662 2017-02-03 Walter Lee <walt@tilera.com>
4665 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
4666 after initial stackframe link reg save.
4667 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
4669 2017-02-03 Jakub Jelinek <jakub@redhat.com>
4672 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
4673 wu for stxssp alternative.
4675 2017-02-03 Martin Sebor <msebor@redhat.com>
4677 PR tree-optimization/79352
4678 * gimple-fold.c (get_range_strlen): Add argument.
4679 (get_range_strlen): Change return type to bool.
4680 (get_maxval_strlen): Pass in a dummy argument.
4681 * gimple-fold.h (get_range_strlen): Change return type to bool.
4682 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
4683 * tree.h (array_at_struct_end_p): Add argument.
4684 * tree.c (array_at_struct_end_p): Handle it.
4686 2017-02-03 Martin Liska <mliska@suse.cz>
4689 * multiple_target.c (create_dispatcher_calls): Redirect edge
4690 from a caller of a dispatcher.
4691 (expand_target_clones): Make the clones local.
4692 (ipa_target_clone): Do both target clones and resolvers.
4693 (ipa_dispatcher_calls): Remove the pass.
4694 (pass_dispatcher_calls::gate): Likewise.
4695 (make_pass_dispatcher_calls): Likewise.
4696 * passes.def (pass_target_clone): Put as very first IPA early
4699 2017-02-03 Martin Liska <mliska@suse.cz>
4701 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
4702 in case of a function with ifunc attribute.
4704 2017-02-03 Martin Liska <mliska@suse.cz>
4706 * cgraph.c (cgraph_node::dump): Dump function version info.
4707 * symtab.c (symtab_node::dump_base): Add missing new line.
4709 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
4711 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
4712 (ifcombine_ifandif): Use it.
4714 2017-02-03 Martin Liska <mliska@suse.cz>
4716 * doc/invoke.texi: Document default value for
4717 use-after-scope-direct-emission-threshold.
4719 2017-02-03 Martin Liska <mliska@suse.cz>
4721 PR tree-optimization/79339
4722 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
4723 (format_floating): Likewise.
4725 2017-02-03 Martin Liska <mliska@suse.cz>
4728 * ipa-prop.c (ipa_node_params_t::insert): Remove current
4730 (ipa_node_params_t::remove): Likewise.
4731 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
4732 initialization from removed ipa_node_params_t::insert.
4733 (ipa_node_params::~ipa_node_params): Move from removed
4734 ipa_node_params_t::release.
4735 * symbol-summary.h (symbol_summary::m_released): New member.
4736 Do not release a summary twice. Do not allow to call finalizer
4737 for types of a summary that live in GGC memory.
4739 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4741 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
4744 2017-02-02 Martin Sebor <msebor@redhat.com>
4747 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
4748 (format_string): Tighten up the range of output for non-constant
4749 strings and correct the expected range for wide non-constant strings.
4751 2017-02-02 Martin Sebor <msebor@redhat.com>
4753 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
4756 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
4758 (-fdump-tree-@var): Add to index and document how to come up
4759 with pass-specific option and dump file names.
4760 (-fdump-passes): Clarify where to look for output.
4762 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
4765 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
4766 statistics of the analyzed path; allow threading for speed when
4767 any of BBs along the path are optimized for speed.
4769 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
4772 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
4773 settings of the virtual registers.
4776 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
4778 * explow.c (get_dynamic_stack_size): Take known alignment of stack
4779 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
4782 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4784 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
4785 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
4787 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4789 * config/s390/s390.md: Add missing comments with the expanded
4791 * config/s390/vector.md: Likewise.
4792 * config/s390/vx-builtins.md: Likewise.
4794 2017-02-02 Jakub Jelinek <jakub@redhat.com>
4797 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
4798 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
4799 conditions on a single line.
4801 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4803 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
4804 __S390_VX__ to __VX__.
4806 2017-02-01 Andrew Pinski <apinski@cavium.com>
4808 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
4809 stmt_info to record_stmt_cost.
4810 (vect_get_known_peeling_cost): Pass stmt_info if known to
4812 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
4813 cpu_vector_cost field into
4814 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
4815 field into vec_int_stmt_cost and vec_fp_stmt_cost.
4816 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
4817 splitting of scalar_stmt_cost and vec_stmt_cost.
4818 (thunderx_vector_cost): Likewise.
4819 (cortexa57_vector_cost): LIkewise.
4820 (exynosm1_vector_cost): Likewise.
4821 (xgene1_vector_cost): Likewise.
4822 (thunderx2t99_vector_cost): Improve after the splitting of the two
4824 (aarch64_builtin_vectorization_cost): Update for the splitting of
4825 scalar_stmt_cost and vec_stmt_cost.
4827 2017-02-01 Torvald Riegel <triegel@redhat.com>
4828 Richard Henderson <rth@redhat.com>
4830 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
4831 conditional on existance of a fast atomic load.
4832 * optabs-query.c (can_atomic_load_p): New function.
4833 * optabs-query.h (can_atomic_load_p): Declare it.
4834 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
4835 no fast atomic load is available for the particular size of access.
4836 (expand_atomic_compare_and_swap): Likewise.
4837 (expand_atomic_load): Likewise.
4838 (expand_atomic_store): Likewise.
4839 (expand_atomic_fetch_op): Likewise.
4840 * testsuite/lib/target-supports.exp
4841 (check_effective_target_sync_int_128): Remove x86 because it provides
4842 no fast atomic load.
4843 (check_effective_target_sync_int_128_runtime): Likewise.
4845 2017-02-01 Richard Biener <rguenther@suse.de>
4847 * graphite.c: Include tree-vectorizer.h for find_loop_location.
4848 (graphite_transform_loops): Provide opt-info for optimized nests.
4849 * tree-parloop.c (parallelize_loops): Provide opt-info for
4852 2017-02-01 Richard Biener <rguenther@suse.de>
4855 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
4858 2017-02-01 Richard Biener <rguenther@suse.de>
4860 PR tree-optimization/71824
4861 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
4862 Verify the loops are valid in the merged SESE region.
4863 (scop_detection::can_represent_loop_1): Check analyzing the
4864 evolution of the number of iterations in the region succeeds.
4866 2017-01-31 Ian Lance Taylor <iant@golang.org>
4868 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
4869 REG_ARGS_SIZE note to 32-bit push insns and call insn.
4871 2017-01-31 David Malcolm <dmalcolm@redhat.com>
4873 PR preprocessor/79210
4874 * input.c (get_substring_ranges_for_loc): Replace line_width
4875 assertion with error-handling.
4877 2017-01-31 Richard Biener <rguenther@suse.de>
4879 PR tree-optimization/77318
4880 * graphite-sese-to-poly.c (extract_affine): Fix assert.
4881 (create_pw_aff_from_tree): Take loop parameter.
4882 (add_condition_to_pbb): Pass loop of the condition to
4883 create_pw_aff_from_tree.
4885 2017-01-31 Jakub Jelinek <jakub@redhat.com>
4887 * config/s390/s390.c (s390_asan_shadow_offset): New function.
4888 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
4890 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4894 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
4896 (convert_int_to_float128): Likewise.
4897 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
4898 (convert_int_to_float128): Likewise.
4899 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
4900 (UNSPEC_IEEE128_CONVERT): Likewise.
4901 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
4902 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
4903 Use local variables for IBM extended format.
4904 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
4905 (fix_trunc<mode>si2_fprs): Likewise.
4906 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
4907 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
4908 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
4909 to know that we can now have integers of all sizes in vector
4911 (fix<uns>_<mode>di2_hw): Likewise.
4912 (float<uns>_<mode>si2_hw): Likewise.
4913 (fix_<mode>si2_hw): Likewise.
4914 (fixuns_<mode>si2_hw): Likewise.
4915 (float<uns>_<mode>di2_hw): Likewise.
4916 (float_<mode>di2_hw): Likewise.
4917 (float_<mode>si2_hw): Likewise.
4918 (floatuns_<mode>di2_hw): Likewise.
4919 (floatuns_<mode>si2_hw): Likewise.
4920 (xscvqp<su>wz_<mode>): Delete, no longer used.
4921 (xscvqp<su>dz_<mode>): Likewise.
4922 (xscv<su>dqp_<mode>): Likewise.
4923 (ieee128_mfvsrd_64bit): Likewise.
4924 (ieee128_mfvsrd_32bit): Likewise.
4925 (ieee128_mfvsrwz): Likewise.
4926 (ieee128_mtvsrw): Likewise.
4927 (ieee128_mtvsrd_64bit): Likewise.
4928 (ieee128_mtvsrd_32bit): Likewise.
4930 2017-01-31 Martin Liska <mliska@suse.cz>
4933 * ipa-prop.c (ipa_free_all_node_params): Call release method
4934 instead of ~sumbol_summary to not to trigger double times
4937 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
4939 PR tree-optimization/71691
4940 * bitmap.h (class auto_bitmap): New.
4941 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
4942 is_maybe_undefined instead of ssa_undefined_value_p.
4944 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4946 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
4947 __S390_ARCH_LEVEL__ to __ARCH__.
4949 2017-01-31 Jakub Jelinek <jakub@redhat.com>
4951 PR tree-optimization/79267
4952 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
4953 if should_remove_lhs_p is true.
4955 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
4958 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
4959 (add_alignment_attribute): New.
4960 (base_type_die): Add alignment attribute.
4961 (subrange_type_die): Likewise.
4962 (modified_type_die): Likewise.
4963 (gen_array_type_die): Likewise.
4964 (gen_descr_array_type_die: Likewise.
4965 (gen_enumeration_type_die): Likewise.
4966 (gen_subprogram_die): Likewise.
4967 (gen_variable_die): Likewise.
4968 (gen_field_die): Likewise.
4969 (gen_ptr_to_mbr_type_die): Likewise.
4970 (gen_struct_or_union_type_die): Likewise.
4971 (gen_subroutine_type_die): Likewise.
4972 (gen_typedef_die): Likewise.
4973 (base_type_cmp): Compare alignment attribute.
4975 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4978 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
4979 (setb_unsigned) New pattern for setb with CCUNS.
4980 * config/rs6000/rs6000.c (expand_block_compare): Use a different
4981 subfc./subfe sequence to avoid overflow problems. Generate a
4982 shorter sequence with cmpld/setb for power9.
4983 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
4984 for generating subfc. instruction.
4985 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
4986 now uses this instruction.
4988 2017-01-30 Ian Lance Taylor <iant@google.com>
4991 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
4992 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
4994 2017-01-30 Martin Sebor <msebor@redhat.com>
4996 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
4997 Move constant to the right of a relational operator.
4998 (get_mpfr_format_length, format_character, format_string): Ditto.
4999 (should_warn_p, maybe_warn): Same.
5001 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
5003 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5006 * asan.c (get_translation_unit_decl): Remove function.
5007 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
5009 2017-01-30 Martin Liska <mliska@suse.cz>
5011 PR gcov-profile/79259
5012 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
5015 2017-01-30 Martin Liska <mliska@suse.cz>
5018 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
5019 Initialize variables with NULL value.
5021 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
5024 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
5026 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
5028 2017-01-30 Richard Biener <rguenther@suse.de>
5030 PR tree-optimization/79276
5031 * tree-vrp.c (process_assert_insertions): Properly adjust common
5032 when removing a duplicate.
5034 * gcc.dg/torture/pr79276.c: New testcase.
5036 2017-01-30 Richard Biener <rguenther@suse.de>
5038 PR tree-optimization/79256
5039 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
5040 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
5042 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
5044 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
5047 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
5048 ("*r<noxa>sbg_<mode>_sll_bitmask")
5049 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
5050 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
5051 Use contiguous_bitmask_nowrap_operand.
5053 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5056 * config/rs6000/altivec.h (vec_xl): Revise #define.
5057 (vec_xst): Likewise.
5059 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
5061 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
5063 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
5065 PR rtl-optimization/79194
5066 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
5067 traps before call to bypass_conditional_jumps.
5069 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
5071 PR tree-optimization/71374
5072 * lra-constraints.c (check_conflict_input_operands): New.
5073 (match_reload): Use it.
5075 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
5078 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
5079 account to calculate conflict_set.
5081 2017-01-27 Bin Cheng <bin.cheng@arm.com>
5083 PR rtl-optimization/78559
5084 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
5085 other_insn in combine.
5087 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
5089 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
5090 uint16_type_node for BT_UINT16.
5092 2017-01-27 David Malcolm <dmalcolm@redhat.com>
5094 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
5095 "RTL Tests" to menu.
5096 (GIMPLE Tests): New node.
5097 (RTL Tests): New node.
5099 2017-01-27 Richard Biener <rguenther@suse.de>
5101 PR tree-optimization/79245
5102 * tree-loop-distribution.c (distribute_loop): Apply cost
5103 modeling also to detected patterns.
5105 2017-01-27 Richard Biener <rguenther@suse.de>
5107 PR tree-optimization/71433
5108 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
5109 (compare_assert_loc): New function.
5110 (process_assert_insertions): Sort and optimize assert locations
5111 to remove duplicates and push down identical assertions on
5112 edges to their destination block.
5114 2017-01-27 Richard Biener <rguenther@suse.de>
5116 PR tree-optimization/79244
5117 * tree-vrp.c (remove_range_assertions): Forcefully propagate
5118 out SSA names even if abnormal.
5120 2017-01-27 Jakub Jelinek <jakub@redhat.com>
5122 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
5123 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
5124 instead of MPFR_RNDN.
5126 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
5129 * arm.c (arm_option_override): Don't call build_target_option_node
5130 until after doing all option overrides.
5131 (arm_valid_target_attribute_tree): Likewise.
5133 2017-01-27 Martin Liska <mliska@suse.cz>
5135 * doc/invoke.texi (-fprofile-arcs): Document profiling support
5136 for {cd}tors and C++ {cd}tors.
5138 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
5140 * config/s390/s390.md ("*setmem_long_and")
5141 ("*setmem_long_and_31z"): Use zero_extend instead of and.
5143 2017-01-26 Martin Sebor <msebor@redhat.com>
5145 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
5148 2017-01-26 Martin Sebor <msebor@redhat.com>
5150 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
5151 HAVE_DFmode before using XFmode or DFmode.
5152 (parse_directive): Avoid using the z length modifier to avoid
5153 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
5156 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
5157 to accept adjustment as an array.
5158 (get_int_range): New function.
5159 (struct directive): Make width and prec arrays.
5160 (directive::set_width, directive::set_precision): Call get_int_range.
5161 (format_integer, format_floating): Handle width and precision ranges.
5162 (format_string, parse_directive): Same.
5164 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5167 * dwarf2out.c (generate_skeleton_bottom_up): For children with
5168 comdat_type_p set, just clone them, but keep the children in the
5172 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
5173 which have direct callers with -fvar-tracking-assignments enabled
5175 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
5176 inside of type units.
5178 2017-01-26 Martin Sebor <msebor@redhat.com>
5181 * gimple-ssa-sprintf.c (struct result_range): Add likely and
5183 (struct format_result): Replace number_chars, number_chars_min,
5184 and number_chars_max with a single member of struct result_range.
5186 (format_result::operator+=): Adjust.
5187 (struct fmtresult): Remove bounded. Handle likely and unlikely
5189 (fmtresult::adjust_for_width_or_precision): New function.
5190 (fmtresult:type_max_digits): New function.
5191 (bytes_remaining): Handle likely and unlikely counters.
5192 (min_bytes_remaining): Remove.
5193 (format_percent): Simplify.
5194 (format_integer, format_floating): Set likely and unlikely counters.
5195 (get_string_length, format_character, format_string): Same.
5196 (format_plain, should_warn_p): New function.
5197 (maybe_warn): Call should_warn_p. Update diagnostic messages
5198 and handle those for all directives, including plain strings.
5199 (format_directive): Handle likely and unlikely counters.
5200 Remove unnecessary quoting from diagnostics. Add an informational
5202 (add_bytes): Remove.
5203 (pass_sprintf_length::compute_format_length): Simplify.
5204 (try_substitute_return_value): Handle likely and unlikely counters.
5206 2017-01-26 Carl Love <cel@us.ibm.com>
5208 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
5209 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
5211 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
5214 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
5215 endianess for subregs into account.
5216 * lra-constraints.c (lra_constraints): Do risky transformations
5217 always on the first iteration.
5218 * lra-lives.c (check_pseudos_live_through_calls): Add arg
5219 last_call_used_reg_set.
5220 (process_bb_lives): Define and use last_call_used_reg_set.
5221 * lra.c (lra): Always continue after lra_constraints on the first
5224 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
5226 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
5228 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
5230 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5232 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
5233 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
5234 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
5235 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
5236 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
5237 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
5238 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
5239 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
5240 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
5242 2017-01-26 Marek Polacek <polacek@redhat.com>
5245 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
5246 for the third operand.
5248 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5251 * omp-low.c (struct omp_context): Add simt_stmt field.
5252 (scan_omp_for): Return omp_context *.
5253 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
5254 context to the _simt_ SIMD stmt.
5255 (lower_omp_for): For combined SIMD with sibling _simt_
5256 SIMD, make sure to use the same decls in _looptemp_
5257 clauses as in the sibling.
5259 2017-01-26 David Sherwood <david.sherwood@arm.com>
5262 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
5265 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5268 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
5269 emit fld b; fld a; if possible.
5271 * brig-builtins.def: Update copyright years.
5272 * config/arm/arm_acle_builtins.def: Update copyright years.
5274 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5277 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
5278 constraint instead of o for the stxsd instruction.
5280 2017-01-25 Carl Love <cel@us.ibm.com>
5282 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
5283 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
5285 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
5287 * doc/invoke.texi (C++ Dialect Options): Fix typo.
5289 2017-01-25 Richard Biener <rguenther@suse.de>
5291 PR tree-optimization/69264
5292 * target.def (vector_alignment_reachable): Improve documentation.
5293 * doc/tm.texi: Regenerate.
5294 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
5296 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
5297 earlier changes with respect to TYPE_USER_ALIGN.
5298 (vector_alignment_reachable_p): Likewise. Improve dumping.
5300 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5303 * config/arm/arm.md (xordi3): Force constant operand into a register
5306 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5308 * doc/invoke.texi (-fstore-merging): Correct default optimization
5309 levels at which it is enabled.
5310 (-O): Move -fstore-merging from list to...
5313 2017-01-25 Richard Biener <rguenther@suse.de>
5316 * omp-expand.c: Include debug.h.
5317 (expand_omp_taskreg): Make sure to generate early debug before
5318 outlining anything from a function.
5319 (expand_omp_target): Likewise.
5320 (grid_expand_target_grid_body): Likewise.
5322 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
5325 * asan.c (get_translation_unit_decl): New function.
5326 (asan_add_global): Extract modules file name from globals
5327 TRANSLATION_UNIT_DECL name.
5329 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
5332 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
5333 for long calls with APCS frame and VFP.
5335 2017-01-24 David Malcolm <dmalcolm@redhat.com>
5337 * cfg.c (original_copy_tables_initialized_p): New function.
5338 * cfg.h (original_copy_tables_initialized_p): New decl.
5339 * cfgrtl.c (relink_block_chain): Guard the call to
5340 free_original_copy_tables with a call to
5341 original_copy_tables_initialized_p.
5342 * cgraph.h (symtab_node::native_rtl_p): New decl.
5343 * cgraphunit.c (symtab_node::native_rtl_p): New function.
5344 (symtab_node::needed_p): Don't assert for early assembly output
5345 for __RTL functions.
5346 (cgraph_node::finalize_function): Set "force_output" for __RTL
5348 (cgraph_node::analyze): Bail out early for __RTL functions.
5349 (analyze_functions): Update assertion to support __RTL functions.
5350 (cgraph_node::expand): Bail out early for __RTL functions.
5351 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
5353 * function.h (struct function): Update comment for field
5355 * gimple-expr.c: Include "tree-pass.h".
5356 (gimple_has_body_p): Return false for __RTL functions.
5357 * Makefile.in (OBJS): Add run-rtl-passes.o.
5358 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
5360 (gcc::pass_manager::get_clean_slate): New accessor.
5361 * passes.c: Include "insn-addr.h".
5362 (should_skip_pass_p): Add logging. Update logic for running
5363 "expand" to be compatible with both __GIMPLE and __RTL. Guard
5364 property-provider override so it is only done for gimple passes.
5366 (skip_pass): New function.
5367 (execute_one_pass): Call skip_pass when skipping passes.
5368 * read-md.c (md_reader::read_char): Support filtering
5369 the input to a subset of line numbers.
5370 (md_reader::md_reader): Initialize fields
5371 m_first_line and m_last_line.
5372 (md_reader::read_file_fragment): New function.
5373 * read-md.h (md_reader::read_file_fragment): New decl.
5374 (md_reader::m_first_line): New field.
5375 (md_reader::m_last_line): New field.
5376 * read-rtl-function.c (function_reader::create_function): Only
5377 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
5378 curr_properties. Set DECL_INITIAL to a dummy block.
5379 (read_rtl_function_body_from_file_range): New function.
5380 * read-rtl-function.h (read_rtl_function_body_from_file_range):
5382 * run-rtl-passes.c: New file.
5383 * run-rtl-passes.h: New file.
5385 2017-01-24 Jeff Law <law@redhat.com>
5387 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
5390 2017-01-24 Bin Cheng <bin.cheng@arm.com>
5392 PR tree-optimization/79159
5393 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
5394 (record_nonwrapping_iv): Improve boundary using above function if no
5395 value range information.
5397 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
5398 Martin Jambor <mjambor@suse.cz>
5400 * brig-builtins.def: New file.
5401 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
5402 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
5403 (DEF_HSAIL_SAT_BUILTIN): Likewise.
5404 (DEF_HSAIL_INTR_BUILTIN): Likewise.
5405 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
5406 * builtin-types.def (BT_INT8): New.
5407 (BT_INT16): Likewise.
5408 (BT_UINT8): Likewise.
5409 (BT_UINT16): Likewise.
5410 (BT_FN_ULONG): Likewise.
5411 (BT_FN_UINT_INT): Likewise.
5412 (BT_FN_UINT_ULONG): Likewise.
5413 (BT_FN_UINT_LONG): Likewise.
5414 (BT_FN_UINT_PTR): Likewise.
5415 (BT_FN_ULONG_PTR): Likewise.
5416 (BT_FN_INT8_FLOAT): Likewise.
5417 (BT_FN_INT16_FLOAT): Likewise.
5418 (BT_FN_UINT32_FLOAT): Likewise.
5419 (BT_FN_UINT16_FLOAT): Likewise.
5420 (BT_FN_UINT8_FLOAT): Likewise.
5421 (BT_FN_UINT64_FLOAT): Likewise.
5422 (BT_FN_UINT16_UINT32): Likewise.
5423 (BT_FN_UINT32_UINT16): Likewise.
5424 (BT_FN_UINT16_UINT16_UINT16): Likewise.
5425 (BT_FN_INT_PTR_INT): Likewise.
5426 (BT_FN_UINT_PTR_UINT): Likewise.
5427 (BT_FN_LONG_PTR_LONG): Likewise.
5428 (BT_FN_ULONG_PTR_ULONG): Likewise.
5429 (BT_FN_VOID_UINT64_UINT64): Likewise.
5430 (BT_FN_UINT8_UINT8_UINT8): Likewise.
5431 (BT_FN_INT8_INT8_INT8): Likewise.
5432 (BT_FN_INT16_INT16_INT16): Likewise.
5433 (BT_FN_INT_INT_INT): Likewise.
5434 (BT_FN_UINT_FLOAT_UINT): Likewise.
5435 (BT_FN_FLOAT_UINT_UINT): Likewise.
5436 (BT_FN_ULONG_UINT_UINT): Likewise.
5437 (BT_FN_ULONG_UINT_PTR): Likewise.
5438 (BT_FN_ULONG_ULONG_ULONG): Likewise.
5439 (BT_FN_UINT_UINT_UINT): Likewise.
5440 (BT_FN_VOID_UINT_PTR): Likewise.
5441 (BT_FN_UINT_UINT_PTR: Likewise.
5442 (BT_FN_UINT32_UINT64_PTR): Likewise.
5443 (BT_FN_INT_INT_UINT_UINT): Likewise.
5444 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
5445 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
5446 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
5447 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
5448 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
5449 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
5450 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
5451 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
5452 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
5453 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
5454 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
5455 * doc/frontends.texi: List BRIG FE.
5456 * doc/install.texi (Testing): Add BRIG tesring requirements.
5457 * doc/invoke.texi (Overall Options): Mention BRIG.
5458 * doc/standards.texi (Standards): Doucment BRIG HSA version.
5460 2017-01-24 Richard Biener <rguenther@suse.de>
5462 PR translation/79208
5463 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
5465 2017-01-24 Martin Jambor <mjambor@suse.cz>
5468 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
5469 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
5472 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
5475 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
5476 casts from signed to unsigned really don't have a range.
5478 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
5480 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
5481 GMP_RNDx for compatiblity.
5483 2017-01-24 Martin Liska <mliska@suse.cz>
5486 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
5487 that would prevent us to call alloca with -1 as argument.
5489 2017-01-24 Jakub Jelinek <jakub@redhat.com>
5491 * dwarf2out.c (output_compilation_unit_header, output_file_names):
5492 Avoid -Wformat-security warning.
5494 2017-01-23 Andrew Pinski <apinski@cavium.com>
5496 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
5499 2017-01-23 Martin Sebor <msebor@redhat.com>
5502 * gimple-ssa-sprintf.c (warn_level): New global.
5503 (format_integer): Use it here and throughout the rest of the file.
5504 Use the same switch to compute sign as base.
5505 (maybe_warn): New function.
5506 (format_directive): Factor out warnings into maybe_warn.
5507 Add debugging output. Use warn_level.
5508 (add_bytes): Use warn_level.
5509 (pass_sprintf_length::compute_format_length): Add debugging output.
5510 (try_substitute_return_value): Same.
5511 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
5514 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
5515 (struct fmtresult, format_integer, format_floating): Adjust.
5516 (fmtresult::fmtresult): Set max correctly in two argument ctor.
5517 (get_string_length, format_string,format_directive): Same.
5518 (pass_sprintf_length::compute_format_length): Same.
5519 (try_substitute_return_value): Simplify slightly.
5522 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
5523 (fmtresult::operator+=): Outlined.
5524 (struct fmtresult): Add ctors.
5525 (struct conversion_spec): Rename...
5526 (struct directive): ...to this. Add and remove data members.
5527 (directive::set_width, directive::set_precision): New functions.
5528 (format_percent): Use fmtresult ctor.
5529 (get_width_and_precision): Remove.
5530 (format_integer): Make naming changes. Avoid computing width and
5532 (format_floating): Same. Adjust indentation.
5533 (format_character, format_none): New functions.
5534 (format_string): Moved character handling to format_character.
5535 (format_directive): Remove arguments, change return type.
5536 (parse_directive): New function.
5537 (pass_sprintf_length::compute_format_length): Move directive
5538 parsing to parse_directive.
5540 2017-01-23 Jakub Jelinek <jakub@redhat.com>
5542 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
5543 (assign_assembler_name_if_needed): ... this.
5544 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
5545 (assign_assembler_name_if_needed): ... this.
5546 (free_lang_data_in_cgraph): Adjust callers.
5547 * cgraphunit.c (cgraph_node::analyze): Likewise.
5548 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
5551 2017-01-23 Richard Biener <rguenther@suse.de>
5553 PR tree-optimization/79088
5554 PR tree-optimization/79188
5555 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
5556 resetting loop bounds after last path deletion. Reset loop
5557 bounds of the target loop, make code match the comments.
5558 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
5559 Make sure loops need no fixups.
5561 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
5563 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
5564 exponent support with double type for first argument.
5565 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
5566 type returned by __builtin_vec_extract_sig,
5567 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
5568 functions from "vector int" to "vector unsigned int" or from
5569 "vector long long int" to "vector unsigned long long int".
5570 Changed type returned by __builtin_vec_extract_exp,
5571 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
5572 functions from "vector int" to "vector unsigned int" or from
5573 "vector long long int" to "vector unsigned long long int".
5574 Changed return type of __builtin_vec_test_data_class,
5575 __builtin_vec_test_data_class_sp, and
5576 __builtin_vec_test_data_class_dp from "vector int" to
5577 "vector bool int" or from "vector long long int" to "vector bool
5578 long long int" and changed second argument type from "unsigned
5579 int" to "int". Added new overloaded function forms "vector float
5580 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
5581 "vector float __builtin_vec_insert_exp_sp (vector float, vector
5582 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
5583 double, vector unsigned long long int)" and "vector double
5584 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
5585 long int)". Changed return type of
5586 __builtin_scalar_test_data_class and
5587 __builtin_scalar_test_data_class_sp and
5588 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
5589 int" and changed second argument from "unsigned int" to "int".
5590 Changed type returned by __builtin_scalar_test_neg,
5591 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
5592 from "int" to "bool int". Added new overloaded function form
5593 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
5594 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
5595 exponent double-precision with floating point first argument.
5596 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
5597 documentation of scalar_test_data_class, scalar_test_neg,
5598 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
5599 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
5600 vec_test_data_class built-in functions to reflect refinements in
5601 their type signatures.
5603 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
5605 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
5607 (aarch64_elf_asm_destructor): Likewise.
5609 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
5611 PR rtl-optimization/78634
5612 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
5613 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
5614 * ifcvt.c (noce_try_cmove): Add missing cost check.
5616 PR rtl-optimization/71724
5617 * combine.c (if_then_else_cond): Look for situations where it is
5618 beneficial to undo the work of one of the recursive calls.
5620 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5622 PR tree-optimization/70754
5623 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5624 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5625 combined stmt before it if not NULL.
5626 (combine_chains): Process refs reversely and compute dominance point
5629 2017-01-23 Martin Liska <mliska@suse.cz>
5631 PR tree-optimization/79196
5632 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
5633 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
5635 (strlen_optimize_stmt): Call the renamed function.
5637 2017-01-23 Michael Matz <matz@suse.de>
5639 PR tree-optimization/78384
5640 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
5642 2017-01-23 Richard Biener <rguenther@suse.de>
5644 PR tree-optimization/79186
5645 * tree-vrp.c (register_new_assert_for): Make sure we've seen
5646 both incoming edges before moving an assert.
5648 2017-01-23 Martin Jambor <mjambor@suse.cz>
5650 * ipa-prop.c (load_from_param_1): Removed.
5651 (load_from_unmodified_param): Bits from load_from_param_1 put back
5653 (load_from_param): Removed.
5654 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
5655 with stmt. Reverted back to use of load_from_unmodified_param.
5657 2017-01-23 Martin Jambor <mjambor@suse.cz>
5660 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
5661 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
5662 field a pointer to garbage collected vector, mark lattices and
5663 ipcp_orig_node with GTY((skip)).
5664 (ipa_get_param_count): Adjust to descriptors being a pointer.
5665 (ipa_get_param): Likewise.
5666 (ipa_get_type): Likewise.
5667 (ipa_get_param_move_cost): Likewise.
5668 (ipa_set_param_used): Likewise.
5669 (ipa_get_controlled_uses): Likewise.
5670 (ipa_set_controlled_uses): Likewise.
5671 (ipa_is_param_used): Likewise.
5672 (ipa_node_params_t): Move into garbage collector. New methods insert
5674 (ipa_node_params_sum): Annotate wth GTY(()).
5675 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
5677 (ipa_load_from_parm_agg): Adjust declaration.
5678 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
5679 * ipa-profile.c (ipa_profile): Likewise.
5680 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
5681 (ipa_populate_param_decls): Make descriptors parameter garbage
5683 (ipa_dump_param): Adjust to descriptors being a pointer.
5684 (ipa_alloc_node_params): Likewise.
5685 (ipa_initialize_node_params): Likewise.
5686 (load_from_param_1): Make descriptors parameter garbage collected.
5687 (load_from_unmodified_param): Likewise.
5688 (load_from_param): Likewise.
5689 (ipa_load_from_parm_agg): Likewise.
5690 (ipa_node_params::~ipa_node_params): Removed.
5691 (ipa_free_all_node_params): Remove call to delete operator.
5692 (ipa_node_params_t::insert): New.
5693 (ipa_node_params_t::remove): Likewise.
5694 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
5695 copy known_csts and known_contexts vectors.
5696 (ipa_read_node_info): Adjust to descriptors being a pointer.
5697 (ipcp_modif_dom_walker): Make m_descriptors field garbage
5699 (ipcp_transform_function): Make descriptors variable garbage
5702 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
5704 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
5705 * config/i386/avx512dqintrin.h: Ditto.
5706 * config/i386/avx512fintrin.h: Ditto.
5707 * gcc/config/i386/i386.c: Handle new builtins.
5708 * config/i386/i386-builtin.def: Add new builtins.
5709 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
5710 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
5712 2017-01-23 Jakub Jelinek <jakub@redhat.com>
5713 Martin Liska <mliska@suse.cz>
5715 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
5716 * asan.c (asan_expand_poison_ifn): Support stores and use
5717 appropriate ASAN report function.
5718 * internal-fn.c (expand_ASAN_POISON_USE): New function.
5719 * internal-fn.def (ASAN_POISON_USE): Declare.
5720 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
5721 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
5722 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
5723 ASAN_POISON calls w/o LHS.
5724 * tree-ssa.c (execute_update_addresses_taken): Create clobber
5725 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
5726 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
5727 * gimplify.c (asan_poison_variables): Add attribute
5728 use_after_scope_memory to variables that really needs to live
5730 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
5731 having the attribute.
5733 2017-01-23 Martin Liska <mliska@suse.cz>
5735 * asan.c (create_asan_shadow_var): New function.
5736 (asan_expand_poison_ifn): Likewise.
5737 * asan.h (asan_expand_poison_ifn): New declaration.
5738 * internal-fn.c (expand_ASAN_POISON): Likewise.
5739 * internal-fn.def (ASAN_POISON): New builtin.
5740 * sanopt.c (pass_sanopt::execute): Expand
5741 asan_expand_poison_ifn.
5742 * tree-inline.c (copy_decl_for_dup_finish): Make function
5744 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
5745 * tree-ssa.c (is_asan_mark_p): New function.
5746 (execute_update_addresses_taken): Rewrite local variables
5747 (identified just by use-after-scope as addressable) into SSA.
5749 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
5751 * doc/install.texi (Specific): opensource.apple.com uses https
5752 now. Remove trailing slash.
5754 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
5756 * README.Portability: Remove note on an Irix compatibility issue.
5758 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
5760 * gcov.c (INCLUDE_ALGORITHM): Define.
5761 (INCLUDE_VECTOR): Define.
5762 No longer include <vector> and <algorithm> directly.
5764 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
5766 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
5768 * doc/invoke.texi (Code Gen Options): Ditto.
5770 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
5773 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
5775 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
5777 rtl-optimization/79125
5778 * cprop.c (local_cprop_pass): Handle cases where we make an
5781 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
5785 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
5786 read from, for big endian.
5788 2017-01-20 Jiong Wang <jiong.wang@arm.com>
5790 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
5791 register pauth builtins for LP64 only.
5793 2017-01-20 Marek Polacek <polacek@redhat.com>
5796 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
5799 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
5801 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
5803 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
5804 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
5805 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
5807 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5810 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
5811 in big-endian mode when they are not a single duplicated value.
5813 2017-01-20 Richard Biener <rguenther@suse.de>
5815 * BASE-VER: Bump to 7.0.1.
5817 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
5819 * omp-low.c (omplow_simd_context): New struct. Use it...
5820 (lower_rec_simd_input_clauses): ...here and...
5821 (lower_rec_input_clauses): ...here to hold common data. Adjust all
5822 references to idx, lane, max_vf, is_simt.
5824 2017-01-20 Graham Markall <graham.markall@embecosm.com>
5826 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
5829 2017-01-20 Martin Jambor <mjambor@suse.cz>
5831 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
5832 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
5834 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
5835 (GTFILES): Rename hsa.c to hsa-common.c.
5836 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
5837 * hsa-dump.c: Likewise.
5838 * hsa-gen.c: Likewise.
5839 * hsa-regalloc.c: Likewise.
5840 * ipa-hsa.c: Likewise.
5841 * omp-expand.c: Likewise.
5842 * omp-low.c: Likewise.
5843 * toplev.c: Likewise.
5845 2017-01-20 Marek Polacek <polacek@redhat.com>
5848 * doc/invoke.texi: Document -Wduplicated-branches.
5849 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
5850 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
5851 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
5852 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
5853 return 0 only when not OEP_LEXICOGRAPHIC.
5854 (fold_build_cleanup_point_expr): Use the expression
5855 location when building CLEANUP_POINT_EXPR.
5856 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
5857 * tree.c (add_expr): Handle error_mark_node.
5859 2017-01-20 Martin Liska <mliska@suse.cz>
5862 * tree-profile.c (init_ic_make_global_vars): Do not call
5864 (gimple_init_gcov_profiler): Likewise.
5866 2017-01-20 Martin Liska <mliska@suse.cz>
5869 * cgraph.h (maybe_create_reference): Remove argument and
5871 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
5873 * ipa-cp.c (create_specialized_node): Likewise.
5874 * symtab.c (symtab_node::maybe_create_reference): Handle
5875 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
5877 2017-01-20 Martin Liska <mliska@suse.cz>
5879 * read-rtl-function.c (function_reader::create_function): Use
5880 build_decl instread of build_decl_stat.
5882 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
5884 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
5885 * config/i386/avx512dqintrin.h: Ditto.
5886 * config/i386/avx512fintrin.h: Ditto.
5887 * config/i386/i386-builtin-types.def: Add new types.
5888 * gcc/config/i386/i386.c: Handle new types.
5889 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
5890 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
5891 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
5892 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
5893 (__builtin_ia32_kshiftridi): New.
5894 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
5896 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
5900 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
5901 define to rs6000_init_stack_protect_guard.
5902 (rs6000_init_stack_protect_guard): New function.
5904 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
5905 Yunqiang Su <yunqiang.su@imgtec.com>
5907 * config.gcc (supported_defaults): Add madd4.
5908 (with_madd4): Add validation.
5909 (all_defaults): Add madd4.
5910 * config/mips/mips.opt (mmadd4): New option.
5911 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
5913 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
5915 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
5916 (ISA_HAS_FUSED_MADD4): Likewise.
5917 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
5918 * gcc/doc/install.texi (--with-madd4): Document the new option.
5920 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5922 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
5923 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
5924 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
5925 (aarch64_init_pauth_hint_builtins): New.
5926 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
5927 (aarch64_expand_builtin): Expand new builtins.
5929 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5931 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
5932 * combine-stack-adj.c (no_unhandled_cfa): Handle
5933 REG_CFA_TOGGLE_RA_MANGLE.
5934 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
5935 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
5936 info for return address signing.
5937 (aarch64_expand_epilogue): Likewise.
5939 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5941 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
5942 * config/aarch64/aarch64-protos.h
5943 (aarch64_return_address_signing_enabled): New declaration.
5944 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
5946 (aarch64_expand_prologue): Sign return address before it's pushed onto
5948 (aarch64_expand_epilogue): Authenticate return address fetched from
5950 (aarch64_override_options): Sanity check for ILP32 and ISA level.
5951 (aarch64_attributes): New function attributes for "sign-return-address".
5952 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
5953 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
5954 ("*do_return"): Generate combined instructions according to key index.
5955 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
5956 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
5958 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
5959 * config/aarch64/aarch64.opt (msign-return-address=): New.
5960 * doc/extend.texi (AArch64 Function Attributes): Documents
5961 "sign-return-address=".
5962 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
5964 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
5966 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
5967 overall option summary.
5969 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5971 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
5972 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
5973 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
5974 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
5976 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
5978 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
5979 -mpower9-minmax by default for -mcpu=power9.
5980 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
5981 128-bit floating point.
5983 2017-01-20 Alan Modra <amodra@gmail.com>
5985 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
5986 optimizing for size.
5988 2017-01-20 Alan Modra <amodra@gmail.com>
5991 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
5992 for strcmp and strncmp from corresponding builtin decl.
5994 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
5996 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
5997 instead of i386/rtems-64.h.
5998 * config/i386/rtems-64.h: Remove.
6000 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
6004 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
6006 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
6008 2017-01-19 Tamar Christina <tamar.christina@arm.com>
6010 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6011 Change int to HOST_WIDE_INT.
6012 * config/aarch64/aarch64-protos.h
6013 (aarch64_simd_gen_const_vector_dup): Likewise.
6014 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
6016 2017-01-19 David Malcolm <dmalcolm@redhat.com>
6018 * langhooks-def.h (lhd_type_for_size): New decl.
6019 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
6020 * langhooks.c (lhd_type_for_size): New function, taken from
6023 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
6025 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
6026 define_bypass for CR latency.
6027 (power9-cracked-alu): Update bypass latency and remove power9-branch.
6028 (power9-alu2): Add define_bypass for CR latency.
6030 (power9-mul): Update insn latency.
6031 (power9-mul-compare): Update insn latency, bypass latency and remove
6034 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6036 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
6038 * config/aarch64/aarch64.md
6039 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
6040 aarch64_nopcrelative_literal_loads.
6041 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
6043 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
6045 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
6047 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
6049 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
6052 * config.gcc (supported_defaults): Add lxc1-sxc1.
6053 (with_lxc1_sxc1): Add validation.
6054 (all_defaults): Add lxc1-sxc1.
6055 * config/mips/mips.opt (mlxc1-sxc1): New option.
6056 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
6058 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
6059 __mips_no_lxc1_sxc1.
6060 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
6061 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
6062 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
6064 2017-01-19 Richard Biener <rguenther@suse.de>
6066 PR tree-optimization/72488
6067 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
6068 sure to restore SSA info.
6069 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
6071 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
6073 PR rtl-optimization/79121
6074 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
6075 of the inner type when shifting an extended value.
6077 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
6080 * symtab.c (symtab_node::equal_address_to): Fix comparing of
6081 interposable aliases.
6083 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
6086 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
6087 Use the evmergelohi instruction.
6088 (mov_si<mode>_e500_subreg4_2_le): Likewise.
6089 (mov_sitf_e500_subreg8_2_be): Likewise.
6090 (mov_sitf_e500_subreg12_2_le): Likewise.
6091 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
6092 (mov_si<mode>_e500_subreg4_2_be): Likewise.
6093 (mov_sitf_e500_subreg8_2_le): Likewise.
6094 (mov_sitf_e500_subreg12_2_be): Likewise.
6096 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6098 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
6099 attribute from vecsimple to vecperm.
6100 (altivec_vbpermq2): Likewise.
6102 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6105 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
6107 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6108 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
6109 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
6110 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
6111 case where N arg is SIZE_MAX.
6112 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
6113 (cmpstrsi): Add pattern.
6115 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
6117 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6118 __builtin_vec_revb builtins.
6119 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
6120 built-in functions to support generation of the ISA 3.0 XXBR<x>
6121 vector byte reverse instructions.
6122 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
6123 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
6124 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
6125 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
6126 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
6127 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
6128 (P9V_BUILTIN_VEC_REVB): Likewise.
6129 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
6130 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
6131 (p9_xxbrq_v16qi): Likewise.
6132 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
6133 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
6134 (p9_xxbrh_v8hi): Likewise.
6135 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
6136 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
6137 vec_revb built-in functions.
6139 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
6141 PR rtl-optimization/78952
6142 * config/i386/i386.md (any_extract): New code iterator.
6143 (*insvqi_2): Use any_extract for source operand.
6144 (*insvqi_3): Use any_shiftrt for source operand.
6146 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
6148 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
6150 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
6152 2017-01-18 Matthias Klose <doko@ubuntu.com>
6154 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
6156 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6158 * config/rs6000/altivec.h (vec_bperm): Change #define.
6159 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
6160 (altivec_vbpermq2): New define_insn.
6161 (altivec_vbpermd): Likewise.
6162 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
6164 (VBPERMD): Likewise.
6165 (VBPERM): New polymorphic function interface.
6166 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
6167 Add entries for P9V_BUILTIN_VEC_VBPERM.
6168 * doc/extend.texi: Add interfaces for vec_bperm.
6170 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6172 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
6173 first letter of error messages.
6174 (s390_resolve_overloaded_builtin): Likewise.
6175 * config/s390/s390.c (s390_expand_builtin): Likewise.
6176 (s390_invalid_arg_for_unprototyped_fn): Likewise.
6177 (s390_valid_target_attribute_inner_p): Likewise.
6178 * config/s390/s390.md ("tabort"): Likewise.
6180 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
6182 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
6183 (ISA_AVOID_DIV_HILO): New macro.
6184 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
6185 (ISA_HAS_DDIV): Likewise.
6187 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6189 * doc/invoke.texi (fabi-version): Correct number of occurrences.
6191 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6193 * doc/invoke.texi (fabi-version): Spelling fix.
6195 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6198 * doc/invoke.texi (fabi-version): Mention mangling fix for
6201 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6204 * doc/invoke.texi (fabi-version): Document discriminator mangling.
6206 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
6209 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
6210 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
6212 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
6214 (stack_protect_test): Ditto.
6215 * config/rs6000/rs6000.opt (mstack-protector-guard=,
6216 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
6218 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
6219 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
6220 -mstack-protector-guard-offset=.
6221 (RS/6000 and PowerPC Options): Ditto.
6223 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
6225 * config/i386/i386.h (MASK_CLASS_P): New define.
6226 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
6227 there are no registers from different register sets also when
6228 mask registers are used. Update function comment.
6229 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
6230 to (*k/*r) and (*k/*km) alternatives.
6232 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
6234 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
6235 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
6236 (EH_RETURN_HANDLER_RTX): New define.
6237 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
6238 Force frame pointer in EH return functions.
6239 (aarch64_expand_epilogue): Add barrier for eh_return.
6240 (aarch64_final_eh_return_addr): Remove.
6241 (aarch64_eh_return_handler_rtx): New function.
6242 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
6244 (aarch64_eh_return_handler_rtx): New prototype.
6246 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6248 * config/rs6000/altivec.h (vec_rlmi): New #define.
6249 (vec_vrlnm): Likewise.
6250 (vec_rlnm): Likewise.
6251 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
6252 (UNSPEC_VRLNM): Likewise.
6253 (VIlong): New mode iterator.
6254 (altivec_vrl<VI_char>mi): New define_insn.
6255 (altivec_vrl<VI_char>nm): Likewise.
6256 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
6259 (RLNM): New polymorphic function entry.
6260 (VRLWMI): New monomorphic function entry.
6262 (RLMI): New polymorphic function entry.
6263 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
6264 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
6265 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
6268 2017-01-17 Jakub Jelinek <jakub@redhat.com>
6271 * dwarf2out.c (field_byte_offset): Restore the
6272 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
6273 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
6274 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
6277 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
6280 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
6282 2017-01-17 Jakub Jelinek <jakub@redhat.com>
6285 * dwarf2out.c (add_data_member_location_attribute): For constant
6286 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
6287 instead of DW_AT_data_member_location, DW_AT_bit_offset and
6288 DW_AT_byte_size attributes.
6290 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
6292 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
6293 after forcing to constant memory when the code model is medium.
6295 2017-01-17 Julia Koval <julia.koval@intel.com>
6298 * config/i386/avx512fintrin.h
6299 (_mm512_i32gather_ps): Change __addr type to void const*.
6300 (_mm512_mask_i32gather_ps): Ditto.
6301 (_mm512_i32gather_pd): Ditto.
6302 (_mm512_mask_i32gather_pd): Ditto.
6303 (_mm512_i64gather_ps): Ditto.
6304 (_mm512_mask_i64gather_ps): Ditto.
6305 (_mm512_i64gather_pd): Ditto.
6306 (_mm512_mask_i64gather_pd): Ditto.
6307 (_mm512_i32gather_epi32): Ditto.
6308 (_mm512_mask_i32gather_epi32): Ditto.
6309 (_mm512_i32gather_epi64): Ditto.
6310 (_mm512_mask_i32gather_epi64): Ditto.
6311 (_mm512_i64gather_epi32): Ditto.
6312 (_mm512_mask_i64gather_epi32): Ditto.
6313 (_mm512_i64gather_epi64): Ditto.
6314 (_mm512_mask_i64gather_epi64): Ditto.
6315 (_mm512_i32scatter_ps): Change __addr type to void*.
6316 (_mm512_mask_i32scatter_ps): Ditto.
6317 (_mm512_i32scatter_pd): Ditto.
6318 (_mm512_mask_i32scatter_pd): Ditto.
6319 (_mm512_i64scatter_ps): Ditto.
6320 (_mm512_mask_i64scatter_ps): Ditto.
6321 (_mm512_i64scatter_pd): Ditto.
6322 (_mm512_mask_i64scatter_pd): Ditto.
6323 (_mm512_i32scatter_epi32): Ditto.
6324 (_mm512_mask_i32scatter_epi32): Ditto.
6325 (_mm512_i32scatter_epi64): Ditto.
6326 (_mm512_mask_i32scatter_epi64): Ditto.
6327 (_mm512_i64scatter_epi32): Ditto.
6328 (_mm512_mask_i64scatter_epi32): Ditto.
6329 (_mm512_i64scatter_epi64): Ditto.
6330 (_mm512_mask_i64scatter_epi64): Ditto.
6331 * config/i386/avx512pfintrin.h
6332 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
6333 (_mm512_mask_prefetch_i32gather_ps): Ditto.
6334 (_mm512_mask_prefetch_i64gather_pd): Ditto.
6335 (_mm512_mask_prefetch_i64gather_ps): Ditto.
6336 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
6337 (_mm512_prefetch_i32scatter_ps): Ditto.
6338 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
6339 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
6340 (_mm512_prefetch_i64scatter_pd): Ditto.
6341 (_mm512_prefetch_i64scatter_ps): Ditto.
6342 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
6343 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
6344 * config/i386/avx512vlintrin.h
6345 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
6346 (_mm_mmask_i32gather_ps): Ditto.
6347 (_mm256_mmask_i32gather_pd): Ditto.
6348 (_mm_mmask_i32gather_pd): Ditto.
6349 (_mm256_mmask_i64gather_ps): Ditto.
6350 (_mm_mmask_i64gather_ps): Ditto.
6351 (_mm256_mmask_i64gather_pd): Ditto.
6352 (_mm_mmask_i64gather_pd): Ditto.
6353 (_mm256_mmask_i32gather_epi32): Ditto.
6354 (_mm_mmask_i32gather_epi32): Ditto.
6355 (_mm256_mmask_i32gather_epi64): Ditto.
6356 (_mm_mmask_i32gather_epi64): Ditto.
6357 (_mm256_mmask_i64gather_epi32): Ditto.
6358 (_mm_mmask_i64gather_epi32): Ditto.
6359 (_mm256_mmask_i64gather_epi64): Ditto.
6360 (_mm_mmask_i64gather_epi64): Ditto.
6361 (_mm256_i32scatter_ps): Change __addr type to void*.
6362 (_mm256_mask_i32scatter_ps): Ditto.
6363 (_mm_i32scatter_ps): Ditto.
6364 (_mm_mask_i32scatter_ps): Ditto.
6365 (_mm256_i32scatter_pd): Ditto.
6366 (_mm256_mask_i32scatter_pd): Ditto.
6367 (_mm_i32scatter_pd): Ditto.
6368 (_mm_mask_i32scatter_pd): Ditto.
6369 (_mm256_i64scatter_ps): Ditto.
6370 (_mm256_mask_i64scatter_ps): Ditto.
6371 (_mm_i64scatter_ps): Ditto.
6372 (_mm_mask_i64scatter_ps): Ditto.
6373 (_mm256_i64scatter_pd): Ditto.
6374 (_mm256_mask_i64scatter_pd): Ditto.
6375 (_mm_i64scatter_pd): Ditto.
6376 (_mm_mask_i64scatter_pd): Ditto.
6377 (_mm256_i32scatter_epi32): Ditto.
6378 (_mm256_mask_i32scatter_epi32): Ditto.
6379 (_mm_i32scatter_epi32): Ditto.
6380 (_mm_mask_i32scatter_epi32): Ditto.
6381 (_mm256_i32scatter_epi64): Ditto.
6382 (_mm256_mask_i32scatter_epi64): Ditto.
6383 (_mm_i32scatter_epi64): Ditto.
6384 (_mm_mask_i32scatter_epi64): Ditto.
6385 (_mm256_i64scatter_epi32): Ditto.
6386 (_mm256_mask_i64scatter_epi32): Ditto.
6387 (_mm_i64scatter_epi32): Ditto.
6388 (_mm_mask_i64scatter_epi32): Ditto.
6389 (_mm256_i64scatter_epi64): Ditto.
6390 (_mm256_mask_i64scatter_epi64): Ditto.
6391 (_mm_i64scatter_epi64): Ditto.
6392 (_mm_mask_i64scatter_epi64): Ditto.
6393 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
6394 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
6395 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
6396 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
6397 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
6398 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
6399 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
6400 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
6401 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
6402 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
6403 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
6404 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
6405 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
6406 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
6407 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
6408 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
6409 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
6410 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
6411 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
6412 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
6413 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
6414 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
6415 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
6416 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
6417 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
6418 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
6419 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
6420 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
6421 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
6422 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
6423 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
6424 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
6425 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
6426 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
6427 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
6428 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
6429 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
6430 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
6431 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
6432 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
6433 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
6434 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
6435 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
6436 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
6437 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
6438 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
6439 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
6440 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
6441 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
6442 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
6443 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
6444 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
6445 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
6446 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
6447 definitions accordingly.
6449 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
6450 Kuan-Lin Chen <kuanlinchentw@gmail.com>
6453 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
6456 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
6459 * ira-conflicts.c (ira_build_conflicts): Update total conflict
6460 hard regs for inner regno.
6462 2017-01-17 Martin Liska <mliska@suse.cz>
6465 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
6466 assumption and add comment.
6468 2017-01-17 Nathan Sidwell <nathan@acm.org>
6470 * ipa-visibility.c (localize_node): New function, broken out of ...
6471 (function_and_variable_visibility): ... here. Call it.
6473 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
6476 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
6477 correctly set frequency of oudgoing edge.
6478 (duplicate_thread_path): Fix profile updating.
6480 2017-01-17 Jakub Jelinek <jakub@redhat.com>
6483 * configure.ac: Add GCC_BASE_VER.
6484 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
6485 version from BASE-VER file.
6486 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
6487 (gcc.o): Depend on $(BASEVER).
6488 * common.opt (dumpfullversion): New option.
6489 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
6490 * doc/invoke.texi: Document -dumpfullversion.
6491 * doc/install.texi: Document --with-gcc-major-version-only.
6492 * configure: Regenerated.
6494 2017-01-17 Richard Biener <rguenther@suse.de>
6496 PR tree-optimization/71433
6497 * tree-vrp.c (register_new_assert_for): Merge same asserts
6498 on all incoming edges.
6499 (process_assert_insertions_for): Handle insertions at the
6502 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
6504 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
6505 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
6507 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
6510 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
6513 2017-01-17 Alan Modra <amodra@gmail.com>
6516 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
6517 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
6518 symbolic stack limit when pic.
6520 2017-01-16 Martin Sebor <msebor@redhat.com>
6522 PR tree-optimization/78608
6523 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
6525 2017-01-16 Jeff Law <law@redhat.com>
6528 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
6529 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
6530 for several include directories that may be relative to sysroot.
6531 * config/i386/x-mingw32 (gplus_includedir): Define.
6532 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
6533 (native_system_includedir): Likewise.
6534 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
6535 override if TARGET_SYSTEM_ROOT is defined.
6536 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
6538 PR tree-optimization/79090
6539 PR tree-optimization/33562
6540 PR tree-optimization/61912
6541 PR tree-optimization/77485
6542 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
6543 and computed trims into the dump file.
6545 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
6547 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
6549 2017-01-16 Jakub Jelinek <jakub@redhat.com>
6552 * gimplify.c (gimplify_init_constructor): If want_value and
6553 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
6557 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
6558 sequence. Formatting fixes.
6559 (doloop_optimize): Formatting fixes.
6562 * gcc.c (debug_level_greater_than_spec_func): New function.
6563 (static_spec_functions): Add debug-level-gt spec function.
6564 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
6566 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
6567 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
6568 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
6569 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
6570 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
6571 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
6573 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
6575 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
6576 QImode fixups to general and mask registers only.
6578 2017-01-16 Carl Love <cel@us.ibm.com>
6580 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
6581 for built-in functions
6582 vector signed char vec_nabs (vector signed char)
6583 vector signed short vec_nabs (vector signed short)
6584 vector signed int vec_nabs (vector signed int)
6585 vector signed long long vec_nabs (vector signed long long)
6586 vector float vec_nabs (vector float)
6587 vector double vec_nabs (vector double)
6588 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
6590 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
6591 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
6592 * doc/extend.texi: Update the documentation file for the new built-in
6595 2017-01-16 Martin Sebor <msebor@redhat.com>
6597 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
6600 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6602 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
6603 UNSPEC_VSX__XXSPLTD to require special splat handling.
6605 2017-01-16 David Malcolm <dmalcolm@redhat.com>
6608 * system.h: Poison strndup.
6610 2017-01-16 Alan Modra <amodra@gmail.com>
6613 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
6616 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
6618 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
6620 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
6622 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
6623 call recog here. Assert that INSN_CODE (insn) is non-negative.
6625 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
6628 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
6630 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
6631 in the currently scheduled RTL fragment.
6633 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
6635 PR rtl-optimization/78751
6636 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
6639 2017-01-14 Jeff Law <law@redhat.com>
6641 PR tree-optimization/79090
6642 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
6643 variable length stores.
6644 (compute_trims): Delete dead assignment to *trim_tail.
6645 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
6648 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
6650 PR rtl-optimization/78626
6651 PR rtl-optimization/78727
6652 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
6653 of a block, and split such blocks after everything else is finished.
6655 2017-01-14 Alan Modra <amodra@gmail.com>
6658 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
6659 target legitimate_combined_insn.
6660 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
6661 (rs6000_legitimate_combined_insn): New function.
6662 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
6664 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
6665 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
6666 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
6668 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
6670 * doc/frontends.texi (G++ and GCC): Remove references to Java.
6672 2017-01-13 Jeff Law <law@redhat.com>
6674 PR tree-optimization/33562
6675 PR tree-optimization/61912
6676 PR tree-optimization/77485
6677 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
6679 (delete_dead_assignment): Likewise.
6680 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
6681 statement to delete_dead_call and delete_dead_assignment.
6683 2017-01-13 David Malcolm <dmalcolm@redhat.com>
6686 * substring-locations.c (format_warning_va): Strengthen case 1 so
6687 that both endpoints of the substring must be within the format
6688 range for just the substring to be printed.
6690 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
6692 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
6693 * config/i386/i386.c (ix86_target_string): Add missing options
6694 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
6695 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
6696 flags_other and ix86_target_other to flags2_other. Display unknown
6698 (ix86_valid_target_attribute_inner_p): Add missing options and
6699 reorder options by implied ISAs, as in ix86_target_string.
6701 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
6703 * hash-table.h (hash_table::too_empty_p): New function.
6704 (hash_table::expand): Use it.
6705 (hash_table::traverse): Likewise.
6706 (hash_table::empty_slot): Use sizeof (value_type) instead of
6707 sizeof (PTR) to convert bytes to elements. Shrink the table
6708 if the current size is excessive for the current number of
6711 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
6713 * ira-costs.c (record_reg_classes): Break from the inner loop
6714 early once alt_fail is known to be true. Update outer loop
6715 handling accordingly.
6717 2017-01-13 Jeff Law <law@redhat.com>
6719 * tree-ssa-dse.c (decrement_count): New function.
6720 (increment_start_addr, maybe_trim_memstar_call): Likewise.
6721 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
6722 when we know the partially dead statement is a mem* function.
6724 PR tree-optimization/61912
6725 PR tree-optimization/77485
6726 * tree-ssa-dse.c: Include expr.h.
6727 (maybe_trim_constructor_store): New function.
6728 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
6730 PR tree-optimization/33562
6731 PR tree-optimization/61912
6732 PR tree-optimization/77485
6733 * doc/invoke.texi: Document new dse-max-object-size param.
6734 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
6735 * tree-ssa-dse.c: Include params.h.
6736 (dse_store_status): New enum.
6737 (initialize_ao_ref_for_dse): New, partially extracted from
6739 (valid_ao_ref_for_dse, normalize_ref): New.
6740 (setup_live_bytes_from_ref, compute_trims): Likewise.
6741 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
6742 (maybe_trim_partially_dead_store): Likewise.
6743 (maybe_trim_complex_store): Likewise.
6744 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
6745 Track what bytes live from the original store. Return tri-state
6746 for dead, partially dead or live.
6747 (dse_dom_walker): Add constructor, destructor and new private members.
6748 (delete_dead_call, delete_dead_assignment): New extracted from
6750 (dse_optimize_stmt): Make a member of dse_dom_walker.
6751 Use initialize_ao_ref_for_dse.
6753 PR tree-optimization/33562
6754 PR tree-optimization/61912
6755 PR tree-optimization/77485
6756 * sbitmap.h (bitmap_count_bits): Prototype.
6757 (bitmap_clear_range, bitmap_set_range): Likewise.
6758 * sbitmap.c (bitmap_clear_range): New function.
6759 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
6761 2017-01-13 Martin Liska <mliska@suse.cz>
6764 * function.c (set_cfun): Add new argument force.
6765 * function.h (set_cfun): Likewise.
6766 * ipa-inline-transform.c (inline_call): Use the function when
6767 strict alising from is dropped for function we inline to.
6769 2017-01-13 Richard Biener <rguenther@suse.de>
6771 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
6772 for dumping GIMPLE INTEGER_CSTs.
6774 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6776 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
6777 to 201112L since C++17.
6779 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
6782 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
6783 if -fsanitize=kernel-address is present.
6785 2017-01-13 Richard Biener <rguenther@suse.de>
6787 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
6788 as _Literal ( type ) number in case usual suffixes do not
6789 preserve all information.
6791 2017-01-13 Richard Biener <rguenther@suse.de>
6793 PR tree-optimization/77283
6794 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
6795 and ssa-iterators.h.
6796 (is_feasible_trace): Implement a cost model based on joiner
6799 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6802 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
6803 char or short to __float128/_Float128 directly.
6805 2017-01-12 Martin Sebor <msebor@redhat.com>
6807 to -Wformat-overflow.
6808 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
6809 (min_bytes_remaining): Same.
6810 (get_string_length): Same.
6811 (format_string): Same.
6812 (format_directive): Same.
6814 (pass_sprintf_length::handle_gimple_call): Same.
6816 2017-01-12 Jakub Jelinek <jakub@redhat.com>
6818 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
6819 info.nowrite calls with no lhs that can't throw. Return bool
6820 whether gsi_remove has been called or not.
6821 (pass_sprintf_length::handle_gimple_call): Return bool whether
6822 try_substitute_return_value called gsi_remove. Formatting fix.
6823 (pass_sprintf_length::execute): Don't use gsi_remove if
6824 handle_gimple_call returned true.
6827 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
6828 be removed due to side-effects, don't remove following barrier nor
6829 turn the successor edge into fallthru edge.
6831 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6834 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
6835 element-reversing loads and stores as not swappable.
6837 2017-01-12 Nathan Sidwell <nathan@acm.org>
6838 Nicolai Stange <nicstange@gmail.com>
6840 * combine.c (try_combine): Don't ignore result of overlap checking
6841 loop. Combine overlap & asm check into single loop.
6843 2017-01-12 Richard Biener <rguenther@suse.de>
6845 * tree-pretty-print.c (dump_generic_node): Provide -gimple
6846 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
6848 2017-01-12 Richard Biener <rguenther@suse.de>
6850 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
6851 and TS_TARGET_OPTION directly derive from TS_BASE.
6852 * tree-core.h (tree_optimization_option): Derive from tree_base.
6853 (tree_target_option): Likewise.
6855 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
6857 * config/i386/i386.c (memory_address_length): Increase len
6858 only when rip_relative_addr_p returns false.
6860 2017-01-11 Julia Koval <julia.koval@intel.com>
6862 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
6863 (OPTION_MASK_ISA_SGX_SET): New.
6864 (ix86_handle_option): Handle OPT_msgx.
6865 * config.gcc: Added sgxintrin.h.
6866 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
6867 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
6868 * config/i386/i386.c (ix86_target_string): Add -msgx.
6870 (ix86_option_override_internal): Handle new options.
6871 (ix86_valid_target_attribute_inner_p): Add sgx.
6872 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
6873 * config/i386/i386.opt: Add msgx.
6874 * config/i386/sgxintrin.h: New file.
6875 * config/i386/x86intrin.h: Add sgxintrin.h.
6877 2017-01-11 Jakub Jelinek <jakub@redhat.com>
6880 * fold-const.c (maybe_nonzero_address): Return 1 for function
6882 (tree_single_nonzero_warnv_p): Don't handle function local objects
6886 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
6889 2017-01-11 David Malcolm <dmalcolm@redhat.com>
6892 * opts.c: Include "spellcheck.h"
6893 (struct string_fragment): New struct.
6894 (struct edit_distance_traits<const string_fragment &>): New
6896 (get_closest_sanitizer_option): New function.
6897 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
6899 2017-01-11 Jakub Jelinek <jakub@redhat.com>
6901 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
6903 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
6904 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
6905 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
6906 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
6907 for initial die_offset if dwarf_split_debug_info.
6908 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
6909 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
6911 (output_skeleton_debug_sections): Formatting fix. Use
6912 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
6913 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
6915 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
6917 * config/arm/cortex-a53.md: Add bypasses for
6919 (cortex_a53_r2f): Only use for transfers.
6920 (cortex_a53_f2r): Likewise.
6921 (cortex_a53_r2f_cvt): Add reservation for conversions.
6922 (cortex_a53_f2r_cvt): Likewise.
6924 2017-01-11 Tamar Christina <tamar.christina@arm.com>
6926 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
6927 to all inlined functions, change static to extern.
6929 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
6932 * config/arm/arm.c (legitimize_pic_address): Handle reference to
6934 (arm_assemble_integer): Likewise.
6936 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
6938 * config.gcc: Use new awk script to check CPU, FPU and architecture
6939 parameters for --with-... options.
6940 * config/arm/parsecpu.awk: New file
6941 * config/arm/arm-cpus.in: New file.
6942 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
6944 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
6946 * config/arm/t-arm: Update dependency rules.
6947 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
6948 of processing .def files.
6949 * config/arm/genopt.sh: Deleted.
6950 * config/arm/gentune.sh: Deleted.
6951 * config/arm/arm-cores.def: Deleted.
6952 * config/arm/arm-arches.def: Deleted.
6953 * config/arm/arm-fpus.def: Deleted.
6954 * config/arm/arm-tune.md: Regenerated.
6955 * config/arm/arm-tables.opt: Regenerated.
6956 * config/arm/arm-cpu.h: New generated file.
6957 * config/arm/arm-cpu-data.h: New generated file.
6958 * config/arm/arm-cpu-cdata.h: New generated file.
6960 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
6963 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
6965 (input_varpool_node): Unpack dynamically_initialized bit.
6967 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
6969 PR rtl-optimization/79032
6970 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
6971 the alignment of the adjusted memory reference against that of MODE,
6972 instead of the alignment of the original memory reference.
6974 2017-01-11 Martin Jambor <mjambor@suse.cz>
6976 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
6978 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
6979 decorated functions.
6981 2017-01-11 Richard Biener <rguenther@suse.de>
6983 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
6984 set range/nonnull info for PHI results. Do not set it on
6985 stmts marked for removal.
6987 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
6989 * expr.c (store_field): In the bitfield case, fetch the return value
6990 from the registers before applying a single big-endian adjustment.
6991 Always do a final load for a BLKmode value not larger than a word.
6993 2017-01-10 David Malcolm <dmalcolm@redhat.com>
6996 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
6997 that we correctly handle column numbers greater than
6998 LINE_MAP_MAX_COLUMN_NUMBER.
7000 2017-01-10 Martin Sebor <msebor@redhat.com>
7003 * gimple-ssa-sprintf.c (get_destination_size): Call
7004 {init,fini}object_sizes.
7005 * tree-object-size.c (addr_object_size): Adjust.
7006 (pass_through_call): Adjust.
7007 (pass_object_sizes::execute): Adjust.
7008 * tree-object-size.h (fini_object_sizes): Declare.
7010 2017-01-10 Martin Sebor <msebor@redhat.com>
7012 PR tree-optimization/78775
7013 * builtins.c (get_size_range): Move...
7014 * calls.c: ...to here.
7015 (alloc_max_size): Accept zero argument.
7016 (operand_signed_p): Remove.
7017 (maybe_warn_alloc_args_overflow): Call get_size_range.
7018 * calls.h (get_size_range): Declare.
7020 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
7022 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
7023 from TI's devices.csv file as of September 2016.
7024 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
7026 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
7028 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
7029 * doc/invoke.texi: Likewise.
7030 * doc/md.texi: Likewise.
7031 * doc/objc.texi: Likewise.
7033 2017-01-10 Joshua Conner <joshconner@google.com>
7035 * config/arm/fuchsia-elf.h: New file.
7036 * config/fuchsia.h: New file.
7037 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
7038 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
7040 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
7042 2016-01-10 Richard Biener <rguenther@suse.de>
7044 PR tree-optimization/79034
7045 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
7046 Propagate out degenerate PHIs in the joiner.
7048 2017-01-10 Martin Liska <mliska@suse.cz>
7050 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
7051 (sort_congruence_classes_by_decl_uid): Likewise.
7052 (sort_congruence_class_groups_by_decl_uid): Likewise.
7053 (sem_item_optimizer::merge_classes): Sort class, groups in these
7054 classes and members in the groups by DECL_UID of declarations.
7055 This would make merge operations stable.
7057 2017-01-10 Martin Liska <mliska@suse.cz>
7059 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
7060 usage of m_classes_vec.
7061 (sem_item_optimizer::~sem_item_optimizer): Likewise.
7062 (sem_item_optimizer::get_group_by_hash): Likewise.
7063 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
7064 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
7065 (sem_item_optimizer::verify_classes): Likewise.
7066 (sem_item_optimizer::process_cong_reduction): Likewise.
7067 (sem_item_optimizer::dump_cong_classes): Likewise.
7068 (sem_item_optimizer::merge_classes): Likewise.
7069 * ipa-icf.h (congruence_class_hash): Rename from
7070 congruence_class_group_hash. Remove declaration of m_classes_vec.
7072 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
7074 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
7075 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
7076 * config.gcc: Add avx512vpopcntdqintrin.h.
7077 * config/i386/avx512vpopcntdqintrin.h: New.
7078 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
7079 * config/i386/i386-builtin-types.def: Add new types.
7080 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
7081 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
7082 __builtin_ia32_vpopcountq_v8di_mask): New.
7083 * config/i386/i386-c.c (ix86_target_macros_internal): Define
7084 __AVX512VPOPCNTDQ__.
7085 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
7086 (PTA_AVX512VPOPCNTDQ): Define.
7087 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
7088 TARGET_AVX512VPOPCNTDQ_P): Define.
7089 * config/i386/i386.opt: Add mavx512vpopcntdq.
7090 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
7091 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
7093 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7096 * predict.def (PRED_CALL): Set to 67.
7098 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
7100 * expr.c (store_field): In the bitfield case, if the value comes from
7101 a function call and is of an aggregate type returned in registers, do
7102 not modify the field mode; extract the value in all cases if the mode
7103 is BLKmode and the size is not larger than a word.
7105 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
7108 * config/i386/cpuid.h: Fix undefined behavior.
7110 2017-01-04 Jeff Law <law@redhat.com>
7112 PR tree-optimization/79007
7113 PR tree-optimization/67955
7114 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
7115 conservative for pt.null when flag_non_call_exceptions is on.
7117 2017-01-09 Jakub Jelinek <jakub@redhat.com>
7119 PR translation/79019
7120 PR translation/79020
7121 * params.def (PARAM_INLINE_MIN_SPEEDUP,
7122 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
7123 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
7125 * config/avr/avr.opt (maccumulate-args): Likewise.
7126 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
7127 * common.opt (freport-bug): Likewise.
7128 * cif-code.def (CIF_FINAL_ERROR): Likewise.
7129 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
7130 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
7131 translatable string.
7132 * config/i386/i386.c (function_value_32): Likewise.
7133 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
7134 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
7136 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
7137 * common/config/msp430/msp430-common.c (msp430_handle_option):
7139 * symtab.c (symtab_node::verify_base): Likewise.
7140 * opts.c (set_debug_level): Likewise.
7141 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
7142 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
7143 missing whitespace to translatable strings.
7144 * config/avr/avr.md (bswapsi2): Fix typo in comment.
7145 * config/sh/superh.h: Likewise.
7146 * config/i386/xopintrin.h: Likewise.
7147 * config/i386/znver1.md: Likewise.
7148 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
7149 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
7150 * double-int.h (struct double_int): Likewise.
7151 * double-int.c (div_and_round_double): Likewise.
7152 * wide-int.cc: Likewise.
7153 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
7154 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
7155 * cfgcleanup.c (crossjumps_occured): Renamed to ...
7156 (crossjumps_occurred): ... this.
7157 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
7160 PR tree-optimization/78899
7161 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
7162 returning bool return struct loop *, NULL for failure and the new
7164 (versionable_outer_loop_p): Don't version outer loop if it has
7165 dont_vectorized bit set.
7166 (tree_if_conversion): When versioning outer loop, ensure
7167 tree_if_conversion is performed also on the inner loop of the
7168 non-vectorizable outer loop copy.
7169 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
7170 LOOP_VECTORIZED in inner loop of the scalar outer loop and
7171 prevent vectorization of it.
7172 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
7173 the outer loop vectorization of the non-scalar version is attempted
7174 before vectorization of the inner loop in scalar version. If
7175 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
7176 vectorization of its inner loop.
7177 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
7178 has 2 inner loops, rename also on edges from bb whose single pred
7179 is outer_loop->header. Fix typo in function comment.
7181 2017-01-09 Martin Sebor <msebor@redhat.com>
7184 * asan.c (asan_emit_stack_protection): Increase local buffer size
7185 to avoid snprintf truncation warning.
7187 2017-01-09 Andrew Pinski <apinski@cavium.com>
7189 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
7190 to reference thunderx2t99 for the tuning structure
7191 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
7193 (thunderx2t99_extra_costs): This.
7194 * config/aarch64/aarch64-tune.md: Regenerate.
7195 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
7196 (vulcan_addrcost_table): This.
7197 (vulcan_regmove_cost): Rename to ...
7198 (thunderx2t99_regmove_cost): This.
7199 (vulcan_vector_cost): Rename to ...
7200 (thunderx2t99_vector_cost): this.
7201 (vulcan_branch_cost): Rename to ...
7202 (thunderx2t99_branch_cost): This.
7203 (vulcan_tunings): Rename to ...
7204 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
7205 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
7207 2017-01-09 Martin Jambor <mjambor@suse.cz>
7211 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
7212 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
7213 (propagate_vr_accross_jump_function): Use the above function for all
7214 value range computations for pass-through jump functions and type
7215 converasion from explicit value range values.
7216 (ipcp_propagate_stage): Do not attempt to deduce types of formal
7217 parameters from TYPE_ARG_TYPES.
7218 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
7219 (ipa_write_node_info): Stream type of the actual argument.
7220 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
7222 2017-01-09 Martin Liska <mliska@suse.cz>
7225 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
7226 (lookup_compiler): Do not show error message with have_E.
7228 2017-01-09 Jakub Jelinek <jakub@redhat.com>
7230 PR tree-optimization/78938
7231 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
7232 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
7233 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
7234 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
7237 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7239 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
7242 2017-01-09 Richard Biener <rguenther@suse.de>
7244 PR tree-optimization/78997
7245 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
7246 name condition properly.
7248 2017-01-09 Richard Biener <rguenther@suse.de>
7251 * dwarf2out.c (is_cxx): New overload with context.
7252 (is_naming_typedef_decl): Use it.
7254 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7256 * invoke.texi (Option Summary): Correct spacing in option lists
7257 and add line breaks to fix over-long lines.
7259 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7263 * extend.texi (Common Variable Attributes): Add xref to GCC
7264 Internals manual to explain mode attribute keywords.
7266 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7269 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
7270 and Preprocessor Options.
7271 (Options for Linking): Document -pthread here....
7272 (RS/6000 and PowerPC Options): ...not here.
7273 (Solaris 2 Options): ...or here.
7274 * doc/cppopts.texi: Document -pthread.
7276 2017-01-08 Martin Sebor <msebor@redhat.com>
7279 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
7280 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
7281 New member functions.
7282 (format_directive): Used them.
7284 (pass_sprintf_length::handle_gimple_call): Same.
7285 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
7286 to avoid truncation for any argument.
7287 (extract_affine_mul): Same.
7288 * tree.c (get_file_function_name): Same.
7290 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7293 * predict.def (PRED_INDIR_CALL): Set to 86.
7295 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
7297 PR preprocessor/54124
7298 * doc/cppopts.texi: Reformat -d subtable to list the full name
7299 of the options. Add cross-reference to the docs for the general
7300 compiler -d options.
7301 * doc/invoke.texi (Developer Options): Add cross-reference to the
7302 preprocessor-specific -d option documentation.
7304 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
7306 PR preprocessor/13498
7307 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
7308 redudant material, and reflect new command-line options.
7309 (System Headers): Likewise.
7311 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
7313 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
7314 -isystem, and -idirafter. Copy-edit.
7315 * doc/cppopts.texi: Copy-edit. Remove contradiction about
7316 default for -ftrack-macro-expansion. Delete obsolete and
7317 badly-formatted implementation details about -fdebug-cpp output.
7318 * doc/cppwarnopts.texi: Copy-edit.
7320 2017-01-07 David Malcolm <dmalcolm@redhat.com>
7323 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
7324 that the transition from a max line width >= 1<<10 to narrower
7325 lines works correctly.
7327 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
7329 * doc/options.texi (PerFunction): New.
7330 * opt-functions.awk (switch_flags): Map both Optimization and
7331 PerFunction to CL_OPTIMIZATION.
7332 * opth-gen.awk: Test for PerFunction flag along with
7334 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
7335 it only when the latter is present. Skip those that don't in
7336 the hash function generator.
7337 * common.opt (fvar-tracking): Mark as PerFunction instead of
7339 (fvar-tracking-assignments): Likewise.
7340 (fvar-tracking-assignments-toggle): Likewise.
7341 (fvar-tracking-uninit): Likewise.
7343 2017-01-07 Jakub Jelinek <jakub@redhat.com>
7345 PR translation/79018
7346 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
7349 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
7352 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
7353 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
7354 TARGET_LONG_JUMP_TABLE_OFFSETS.
7355 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
7356 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
7357 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
7358 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
7359 * config/m68k/m68k.md (tablejump expander): Likewise.
7360 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
7361 TARGET_LONG_JUMP_TABLE_OFFSETS.
7362 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
7363 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
7365 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
7366 David Holsgrove <david.holsgrove@xilinx.com>
7368 * common/config/microblaze/microblaze-common.c
7369 (TARGET_EXCEPT_UNWIND_INFO): Remove.
7370 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
7372 * config/microblaze/microblaze.c (microblaze_must_save_register)
7373 (microblaze_expand_epilogue, microblaze_return_addr): Handle
7375 (microblaze_eh_return): New function.
7376 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
7377 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
7378 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
7379 * config/microblaze/microblaze.md (eh_return): New pattern.
7381 2017-01-06 Jakub Jelinek <jakub@redhat.com>
7383 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
7384 GCC_DIAGNOSTIC_STRINGIFY): Define.
7386 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
7388 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7390 * config/arm/arm.md (<mcrr>): New.
7392 * config/arm/arm.c (arm_arch5te): New.
7393 (arm_option_override): Set arm_arch5te.
7394 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
7396 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
7397 (arm_mcrr_qualifiers): ... this. New.
7398 (MRRC_QUALIFIERS): Define to...
7399 (arm_mrrc_qualifiers): ... this. New.
7400 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
7402 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
7403 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
7404 (MRRCI, mrrc, MRRC): New.
7405 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
7406 VUNSPEC_MRRC2): New.
7408 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7410 * config/arm/arm.md (<mcr>): New.
7412 * config/arm/arm.c (arm_coproc_builtin_available): Add
7413 support for mcr, mrc, mcr2 and mrc2.
7414 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
7415 (arm_mcr_qualifiers): ... this. New.
7416 (MRC_QUALIFIERS): Define to ...
7417 (arm_mrc_qualifiers): ... this. New.
7418 (MCR_QUALIFIERS): Define to ...
7419 (arm_mcr_qualifiers): ... this. New.
7420 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
7422 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
7423 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
7424 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
7427 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7429 * config/arm/arm.md (*ldc): New.
7433 * config/arm/arm.c (arm_coproc_builtin_available): Add
7434 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
7435 (arm_coproc_ldc_stc_legitimate_address): New.
7436 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
7437 'qualifier_const_pointer'.
7438 (LDC_QUALIFIERS): Define to...
7439 (arm_ldc_qualifiers): ... this. New.
7440 (STC_QUALIFIERS): Define to...
7441 (arm_stc_qualifiers): ... this. New.
7442 * config/arm/arm-protos.h
7443 (arm_coproc_ldc_stc_legitimate_address): New.
7444 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
7445 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
7446 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
7447 stc2, stcl, stc2l): New.
7448 * config/arm/constraints.md (Uz): New.
7449 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
7450 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
7451 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
7452 VUNSPEC_STC2L): New.
7454 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7456 * config/arm/arm.md (<cdp>): New.
7457 * config/arm/arm.c (neon_const_bounds): Rename this ...
7458 (arm_const_bounds): ... this.
7459 (arm_coproc_builtin_available): New.
7460 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
7461 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
7462 (CDP_QUALIFIERS): Define to...
7463 (arm_cdp_qualifiers): ... this. New.
7465 (arm_expand_builtin_args): Add case for 6 arguments.
7466 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
7467 (arm_const_bounds): ... this.
7468 (arm_coproc_builtin_available): New.
7469 * config/arm/arm_acle.h (__arm_cdp): New.
7471 * config/arm/arm_acle_builtins.def (cdp): New.
7473 * config/arm/iterators.md (CDPI,CDP,cdp): New.
7474 * config/arm/neon.md: Rename all 'neon_const_bounds' to
7476 * config/arm/types.md (coproc): New.
7477 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
7478 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
7479 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
7480 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
7482 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7484 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
7485 (UBINOP_QUALIFIERS): New.
7487 (acle_builtin_data): New. Change comment.
7488 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
7489 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
7490 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
7491 arm_acle_builtins.def.
7492 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
7493 (arm_init_acle_builtins): New.
7494 (CRC32_BUILTIN): Remove.
7495 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
7496 crc32cb, crc32ch and crc32cw.
7497 (arm_init_crc32_builtins): Remove.
7498 (arm_init_builtins): Use arm_init_acle_builtins rather
7499 than arm_init_crc32_builtins.
7500 (arm_expand_acle_builtin): New.
7501 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
7502 * config/arm/arm_acle_builtins.def: New.
7504 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7506 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
7507 (arm_builtin_datum): ... this.
7508 (arm_init_neon_builtin): Rename to ...
7509 (arm_init_builtin): ... this. Add a new parameters PREFIX
7510 and USE_SIG_IN_NAME.
7511 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
7512 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
7513 'arm_builtin_datum'.
7514 (arm_init_vfp_builtins): Likewise.
7515 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
7516 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
7517 (arm_expand_neon_args): Rename to ...
7518 (arm_expand_builtin_args): ... this. Rename builtin_arg
7519 enum values and differentiate between ARG_BUILTIN_MEMORY
7520 and ARG_BUILTIN_NEON_MEMORY.
7521 (arm_expand_neon_builtin_1): Rename to ...
7522 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
7523 values, arm_expand_builtin_args and add bool parameter NEON.
7524 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
7525 (arm_expand_vfp_builtin): Likewise.
7526 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
7528 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7531 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
7532 * predict.c (tree_estimate_probability_bb): Reverse direction of
7533 polymorphic call predictor.
7535 2017-01-06 David Malcolm <dmalcolm@redhat.com>
7537 * passes.c (execute_one_pass): Split out pass-skipping logic into...
7538 (determine_pass_name_match): ...this new function and...
7539 (should_skip_pass_p): ...this new function.
7541 2017-01-06 Nathan Sidwell <nathan@acm.org>
7543 * ipa-visibility.c (function_and_variable_visibility): Reformat
7544 comments and long lines. Remove extrneous if.
7545 * symtab.c (symtab_node::make_decl_local): Fix code format.
7546 (symtab_node::set_section_for_node): Fix comment typo.
7548 2017-01-06 Martin Liska <mliska@suse.cz>
7551 * lra-constraints.c: Rename invariant to lra_invariant.
7552 * predict.c (set_even_probabilities): Initialize e to NULL.
7554 2017-01-05 Martin Sebor <msebor@redhat.com>
7556 PR tree-optimization/78910
7557 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
7558 (format_integer): Correct off-by-one error in the handling
7559 of precision with negative numbers in signed conversions..
7561 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
7563 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
7565 2017-01-05 Jakub Jelinek <jakub@redhat.com>
7567 PR tree-optimization/71016
7568 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
7569 factor_out_conditional_conversion. Formatting fix.
7570 (factor_out_conditional_conversion): Add cond_stmt argument.
7571 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
7572 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
7575 2017-01-05 David Malcolm <dmalcolm@redhat.com>
7577 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
7578 read-rtl-function.o, and selftest-rtl.o.
7579 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
7580 (selftest::aarch64_test_loading_full_dump): New function.
7581 (selftest::aarch64_run_selftests): New function.
7582 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
7583 selftest::aarch64_run_selftests.
7584 * config/i386/i386.c
7585 (selftest::ix86_test_loading_dump_fragment_1): New function.
7586 (selftest::ix86_test_loading_call_insn): New function.
7587 (selftest::ix86_test_loading_full_dump): New function.
7588 (selftest::ix86_test_loading_unspec): New function.
7589 (selftest::ix86_run_selftests): Call the new functions.
7590 * emit-rtl.c (maybe_set_max_label_num): New function.
7591 * emit-rtl.h (maybe_set_max_label_num): New decl.
7592 * function.c (instantiate_decls): Guard call to
7593 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
7594 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
7596 * gensupport.c (gen_reader::gen_reader): Pass "false"
7597 for new "compact" param of rtx_reader.
7598 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
7599 rather than an empty string for NULL strings.
7600 * read-md.c: Potentially include config.h rather than bconfig.h.
7601 Wrap include of errors.h with #ifdef GENERATOR_FILE.
7602 (have_error): New global, copied from errors.c.
7603 (md_reader::read_name): Rename to...
7604 (md_reader::read_name_1): ...this, adding "out_loc" param,
7605 and converting "missing name or number" to returning false, rather
7607 (md_reader::read_name): Reimplement in terms of read_name_1.
7608 (md_reader::read_name_or_nil): New function.
7609 (md_reader::read_string): Handle "(nil)" by returning NULL.
7610 (md_reader::md_reader): Add new param "compact".
7611 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
7612 (md_reader::read_file): New method.
7613 * read-md.h (md_reader::md_reader): Add new param "compact".
7614 (md_reader::read_file): New method.
7615 (md_reader::is_compact): New accessor.
7616 (md_reader::read_name): Convert return type from void to file_location.
7617 (md_reader::read_name_or_nil): New decl.
7618 (md_reader::read_name_1): New decl.
7619 (md_reader::m_compact): New field.
7620 (noop_reader::noop_reader): Pass "false" for new "compact" param
7622 (rtx_reader::rtx_reader): Add new "compact" param.
7623 (rtx_reader::read_rtx_operand): Make virtual and convert return
7624 type from void to rtx.
7625 (rtx_reader::read_until): New decl.
7626 (rtx_reader::handle_any_trailing_information): New virtual function.
7627 (rtx_reader::postprocess): New virtual function.
7628 (rtx_reader::finalize_string): New virtual function.
7629 (rtx_reader::m_in_call_function_usage): New field.
7630 (rtx_reader::m_reuse_rtx_by_id): New field.
7631 * read-rtl-function.c: New file.
7632 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
7633 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
7634 (selftest::verify_three_block_rtl_cfg): New decl.
7635 * read-rtl-function.h: New file.
7636 * read-rtl.c: Potentially include config.h rather than bconfig.h.
7637 For host, include function.h, memmodel.h, and emit-rtl.h.
7638 (one_time_initialization): New function.
7639 (struct compact_insn_name): New struct.
7640 (compact_insn_names): New array.
7641 (find_code): Handle insn codes in compact dumps.
7642 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
7643 (bind_subst_iter_and_attr): Likewise.
7644 (add_condition_to_string): Likewise.
7645 (add_condition_to_rtx): Likewise.
7646 (apply_attribute_uses): Likewise.
7647 (add_current_iterators): Likewise.
7648 (apply_iterators): Likewise.
7649 (initialize_iterators): Guard usage of apply_subst_iterator with
7650 #ifdef GENERATOR_FILE.
7651 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
7652 (md_reader::read_mapping): Likewise.
7653 (add_define_attr_for_define_subst): Likewise.
7654 (add_define_subst_attr): Likewise.
7655 (read_subst_mapping): Likewise.
7656 (check_code_iterator): Likewise.
7657 (rtx_reader::read_rtx): Likewise. Move one-time initialization
7659 (one_time_initialization): New function.
7660 (rtx_reader::read_until): New method.
7661 (read_flags): New function.
7662 (parse_reg_note_name): New function.
7663 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
7664 Handle reuse_rtx ids.
7665 Wrap iterator lookup within #ifdef GENERATOR_FILE.
7666 Add parsing support for RTL dumps, mirroring the special-cases in
7667 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
7668 values, and calling handle_any_trailing_information.
7669 (rtx_reader::read_rtx_operand): Convert return type from void
7670 to rtx, returning return_rtx. Handle case 'e'. Call
7671 finalize_string on XSTR and XTMPL fields.
7672 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
7673 "(nil)" values were omitted. Call the postprocess vfunc on the
7675 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
7676 class ctor. Initialize m_in_call_function_usage. Call
7677 one_time_initialization.
7678 * rtl-tests.c (selftest::test_uncond_jump): Call
7679 set_new_first_and_last_insn.
7680 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
7681 * selftest-rtl.c: New file.
7682 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
7683 (selftest::get_insn_by_uid): New decl.
7684 * selftest-run-tests.c (selftest::run_tests): Call
7685 read_rtl_function_c_tests.
7686 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
7687 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
7690 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
7692 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
7693 operands in a special way. Assert that pos+len <= mode precision.
7695 2017-01-05 Jakub Jelinek <jakub@redhat.com>
7697 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
7698 3 argument Alias with unlimited for the negative form.
7699 (fno-vect-cost-model): Removed.
7701 2017-01-05 Martin Liska <mliska@suse.cz>
7703 * hsa-gen.c (gen_hsa_divmod): New function.
7704 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
7706 2017-01-05 Martin Liska <mliska@suse.cz>
7709 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
7712 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7714 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
7715 small constant length operands.
7717 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7719 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
7720 between loop iterations.
7722 2017-01-05 Martin Liska <mliska@suse.cz>
7725 * gimplify.c (gimplify_decl_expr): Compare to
7726 asan_poisoned_variables instread of checking flags.
7727 (gimplify_target_expr): Likewise.
7728 (gimplify_expr): Likewise.
7729 (gimplify_function_tree): Conditionally initialize
7730 asan_poisoned_variables.
7732 2017-01-04 Jeff Law <law@redhat.com>
7734 PR tree-optimizatin/78812
7735 * rtl.h (contains_mem_rtx_p): Prototype.
7736 * ifcvt.c (containts_mem_rtx_p): Move from here to...
7737 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
7738 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
7739 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
7740 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
7742 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7744 * input.c (assert_char_at_range): Default-initialize actual_range.
7746 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7748 * df-scan.c (df_ref_create_structure): Make regno unsigned,
7749 to match the caller.
7751 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7753 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
7754 insns after final jump in test to emit dummy move.
7756 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7758 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
7759 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
7761 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7763 * multiple_target.c (create_dispatcher_calls): Init e_next.
7764 * tree-ssa-loop-split.c (split_loop): Init border.
7765 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
7768 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
7773 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
7774 (altivec_register_operand): Do not return true if the operand
7775 contains a SUBREG mixing SImode and SFmode.
7776 (vsx_register_operand): Likewise.
7777 (vsx_reg_sfsubreg_ok): New predicate.
7778 (vfloat_operand): Do not return true if the operand contains a
7779 SUBREG mixing SImode and SFmode.
7780 (vint_operand): Likewise.
7781 (vlogical_operand): Likewise.
7782 (gpc_reg_operand): Likewise.
7783 (int_reg_operand): Likewise.
7784 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
7785 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
7786 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
7788 (rs6000_emit_move_si_sf_subreg): New helper function.
7789 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
7790 fixup SUBREGs involving SImode and SFmode.
7791 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
7792 numbers for the new peephole2 optimization.
7793 (peephole2 for SFmode unions): New peephole2 to optimize cases in
7794 the GLIBC math library that do AND/IOR/XOR operations on single
7795 precision floating point.
7796 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
7797 target macros to say whether we need to avoid SUBREGs mixing
7799 (TARGET_ALLOW_SF_SUBREG): Likewise.
7800 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
7801 (UNSPEC_SI_FROM_SF): Likewise.
7802 (iorxor): Change spacing.
7803 (and_ior_xor): New iterator for AND, IOR, and XOR.
7804 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
7805 (movdi_from_sf_zero_ext): Likewise.
7806 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
7807 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
7808 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
7809 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
7810 (fms<mode>4): Likewise.
7811 (fnma<mode>4): Likewise.
7812 (fnms<mode>4): Likewise.
7813 (nfma<mode>4): Likewise.
7814 (nfms<mode>4): Likewise.
7816 2017-01-04 Marek Polacek <polacek@redhat.com>
7819 * doc/invoke.texi: Document -Wpointer-compare.
7821 2017-01-04 Jakub Jelinek <jakub@redhat.com>
7823 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
7826 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
7827 descriptions for -gdwarf-5 and emit them as uleb128 instead of
7830 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
7833 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
7834 documentation of the powerpc_popcntb_ok attribute.
7835 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
7836 code to issue warning messages if a requested CPU configuration is
7837 not supported by the binary (assembler and loader) toolchain.
7838 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
7839 made to define a built-in function that has been disabled.
7840 (paired_init_builtins): Add assertion to prevent ICE if attempt is
7841 made to define a built-in function that has been disabled.
7842 (altivec_init_builtins): Add comment explaining why definition
7843 of the DST built-in functions is not preceded by an assertion
7844 check. Add assertions to prevent ICE if attempts are made to
7845 define an altivec predicate or an abs* built-in function that has
7847 (htm_init_builtins): Add comment explaining why definition of the
7848 htm built-in functions is not preceded by an assertion check.
7850 2017-01-04 Jeff Law <law@redhat.com>
7852 PR tree-optimizatin/67955
7853 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
7854 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
7855 the points-to solution does not include pt_null. Use DECL_PT_UID
7858 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
7860 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
7861 Use gen_int_mode instead of gen_lopwart for const_int operands.
7863 2017-01-04 Jakub Jelinek <jakub@redhat.com>
7865 PR tree-optimization/71563
7866 * match.pd: Simplify X << Y into X if Y is known to be 0 or
7867 out of range value - has low bits known to be zero.
7869 2017-01-04 Alan Modra <amodra@gmail.com>
7871 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
7872 * configure: Regenerate.
7873 * config.in: Regenerate.
7875 2017-01-04 Jakub Jelinek <jakub@redhat.com>
7878 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
7879 a substring of the message, but strcmp with the whole message. Ifdef
7880 ENABLE_NLS, translate the message first using dgettext.
7882 2017-01-03 Jeff Law <law@redhat.com>
7884 PR tree-optimizatin/78856
7885 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
7886 (mark_threaded_blocks): Remove code to truncate thread paths that
7887 cross multiple loop headers. Instead invalidate the cached loop
7888 iteration information and handle case of a thread path walking
7889 into an irreducible region.
7891 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7894 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
7895 assertions. Add support for doing the signbit if the IEEE 128-bit
7896 floating point value is in a GPR.
7897 * config/rs6000/rs6000.md (Fsignbit): Delete.
7898 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
7899 Update the length attribute if the value is in a GPR.
7900 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
7901 the sign or zero extension instruction, since the value is always 0/1.
7902 (signbit<mode>2_dm2): Delete using <Fsignbit>.
7905 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
7906 extracting SImode to a GPR register so that we can generate a
7907 store, limit the vector to be in a traditional Altivec register
7908 for the vextuwrx instruction.
7910 2017-01-03 Ian Lance Taylor <iant@google.com>
7912 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
7914 2017-01-03 Martin Sebor <msebor@redhat.com>
7916 PR tree-optimization/78696
7917 * gimple-ssa-sprintf.c (format_floating): Correct handling of
7918 precision. Use MPFR for %f for greater fidelity. Correct handling
7920 (pass_sprintf_length::compute_format_length): Set width and precision
7921 specified by asrerisk to void_node for vararg functions.
7922 (try_substitute_return_value): Adjust dump output.
7924 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
7926 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
7928 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
7930 * doc/invoke.texi (SPARC options): Document -mlra as the default.
7931 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
7932 -mlra/-mno-lra was passed to the compiler.
7934 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
7936 PR rtl-optimization/65618
7937 * emit-rtl.c (try_split): Move initialization of "before" and
7938 "after" to just before the call to emit_insn_after_setloc.
7940 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
7942 * doc/md.texi (Standard Names): Remove reference to Java frontend.
7944 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
7946 * dwarf2out.c (gen_enumeration_type_die): When
7947 -gno-strict-dwarf, add a DW_AT_encoding attribute.
7949 2017-01-03 Jakub Jelinek <jakub@redhat.com>
7951 PR tree-optimization/78965
7952 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
7953 Change first argument from const call_info & to call_info &. For %n
7954 set info.nowrite to false.
7957 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
7958 possibly throwing calls.
7960 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
7961 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
7962 and fns handling, rather than in a separate case SSA_NAME.
7964 2017-01-02 Jeff Law <law@redhat.com>
7966 * config/darwin-driver.c (darwin_driver_init): Const-correctness
7967 fixes for first_period and second_period variables.
7969 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
7972 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
7973 (*insvqi_1): New insn pattern.
7974 (*insvqi_1_mem_rex64): Ditto.
7976 (*insvqi_3): Rename from *insvqi.
7978 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
7980 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
7982 * doc/cfg.texi (Edges): Remove reference to Java.
7983 (Maintaining the CFG): Ditto.
7985 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7988 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
7989 transparent aliases.
7991 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7994 * predict.def (PRED_CALL): Update hitrate.
7995 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
7996 * predict.c (tree_estimate_probability_bb): Split CALL predictor
7997 into direct/indirect/polymorphic variants.
7999 2017-01-01 Jakub Jelinek <jakub@redhat.com>
8001 Update copyright years.
8003 * gcc.c (process_command): Update copyright notice dates.
8004 * gcov-dump.c (print_version): Ditto.
8005 * gcov.c (print_version): Ditto.
8006 * gcov-tool.c (print_version): Ditto.
8007 * gengtype.c (create_file): Ditto.
8008 * doc/cpp.texi: Bump @copying's copyright year.
8009 * doc/cppinternals.texi: Ditto.
8010 * doc/gcc.texi: Ditto.
8011 * doc/gccint.texi: Ditto.
8012 * doc/gcov.texi: Ditto.
8013 * doc/install.texi: Ditto.
8014 * doc/invoke.texi: Ditto.
8016 Copyright (C) 2017 Free Software Foundation, Inc.
8018 Copying and distribution of this file, with or without modification,
8019 are permitted in any medium without royalty provided the copyright
8020 notice and this notice are preserved.