1 2017-04-13 Martin Liska <mliska@suse.cz>
4 * gcov-io.c (gcov_write_string): Copy to buffer just when
5 allocated size is greater than zero.
7 2017-04-13 Jakub Jelinek <jakub@redhat.com>
10 * dwarf2out.c (decls_for_scope): Ignore declarations of
11 current_function_decl in BLOCK_NONLOCALIZED_VARS.
13 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
16 * ipa-visibility.c (non_local_p): Fix typos.
17 (localize_node): When localizing symbol in same comdat group,
18 dissolve the group only when we know external symbols are going
20 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
22 2017-04-12 Jakub Jelinek <jakub@redhat.com>
24 PR tree-optimization/79390
25 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26 order does not result in usable sequence, retry with reversed operand
32 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
33 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34 op0 instead of fold_convert_loc (loc, type, arg0).
36 2017-04-12 Jeff Law <law@redhat.com>
38 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
39 has a delay slot in the generated code.
41 * config/cris/cris.md (cris_preferred_reload_class): Return
42 GENNONACR_REGS rather than GENERAL_REGS.
44 2017-04-12 Jakub Jelinek <jakub@redhat.com>
47 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
48 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
49 signedness of the result type.
51 2017-04-12 Richard Biener <rguenther@suse.de>
52 Jeff Law <law@redhat.com>
54 PR tree-optimization/80359
55 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
56 trim stores to TARGET_MEM_REFs.
58 2017-04-12 Richard Biener <rguenther@suse.de>
60 PR tree-optimization/79390
61 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
62 threading case even more.
64 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
67 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
68 for quad_address_p for TImode, instead of just not indexed_address.
70 2017-04-12 Richard Biener <rguenther@suse.de>
71 Bernd Edlinger <bernd.edlinger@hotmail.de>
74 * alias.c (component_uses_parent_alias_set_from): Handle
75 TYPE_TYPELESS_STORAGE.
76 (get_alias_set): Likewise.
77 * tree-core.h (tree_type_common): Add typeless_storage flag.
78 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
79 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
80 for types containing members with TYPE_TYPELESS_STORAGE.
81 (place_field): Likewise.
82 (layout_type): Likewise for ARRAY_TYPE.
83 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
84 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
85 TYPE_TYPELESS_STORAGE.
86 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
88 2017-04-12 Jakub Jelinek <jakub@redhat.com>
91 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
92 first argument to type.
94 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
98 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
99 CONST0_RTX (mode) rather than const0_rtx where appropriate.
100 (rs6000_expand_binop_builtin): Likewise.
101 (rs6000_expand_ternop_builtin): Likewise; also add missing
102 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
104 * doc/extend.texi: Document that vec_xxpermdi's third argument
107 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
109 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
110 Use shift_const cost parameter when calculating gain of STV shifts.
112 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
114 PR rtl-optimization/70478
115 * lra-constraints.c (process_alt_operands): Check memory for
116 disfavoring memory insn operand.
118 2017-04-11 Jakub Jelinek <jakub@redhat.com>
121 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
122 left shift in unsigned HOST_WIDE_INT type.
124 PR rtl-optimization/80385
125 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
126 (not (neg X)) into (plus X -1) for complex or non-integral modes.
129 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
130 if they have any depend clauses.
132 2017-04-11 Martin Liska <mliska@suse.cz>
135 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
136 * ipa-split.c (split_function): Create a local comdat symbol
137 if caller is in a comdat group.
139 2017-04-11 Martin Liska <mliska@suse.cz>
142 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
145 2017-04-11 Martin Sebor <msebor@redhat.com>
148 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
149 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
151 (directive::set_width, directive::set_precision, format_character):
153 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
156 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
159 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
160 conflict, set target->arch_name instead of target->cpu_name.
162 2017-04-11 Richard Biener <rguenther@suse.de>
164 PR tree-optimization/80374
165 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
166 build_zero_cst, remove fold_convertible_p check again.
168 2017-04-11 Martin Liska <mliska@suse.cz>
171 * ubsan.c (instrument_object_size): Do not instrument register
174 2017-04-11 Jakub Jelinek <jakub@redhat.com>
177 * config/i386/i386-builtin-types.def
178 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
179 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
180 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
181 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
182 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
183 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
184 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
185 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
186 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
187 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
188 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
189 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
190 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
191 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
192 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
193 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
194 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
195 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
196 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
197 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
198 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
199 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
200 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
201 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
202 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
203 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
204 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
205 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
206 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
207 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
208 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
209 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
210 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
211 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
212 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
213 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
214 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
215 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
216 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
217 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
218 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
219 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
220 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
221 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
222 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
223 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
225 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
226 flag to second_arg_count, handle 4 argument function type _COUNT
227 aliases, handle second_arg_count on second argument rather than last.
229 2017-04-10 Jeff Law <law@redhat.com>
231 PR tree-optimization/80374
232 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
233 record anything if we can not convert integer_zero_node to the
236 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
239 * config/rs6000/rs6000.c (rs6000_option_override_internal):
240 Enhance special handling given to the TARGET_P9_MINMAX option in
241 relation to certain other options.
243 2017-04-10 Bin Cheng <bin.cheng@arm.com>
245 PR tree-optimization/80153
246 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
247 remove POINTER_PLUS_EXPR's base part directly, rather than through
250 2017-04-10 Richard Biener <rguenther@suse.de>
251 Bin Cheng <bin.cheng@arm.com>
253 PR tree-optimization/80153
254 * tree-affine.c (aff_combination_to_tree): Get base pointer from
255 the first element of pointer type aff_tree. Build result expr in
257 (add_elt_to_tree): Convert to type unconditionally. Remove other
259 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
260 (rewrite_use_nonlinear_expr): Check invariant using iv information.
262 2017-04-10 Richard Biener <rguenther@suse.de>
264 * tree-ssa-structalias.c (find_func_aliases): Properly handle
267 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
269 PR rtl-optimization/70478
270 * lra-constraints.c (curr_small_class_check): New.
271 (update_and_check_small_class_inputs): New.
272 (process_alt_operands): Update curr_small_class_check. Disfavor
273 alternative insn memory operands. Check available regs for small
276 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
279 * config/mips/mips.opt (-mvirt): Update description.
280 * doc/invoke.texi (-mvirt): Likewise.
282 2017-04-10 Richard Biener <rguenther@suse.de>
285 * fold-const.c (fold_binary_loc): Look at unstripped ops when
286 looking for NEGATE_EXPR in -A / -B to A / B folding.
288 2017-04-10 Martin Liska <mliska@suse.cz>
290 PR gcov-profile/80224
291 * gcov.c (print_usage): Fix usage string.
292 (get_gcov_intermediate_filename): Remove.
293 (output_gcov_file): Use both for normal and intermediate format.
294 (generate_results): Do not initialize special file for
297 2017-04-10 Richard Biener <rguenther@suse.de>
299 PR tree-optimization/80304
300 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
303 2017-04-10 Nathan Sidwell <nathan@acm.org>
306 * config/rs6000/rs6000.c (rs6000_vector_type): New.
307 (rs6000_init_builtins): Use it.
309 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
311 * config/arm/arm.md (<mrc>): Add mode to SET source.
314 2017-04-10 Richard Biener <rguenther@suse.de>
317 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
319 2017-04-10 Jakub Jelinek <jakub@redhat.com>
322 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
323 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
324 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
325 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
326 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
327 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
328 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
329 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
330 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
331 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
332 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
333 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
334 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
335 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
336 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
337 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
338 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
339 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
340 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
341 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
342 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
343 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
344 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
346 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
348 PR rtl-optimization/70478
349 * lra-constraints.c: Reverse the last patch.
351 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
353 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
354 Add comment for WCHAR_T.
356 2017-04-08 Martin Liska <mliska@suse.cz>
359 2017-04-07 Martin Liska <mliska@suse.cz>
362 * ipa-split.c (split_function): Add function part to a same comdat
365 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
368 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
370 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
372 * rs6000/rs6000.c (vec_load_pendulum): Rename...
373 (vec_pairing): ...to this.
374 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
375 (rs6000_sched_init): Adjust for name change.
376 (struct rs6000_sched_context): Likewise.
377 (rs6000_init_sched_context): Likewise.
378 (rs6000_set_sched_context): Likewise.
380 2017-04-07 Jakub Jelinek <jakub@redhat.com>
386 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
388 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
389 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
390 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
392 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
394 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
396 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
398 PR rtl-optimization/70703
399 * ira-color.c (update_conflict_hard_regno_costs): Use
400 int64_t instead of HOST_WIDE_INT.
402 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
404 PR rtl-optimization/70478
405 * lra-constraints.c (process_alt_operands): Disfavor alternative
406 insn memory operands.
408 2017-04-07 Jeff Law <law@redhat.com>
410 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
411 CALL and NOTE_INSN_CALL_ARG_LOCATION.
413 2017-04-07 Martin Liska <mliska@suse.cz>
416 * config/aarch64/aarch64.c (aarch64_process_target_attr):
417 Show error message instead of an ICE.
419 2017-04-07 Martin Liska <mliska@suse.cz>
422 * ipa-split.c (split_function): Add function part to a same comdat
425 2017-04-07 Richard Biener <rguenther@suse.de>
428 * tree.c (get_unwidened): Also handle ! for_type case for
430 * convert.c (do_narrow): Split out from ...
431 (convert_to_integer_1): ... here. Do not pass final truncation
432 type to get_unwidened for TRUNC_DIV_EXPR.
434 2017-04-07 Richard Biener <rguenther@suse.de>
436 * tree-affine.c (wide_int_ext_for_comb): Take type rather
438 (aff_combination_const): Adjust.
439 (aff_combination_scale): Likewise.
440 (aff_combination_add_elt): Likewise.
441 (aff_combination_add_cst): Likewise.
442 (aff_combination_convert): Likewise.
443 (add_elt_to_tree): Likewise. Remove unused argument.
444 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
446 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
448 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
450 * config/arm/arm.c (arm_default_short_enums): Use
451 ARM_DEFAULT_SHORT_ENUMS.
452 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
454 2017-04-06 Jakub Jelinek <jakub@redhat.com>
457 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
458 members with redundant out-of-class redeclaration.
460 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
463 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
464 * config/i386/i386.md (*zero_extendsidi2):
465 Add (?*x,*x) and (?*v,*v) alternatives.
467 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
470 * config/i386/i386.c (ix86_expand_builtin)
471 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
472 mode from insn data. Convert operands to insn operand mode.
473 Copy operands that don't satisfy insn predicate to a register.
475 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
477 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
480 2017-04-06 Richard Biener <rguenther@suse.de>
482 PR tree-optimization/80334
483 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
484 preserve alignment of accesses.
486 2017-04-06 Richard Biener <rguenther@suse.de>
488 PR tree-optimization/80262
489 * tree-sra.c (build_ref_for_offset): Preserve address-space
491 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
492 Drop useless address-space information on MEM_REF offsets.
494 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
496 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
498 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
500 PR rtl-optimization/70703
501 * ira-color.c (update_conflict_hard_regno_costs): Use
502 HOST_WIDE_INT instead of long.
504 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
507 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
508 not defined for x86_64 target. Add -mmmx target option when __SSE2__
510 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
511 for x86_64 target. Handle -m3dnowa option.
513 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
515 PR rtl-optimization/70703
516 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
517 (update_conflict_hard_regno_costs): Use long instead of unsigned
518 arithmetic for cost calculation.
520 2017-04-05 Jakub Jelinek <jakub@redhat.com>
521 Bernd Edlinger <bernd.edlinger@hotmail.de>
524 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
527 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
530 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
531 ptr_mode with Pmode throughout.
532 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
533 into probe_stack_range and use DImode.
535 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
538 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
539 call_eh_return is true.
541 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
543 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
544 Initialize last_match_fntype_index.
546 2017-04-05 Jakub Jelinek <jakub@redhat.com>
549 * tree-nvr.c: Include internal-fn.h.
550 (pass_return_slot::execute): Ignore internal calls without
553 2017-04-04 Jakub Jelinek <jakub@redhat.com>
554 Richard Biener <rguenther@suse.de>
557 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
558 captures used multiple times, except for the last use.
559 * generic-match-head.c: Include gimplify.h.
561 2017-04-04 Jakub Jelinek <jakub@redhat.com>
563 PR tree-optimization/79390
564 * target.h (struct noce_if_info): Declare.
565 * targhooks.h (default_noce_conversion_profitable_p): Declare.
566 * target.def (noce_conversion_profitable_p): New target hook.
567 * ifcvt.h (struct noce_if_info): New type, moved from ...
568 * ifcvt.c (struct noce_if_info): ... here.
569 (noce_conversion_profitable_p): Renamed to ...
570 (default_noce_conversion_profitable_p): ... this. No longer
572 (noce_try_store_flag_constants, noce_try_addcc,
573 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
574 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
575 instead of noce_conversion_profitable_p.
576 * config/i386/i386.c: Include ifcvt.h.
577 (ix86_option_override_internal): Don't override
578 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
579 (ix86_noce_conversion_profitable_p): New function.
580 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
581 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
582 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
583 * doc/tm.texi: Regenerated.
585 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
587 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
590 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
593 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
594 instructions for small multiply cores.
596 2017-04-04 Jeff Law <law@redhat.com>
598 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
600 (mips_expand_vec_perm_const): Initialize elements in orig_perm
601 that are not set by the loop over the elements.
603 2017-04-04 Jakub Jelinek <jakub@redhat.com>
606 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
607 int mode, convert_modes it to mode as unsigned, otherwise use
608 lowpart_subreg to mode rather than SImode.
609 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
610 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
611 Use DImode instead of SImode for the shift count operand.
612 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
615 2017-04-04 Richard Biener <rguenther@suse.de>
618 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
619 arithmetic done for the negate or the plus. Simplify.
620 (A - (-B) -> A + B): Likewise.
621 * fold-const.c (split_tree): Make sure to not negate pointers.
623 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
625 PR rtl-optimization/60818
626 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
627 a compare of comparisons with the thing compared if this results
628 in a different machine mode.
630 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
632 * alias.c (base_alias_check): Fix typo in comment.
633 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
634 * cgraphunit.c (symbol_table::compile): Likewise.
635 * collect2.c (maybe_run_lto_and_relink): Likewise.
636 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
637 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
638 * config/avr/avr.c (avr_map_op_t): Likewise.
639 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
640 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
641 * config/epiphany/epiphany.md (movcc): Likewise.
642 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
643 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
645 * config/mips/mips.c (mips_save_restore_reg): Likewise.
646 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
647 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
648 * config/sh/sh.c (sh_rtx_costs): Likewise.
649 * fold-const.c (fold_truth_andor): Likewise.
650 * genautomata.c (collapse_flag): Likewise.
651 * gengtype.h (struct type::u::s): Likewise.
652 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
653 * input.c (FORMAT_AMOUNT): Likewise.
654 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
655 (known_aggs_to_agg_replacement_list): Likewise.
656 * ipa-inline-analysis.c: Likewise.
657 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
658 * ipa-polymorphic-call.c
659 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
660 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
661 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
663 * modulo-sched.c (apply_reg_moves): Likewise.
664 * omp-expand.c (build_omp_regions_1): Likewise.
665 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
666 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
667 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
668 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
669 * value-prof.c: Likewise.
670 * var-tracking.c (val_reset): Likewise.
672 2017-04-03 Richard Biener <rguenther@suse.de>
674 PR tree-optimization/80275
675 * fold-const.c (split_address_to_core_and_offset): Handle
678 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
680 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
681 descriptors is at least equal to that of functions.
683 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
685 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
687 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
690 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
691 (mov<IMOD4:mode>): New expander.
692 (*mov<IMOD4:mode>_internal): New insn and split pattern.
694 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
696 PR rtl-optimization/79405
697 * fwprop.c (propagations_left): New variable.
698 (forward_propagate_into): Decrement it.
699 (fwprop_init): Initialize it.
700 (fw_prop): If the variable has reached zero, stop propagating.
701 (fwprop_addr): Ditto.
703 2017-03-31 Jakub Jelinek <jakub@redhat.com>
706 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
707 a FUNCTION_DECL, pass it as decl instead of origin to
710 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
712 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
715 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
718 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
719 TARGET_VSX_SMALL_INTEGER.
721 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
723 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
724 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
726 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
728 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
729 extraction from odd-numbered MSA register.
731 2017-03-31 Jakub Jelinek <jakub@redhat.com>
734 * expmed.c (store_bit_field_1): Don't attempt to create
735 a word subreg out of hard registers wider than word if they
736 have HARD_REGNO_NREGS of 1 for their mode.
739 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
740 conversions to integer types wider than word and pointer.
743 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
744 (rtx_equal_for_cselib_p): Pass 0 to it.
745 * cselib.c (cselib_hasher::equal): Likewise.
746 (rtx_equal_for_cselib_1): Add depth argument. If depth
747 is 128, don't look up VALUE locs and punt. Increment
748 depth in recursive calls when walking VALUE locs.
750 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
752 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
753 (make_gcov_file_name): Use the canonical path name for generating
755 (read_line): Fix handling of files with ascii null bytes.
757 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
759 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
760 to initialise a vector register instead
761 of using a const_int.
763 2017-03-30 Jakub Jelinek <jakub@redhat.com>
766 * gimplify.c (omp_default_clause): Use %qs instead of %s in
769 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
772 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
773 (dfp_diex_<mode>): Update mode of operand 1.
774 * doc/extend.texi (dxex, dxexq): Document change to return type.
775 (diex, diexq): Document change to argument type.
777 2017-03-30 Martin Jambor <mjambor@suse.cz>
780 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
781 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
782 it reflects the signature changes performed at the callee side.
783 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
784 to cgraph_build_function_type_skip_args.
785 (build_function_decl_skip_args): Adjust call to the above function.
787 2017-03-30 Jakub Jelinek <jakub@redhat.com>
791 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
792 register as dest whenever it is a MEM not rtx_equal_p to the
793 corresponding dup operand, and when forcing into reg move the
794 reg into the memory afterwards.
795 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
796 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
797 for the force_reg mode.
798 (avx512vl_vextractf128<mode>): Use register as dest either
799 always when a MEM, or when it is a MEM not rtx_equal_p to the
800 corresponding dup operand, or even not when it is a CONST_VECTOR
801 depending on the mode and lo vs. hi.
802 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
804 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
805 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
806 Likewise. Require that operands[2] is even.
807 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
808 Remove extraneous parens. Require that operands[2] is a multiple
810 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
811 operands[0] is a MEM if <mask_applied>, the predicates/constraints
812 disallow memory then.
814 2017-03-30 Richard Biener <rguenther@suse.de>
816 PR tree-optimization/77498
817 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
818 to non-constants over backedges.
820 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
822 PR rtl-optimization/80233
823 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
824 as last_combined_insn. Do not test for BARRIER_P separately.
826 2017-03-29 Andreas Schwab <schwab@suse.de>
829 * calls.c (prepare_call_address): Convert funexp to Pmode before
832 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
834 PR tree-optimization/80158
835 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
836 Handle possible future case of more than one alternate
838 (replace_rhs_if_not_dup): Likewise.
839 (replace_one_candidate): Likewise.
841 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
843 PR rtl-optimization/80193
844 * ira.c (ira): Do not check allocation for LRA.
846 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
848 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
849 (nvptx_output_simt_exit): Declare.
850 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
851 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
852 (init_softstack_frame): Move initialization of crtl->is_leaf to...
853 (nvptx_declare_function_name): ...here. Emit declaration of local
854 memory space buffer for omp_simt_enter insn.
855 (nvptx_output_unisimt_switch): New.
856 (nvptx_output_softstack_switch): New.
857 (nvptx_output_simt_enter): New.
858 (nvptx_output_simt_exit): New.
859 * config/nvptx/nvptx.h (struct machine_function): New fields
860 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
861 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
862 (UNSPECV_SIMT_EXIT): Ditto.
863 (omp_simt_enter_insn): New insn.
864 (omp_simt_enter): New expansion.
865 (omp_simt_exit): New insn.
866 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
868 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
869 (expand_GOMP_SIMT_ENTER_ALLOC): New.
870 (expand_GOMP_SIMT_EXIT): New.
871 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
872 (GOMP_SIMT_ENTER_ALLOC): Ditto.
873 (GOMP_SIMT_EXIT): Ditto.
874 * target-insns.def (omp_simt_enter): New insn.
875 (omp_simt_exit): Ditto.
876 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
878 (lower_rec_simd_input_clauses): Implement SIMT privatization.
879 (lower_rec_input_clauses): Likewise.
880 (lower_lastprivate_clauses): Handle SIMT privatization.
882 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
883 (ompdevlow_adjust_simt_enter): New.
884 (find_simtpriv_var_op): New.
885 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
886 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
888 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
889 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
890 (copy_decl_for_dup_finish): Ditto.
892 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
894 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
897 * config/i386/i386.c (ix86_option_override_internal): Always
898 allow -mpreferred-stack-boundary=3 for 64-bit targets.
900 2017-03-28 Bin Cheng <bin.cheng@arm.com>
902 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
904 2017-03-28 Bin Cheng <bin.cheng@arm.com>
906 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
907 mark new edge's irreducible flag accordign to it.
908 (vect_do_peeling): Check loop preheader edge's irreducible flag
909 and pass it to function slpeel_add_loop_guard.
911 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
913 PR tree-optimization/80218
914 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
915 Update block frequencies and counts.
917 2017-03-28 Richard Biener <rguenther@suse.de>
919 PR tree-optimization/78644
920 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
921 of a simplification result we may not use it at all.
923 2017-03-28 Richard Biener <rguenther@suse.de>
926 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
927 without arguments, generate default definition of a SSA name.
929 2017-03-28 Richard Biener <rguenther@suse.de>
932 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
933 TYPE_REF_CAN_ALIAS_ALL references.
934 * fold-const.c (fold_indirect_ref_1): Likewise.
936 2017-03-28 Martin Liska <mliska@suse.cz>
939 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
940 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
942 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
943 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
945 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
946 (EXTRA_SPECS): Define.
947 (SUBTARGET_EXTRA_SPECS): Likewise.
948 (SUBTARGET_CPP_SPEC): Likewise.
949 * config/arc/elf.h (EXTRA_SPECS): Renamed to
950 SUBTARGET_EXTRA_SPECS.
951 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
953 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
955 * config/arc/simdext.md (vst64_insn): Update pattern.
956 (vld32wh_insn): Likewise.
957 (vld32wl_insn): Likewise.
958 (vld64_insn): Likewise.
959 (vld32_insn): Likewise.
961 2017-03-28 Marek Polacek <polacek@redhat.com>
964 * fold-const.c (fold_comparison): Use protected_set_expr_location
965 instead of SET_EXPR_LOCATION.
967 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
969 * tree.c (add_expr): Avoid name lookup warning.
971 2017-03-27 Jeff Law <law@redhat.com>
973 PR tree-optimization/80216
974 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
975 function name. Limit recursion depth.
976 (record_temporary_equivalences): Corresponding changes.
978 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
980 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
983 2017-03-27 Jakub Jelinek <jakub@redhat.com>
986 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
988 * cfgcleanup.c (reg_note_cfa_p): New array.
989 (insns_have_identical_cfa_notes): New function.
990 (old_insns_match_p): Don't cross-jump in between /f
991 and non-/f instructions. If both i1 and i2 are frame related,
992 verify all CFA notes, their order and content.
994 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
997 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
998 HImode and SImode with zero extend to DImode to one insn.
999 (bswap<mode>2_extenddi): Likewise.
1000 (bswapsi2_extenddi): Likewise.
1001 (bswaphi2_extendsi): Likewise.
1002 (bswaphi2): Combine bswap HImode and SImode into one insn.
1003 Separate memory insns from swapping register.
1004 (bswapsi2): Likewise.
1005 (bswap<mode>2): Likewise.
1006 (bswaphi2_internal): Delete, no longer used.
1007 (bswapsi2_internal): Likewise.
1008 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
1009 store, and gpr<-gpr swap insns.
1010 (bswap<mode>2_store): Likewise.
1011 (bswaphi2_reg): Register only splitter, combine with the splitter.
1012 (bswaphi2 splitter): Likewise.
1013 (bswapsi2_reg): Likewise.
1014 (bswapsi2 splitter): Likewise.
1015 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
1016 the insns into load, store, and register/register insns.
1017 (bswapdi2_ldbrx): Likewise.
1018 (bswapdi2_load): Likewise.
1019 (bswapdi2_store): Likewise.
1020 (bswapdi2_reg): Likewise.
1022 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
1024 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
1025 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
1027 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1030 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
1032 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1033 special handling for target option conflicts between dform
1034 options (-mpower9-dform, -mpower9-dform-vector,
1035 -mpower9-dform-scalar) and -mno-direct-move.
1037 2017-03-27 Richard Biener <rguenther@suse.de>
1039 PR tree-optimization/80181
1040 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
1042 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1044 * config/arc/predicates.md (move_double_src_operand): Replace the
1045 call to move_double_src_operand with a call to address_operand.
1047 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1049 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
1050 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
1051 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
1053 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1055 * config/arc/predicates.md (long_immediate_loadstore_operand):
1056 Consider scaled addresses cases.
1058 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1060 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
1061 restored when in interrupt.
1062 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
1063 doesn't have delay slot.
1065 2017-03-27 Richard Biener <rguenther@suse.de>
1068 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
1069 inlined thunk clones.
1071 2017-03-27 Jakub Jelinek <jakub@redhat.com>
1074 * asan.c (instrument_derefs): Copy over last operand from
1075 original COMPONENT_REF to the new COMPONENT_REF with
1076 DECL_BIT_FIELD_REPRESENTATIVE.
1077 * ubsan.c (instrument_object_size): Likewise.
1079 2017-03-27 Richard Biener <rguenther@suse.de>
1081 PR tree-optimization/80170
1082 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
1083 sure DR/SCEV didnt fold in constants we do not see when looking
1084 at the reference base alignment.
1086 2017-03-27 Richard Biener <rguenther@suse.de>
1089 * gimple-fold.c (fold_ctor_reference): Properly guard against
1090 NULL return value from canonicalize_constructor_val.
1092 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
1095 * config/i386/i386.c (ix86_expand_builtin)
1096 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
1097 flags reg setting and flags reg using instructions.
1098 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
1099 clobbering instructions to zero extend op2.
1101 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
1103 * doc/install.texi (Configuration) <--with-aix-soname>:
1104 Update link to AIX ld.
1106 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
1108 PR rtl-optimization/80160
1109 PR rtl-optimization/80159
1110 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
1111 reg_alternate_class into account.
1113 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
1116 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
1117 to consider in curr_insn_transform.
1119 2017-03-24 Jakub Jelinek <jakub@redhat.com>
1121 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
1122 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
1123 and emit_mode_inner.
1125 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1127 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
1128 argument to the overloaded builtin variants. Use the new flag to
1129 deprecate certain builtin variants.
1130 * config/s390/s390-builtin-types.def: Add new builtin types.
1131 * config/s390/s390-builtins.h: Support new flags field for
1132 overloaded builtins.
1133 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
1134 (s390_macro_to_expand): Enable vector float data type.
1135 (s390_cpu_cpp_builtins_internal): Indicate support of the new
1136 builtins by incrementing the __VEC__ version number.
1137 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
1139 (s390_resolve_overloaded_builtin): Emit error messages depending
1140 on the builtin flags.
1141 * config/s390/s390.c (s390_expand_builtin): Support additional
1142 flags argument. Change error message to match the messages
1143 emitted in s390-c.c.
1144 * config/s390/s390.md: New UNSPEC_* constants.
1145 (op_type): Add new instruction types.
1146 * config/s390/vecintrin.h: Add new builtins and test data class
1148 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
1149 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
1150 (VEC_INEXACT, VEC_NOINEXACT): New constants.
1151 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
1152 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
1153 ("vec_mergel<mode>"): V_HW -> VEC_HW.
1155 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
1156 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
1157 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
1158 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
1160 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
1161 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
1162 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
1163 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
1165 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
1166 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
1167 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
1168 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
1169 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
1170 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
1171 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
1173 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
1174 ("vec_scatter_element<V_HW_4:mode>_DI")
1175 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
1176 ("vec_fpint<mode>", "vflls")
1177 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
1178 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
1179 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
1180 ("*vec_cmphe<mode>_cc"): ... these.
1182 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
1183 mode constant instead of magic value.
1185 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1187 * config/s390/s390.c (s390_expand_vec_compare): Support other
1188 vector floating point modes than just V2DF.
1189 (s390_expand_vcond): Likewise.
1190 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
1191 (s390_cannot_change_mode_class): Prevent mode changes between TF
1192 and V1TF in vector registers.
1193 * config/s390/s390.md (DF, SF): New mode attributes.
1194 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
1195 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
1196 SFmode support for VRs.
1197 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
1199 (VFT, VF_HW): New mode iterators.
1200 (vw, sdx): New mode attributes.
1201 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
1202 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
1203 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
1204 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
1205 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
1206 also the new vector floating point modes. Renaming to ...
1208 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
1209 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
1210 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
1211 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
1212 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
1213 ("vec_unordered<mode>"): ... these.
1215 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
1216 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
1217 ("*vec_extendv2df"): New insn definitions.
1219 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1221 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
1222 ("mulditi3_2", "*muldi3_sign"): New patterns.
1223 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
1224 rename the pattern definition.
1226 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1228 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
1230 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
1232 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1234 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
1235 instruction if possible.
1236 * config/s390/vector.md (vec_halfnumelts): New mode
1238 ("*vec_vllezlf<mode>"): New pattern.
1240 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1242 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
1243 ("popcountv4si2", "popcountv2di2"): Rename to ...
1244 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
1245 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
1247 ("popcount<mode>2_vxe"): New pattern.
1249 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1251 * common/config/s390/s390-common.c (processor_flags_table): Add
1253 * config.gcc: Add arch12.
1254 * config/s390/driver-native.c (s390_host_detect_local_cpu):
1255 Default to arch12 for unknown CPU model numbers.
1256 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
1257 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
1258 PROCESSOR_max sanity check.
1259 * config/s390/s390-opts.h (enum processor_type): Add
1261 * config/s390/s390.c (processor_table): Add arch12.
1262 (s390_expand_builtin): Add check for B_VXE flag.
1263 (s390_issue_rate): Add PROCESSOR_ARCH12.
1264 (s390_get_sched_attrmask): Likewise.
1265 (s390_get_unit_mask): Likewise.
1266 (s390_sched_score): Enable z13 scheduling for arch12.
1267 (s390_sched_reorder): Likewise.
1268 (s390_sched_variable_issue): Likewise.
1269 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
1271 (s390_tune_attr): Use z13 scheduling also for arch12.
1272 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
1273 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
1274 (TARGET_VXE_P): New macros.
1275 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
1276 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
1277 * config/s390/s390.opt: Add arch12 as processor_type.
1279 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1281 * config/s390/s390.md
1282 ("fixuns_truncdddi2", "fixuns_trunctddi2")
1283 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
1284 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
1286 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
1287 Rename expanders to ...
1289 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
1290 ("fixuns_truncdddi2_emu"): ... these.
1292 ("fixuns_trunc<mode>si2_emu"): New expander.
1294 ("*fixuns_truncdfdi2_z13"): Rename to ...
1295 ("*fixuns_truncdfdi2_vx"): ... this.
1297 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1299 * config/s390/2964.md: Remove the single element vector compare
1300 instructions which are no longer used.
1301 * config/s390/s390.c (s390_select_ccmode): Remove handling of
1303 (s390_canonicalize_comparison): Remove handling of DFmode
1305 (s390_expand_vec_compare_scalar): Remove function.
1306 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
1307 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
1309 ("*cmp<mode>_ccs"): Add wfcdb instruction.
1311 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1313 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
1315 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
1316 will anyway by matched by mov<mode>_64dfp.
1318 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1320 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
1321 vlef/vstef. Add missing operand to vleif.
1323 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1325 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
1326 pair for all vector types with 64 bit elements.
1327 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
1328 * config/s390/vector.md (V_HW_64): ... here.
1329 (V_128_NOSINGLE): New mode iterator.
1330 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
1331 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
1332 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
1333 ("*vec_load_pairv2di"): Change to ...
1334 ("*vec_load_pair<mode>"): ... this one.
1336 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1338 * config/s390/constraints.md: Add comments.
1339 (jKK): Reject element sizes > 8 bytes.
1340 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
1342 * config/s390/s390.md: Add the s_operand checks formerly in
1343 s390_split_ok_p to various splitters where they are still
1345 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
1346 for 128 bit vectors. Plus two splitters.
1348 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1350 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
1353 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1356 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
1357 error if the boundary argument is not constant.
1359 2017-03-24 Jakub Jelinek <jakub@redhat.com>
1361 PR rtl-optimization/80112
1362 * loop-doloop.c (doloop_condition_get): Don't check condition
1363 if cmp isn't SET with IF_THEN_ELSE src.
1365 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1367 PR tree-optimization/80158
1368 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
1369 replacing a candidate statement, also replace it for the
1370 candidate's alternate interpretation.
1371 (replace_rhs_if_not_dup): Likewise.
1372 (replace_one_candidate): Likewise.
1374 2017-03-24 Richard Biener <rguenther@suse.de>
1376 PR tree-optimization/80167
1377 * graphite-isl-ast-to-gimple.c
1378 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
1380 (translate_isl_ast_to_gimple::get_rename): Likewise.
1382 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
1384 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
1385 handling of certain combinations of target options, including the
1386 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
1387 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
1389 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1392 * config/arm/arm.md (*load_multiple): Add reload_completed to
1395 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1396 Richard Biener <rguenth@suse.de>
1398 PR tree-optimization/79908
1399 PR tree-optimization/80136
1400 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
1401 been cast away, gimplify_and_add suffices.
1403 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
1405 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
1407 2017-03-23 Richard Biener <rguenther@suse.de>
1409 PR tree-optimization/80032
1410 * gimplify.c (gimple_push_cleanup): Forced unconditional
1411 cleanups still have to go to the conditional_cleanups
1414 2017-03-22 Jakub Jelinek <jakub@redhat.com>
1416 PR tree-optimization/80072
1417 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
1419 (next_operand_entry_id): Change type to unsigned int.
1420 (sort_by_operand_rank): Make sure to return the right return value
1421 even if unsigned fields are bigger than INT_MAX.
1422 (struct oecount): Change cnt and id type to unsigned int.
1423 (oecount_hasher::equal): Formatting fix.
1424 (oecount_cmp): Make sure to return the right return value
1425 even if unsigned fields are bigger than INT_MAX.
1426 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
1429 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
1430 TREE_READONLY on result if writing it more than once.
1433 * doc/invoke.texi (-fsanitize=thread): Document that with
1434 -fnon-call-exceptions atomics are not able to throw
1438 * tsan.c: Include tree-eh.h.
1439 (instrument_builtin_call): Call maybe_clean_eh_stmt or
1440 maybe_clean_or_replace_eh_stmt where needed.
1441 (instrument_memory_accesses): Add cfg_changed argument.
1442 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
1443 if it returned true.
1444 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
1446 PR rtl-optimization/63191
1447 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
1448 wrapper function, moved the whole old content into ...
1449 (ix86_delegitimize_address_1): ... this. New inline function.
1450 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
1451 true as last argument instead of ix86_delegitimize_address.
1453 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
1455 * config/aarch64/aarch64.c (generic_branch_cost): Copy
1456 cortexa57_branch_cost.
1458 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
1460 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
1462 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1465 * doc/md.texi (Constraints): Document wA constraint.
1466 * config/rs6000/constraints.md (wA): New.
1467 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
1468 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
1469 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
1470 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
1472 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
1475 * gimplify.c (is_oacc_declared): New function.
1476 (oacc_default_clause): Use it to set default flags for acc declared
1477 variables inside parallel regions.
1478 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
1480 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
1481 declare attribute to any decl as necessary.
1483 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1486 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
1487 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
1488 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
1489 (arm_arch_lpae): This.
1490 * config/arm/arm.c (arm_arch7ve): Rename into ...
1491 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
1492 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
1495 2017-03-22 Martin Liska <mliska@suse.cz>
1498 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
1499 error message instead of an ICE.
1501 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1503 * doc/extend.texi (6.11 Additional Floating Types): Revise.
1505 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
1507 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
1509 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1512 2017-03-21 Martin Sebor <msebor@redhat.com>
1514 * doc/extend.texi: Use "cannot" instead of "can't."
1515 * doc/hostconfig.texi: Same.
1516 * doc/install.texi: Same.
1517 * doc/invoke.texi: Same.
1518 * doc/loop.texi: Same.
1519 * doc/md.texi: Same.
1520 * doc/objc.texi: Same.
1521 * doc/rtl.texi: Same.
1522 * doc/tm.texi: Same.
1523 * doc/tm.texi.in: Same.
1524 * doc/trouble.texi: Same.
1526 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
1529 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
1530 (collect_checksum_attributes): Set it.
1531 (die_checksum_ordered): Use it.
1533 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1535 PR tree-optimization/79908
1536 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
1537 change: For a VA_ARG whose LHS has been cast away, use
1538 force_gimple_operand to construct the side effects.
1540 2017-03-21 David Malcolm <dmalcolm@redhat.com>
1542 PR translation/80001
1543 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
1544 more amenable to translation.
1545 (oacc_loop_auto_partitions): Likewise.
1547 2017-03-21 Marek Polacek <polacek@redhat.com>
1548 Martin Sebor <msebor@redhat.com>
1550 PR tree-optimization/80109
1551 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
1554 2017-03-21 Jakub Jelinek <jakub@redhat.com>
1555 Segher Boessenkool <segher@kernel.crashing.org>
1558 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
1559 check reg_used_between_p between insn and one of succ or succ2
1560 depending on if succ is artificial insn not inserted into insn
1563 2017-03-21 Martin Liska <mliska@suse.cz>
1565 PR gcov-profile/80081
1566 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
1567 * doc/gcc.texi: Include gcov-dump stuff.
1568 * doc/gcov-dump.texi: New file.
1570 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
1572 PR rtl-optimization/79150
1573 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
1574 conditional jump, if the jump is the last insn of the loop.
1576 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1577 Richard Biener <rguenth@suse.de>
1579 PR tree-optimization/79908
1580 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
1581 been cast away, use force_gimple_operand to construct the side
1584 2017-03-21 Martin Liska <mliska@suse.cz>
1587 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
1590 2017-03-21 Brad Spengler <spender@grsecurity.net>
1593 * plugin.c (htab_hash_plugin): New function.
1594 (add_new_plugin): Use it and adjust.
1595 (parse_plugin_arg_opt): Adjust.
1596 (init_one_plugin): Likewise.
1598 2017-03-21 Richard Biener <rguenther@suse.de>
1600 PR tree-optimization/80032
1601 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
1602 if set force the cleanup to happen unconditionally.
1603 (gimplify_target_expr): Push inserted clobbers with force_uncond
1604 to avoid them being removed by control-dependent DCE.
1606 2017-03-21 Richard Biener <rguenther@suse.de>
1608 PR tree-optimization/80122
1609 * tree-inline.c (copy_bb): Do not expans va-arg packs or
1610 va_arg_pack_len when the inlined call stmt requires pack
1612 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
1614 2017-03-21 Jakub Jelinek <jakub@redhat.com>
1617 * tsan.c (instrument_builtin_call): If the memory model argument
1618 is not a constant, assume it is valid.
1621 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
1624 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
1626 PR rtl-optimization/79910
1627 * combine.c (can_combine_p): Do not allow combining an I0 or I1
1628 if its dest is used by an insn before I2 (other than the combined
1629 insns themselves, which are properly handled already).
1631 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
1634 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
1636 * combine.c (record_used_regs): New static function.
1637 (try_combine): Handle situations where there is an additional
1638 instruction between I2 and I3 which needs to have a LOG_LINK
1642 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
1644 * combine.c (try_combine): Delete redundant i1 test. Call
1645 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
1647 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1650 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
1653 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1655 PR tree-optimization/80054
1656 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
1657 the optimization if a PHI or any of its arguments is not dominated
1658 by the candidate's basis. Use gphi* rather than gimple* as
1660 (replace_profitable_candidates): Clean up a gimple* variable that
1661 should be a gphi* variable.
1663 2017-03-20 Martin Sebor <msebor@redhat.com>
1666 * doc/extend.texi (attribute constructor): Document present limitation.
1668 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
1671 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
1672 __POWER9_VECTOR__ #ifdef control, change template definition to
1673 use Power9-specific built-in function.
1674 (vec_any_eq): Likewise.
1675 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
1676 to control outcomes from this test.
1677 (vector_ae_<mode>p): For VEC_F modes, likewise.
1679 2017-03-20 Ian Lance Taylor <iant@google.com>
1681 * config/i386/i386.c (ix86_function_regparm): Save an extra
1682 register for -fsplit-stack with DECL_STATIC_CHAIN.
1684 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
1687 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
1688 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
1690 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
1692 * config/riscv/riscv.c (riscv_print_operand): Use "fence
1694 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
1697 2017-03-20 Marek Polacek <polacek@redhat.com>
1700 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
1702 2017-03-20 Richard Biener <rguenther@suse.de>
1704 PR tree-optimization/80113
1705 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
1706 allocate extra SSA name for PHI def.
1707 (add_close_phis_to_outer_loops): Likewise.
1708 (add_close_phis_to_merge_points): Likewise.
1709 (copy_loop_close_phi_args): Likewise.
1710 (copy_cond_phi_nodes): Likewise.
1712 2017-03-20 Martin Liska <mliska@suse.cz>
1715 * tree-chkp.c (chkp_build_returned_bound): Do not build
1716 returned bounds for a LHS that's not a BOUNDED_P type.
1718 2017-03-20 Martin Liska <mliska@suse.cz>
1722 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
1723 COMPLEX_CST and VECTOR_CST.
1725 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1728 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
1729 target operand. A new splitter adds the clobber statement in case
1730 the target operand is dead anyway.
1732 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
1734 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
1735 to age-old versions of binutils and glibc.
1737 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
1739 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
1741 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
1743 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
1745 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
1747 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
1748 requirement for binutils 2.13.
1750 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
1752 * combine.c (try_combine): Delete redundant i1 test. Call
1753 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
1755 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
1757 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
1758 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
1760 <riscv64-*-elf>: Re-arrange section
1761 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
1762 <riscv32-*-linux>: Likewise.
1763 <riscv64-*-elf>: Likewise
1764 <riscv64-*-linux>: Likewise.
1766 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
1769 * aarch64.opt(verbose-cost-dump): Fix typo.
1771 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
1774 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
1775 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
1777 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
1779 * reload.c (find_reloads): When reloading a nonoffsettable address,
1780 use RELOAD_OTHER for it and its address reloads.
1782 PR rtl-optimization/79910
1783 * combine.c (record_used_regs): New static function.
1784 (try_combine): Handle situations where there is an additional
1785 instruction between I2 and I3 which needs to have a LOG_LINK
1788 2017-03-17 Jeff Law <law@redhat.com>
1790 PR tree-optimization/71437
1791 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
1792 conditional in the hash table first.
1793 (vrp_dom_walker::before_dom_children): Extract condition from
1794 ASSERT_EXPR. Record condition, its inverion and any implied
1797 2017-03-17 Marek Polacek <polacek@redhat.com>
1798 Markus Trippelsdorf <markus@trippelsdorf.de>
1800 PR tree-optimization/80079
1801 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
1804 2017-03-17 Richard Biener <rguenther@suse.de>
1807 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
1808 Properly verify the LHS before the RHS possibly claims to be
1810 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
1813 2017-03-17 Martin Jambor <mjambor@suse.cz>
1815 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
1816 (List of -O2 options): Likewise.
1817 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
1820 2017-03-17 Tom de Vries <tom@codesourcery.com>
1822 * gcov-dump.c (print_usage): Print bug_report_url.
1824 2017-03-17 Richard Biener <rguenther@suse.de>
1827 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
1828 (parser::peek): Likewise.
1830 2017-03-17 Richard Biener <rguenther@suse.de>
1832 PR tree-optimization/80048
1833 * sese.c (free_sese_info): Properly release rename_map and
1834 copied_bb_map elements.
1836 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
1838 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
1839 Add linked-list forward and backlinks. Insert on
1840 construction, remove on destruction.
1841 (class pass_store_merging): Add m_stores_head field.
1842 (pass_store_merging::terminate_and_process_all_chains):
1843 Iterate over m_stores_head list.
1844 (pass_store_merging::terminate_all_aliasing_chains):
1846 (pass_store_merging::execute): Check for debug stmts first.
1847 Push new chains onto the m_stores_head stack.
1849 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1852 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
1853 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
1854 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
1856 2017-03-16 Jeff Law <law@redhat.com>
1858 PR tree-optimization/71437
1859 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
1860 member function. Implementation moved into after_dom_children
1861 member function and into the threader's thread_outgoing_edges
1863 (dom_opt_dom_walker::after_dom_children): Simplify by moving
1864 some code into new thread_outgoing_edges.
1865 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
1866 definition. Simplify marker handling (do it here). Assume we always
1867 have the available expression and the const/copies tables.
1868 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
1870 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
1871 * tree-vrp.c (equiv_stack): No longer file scoped.
1872 (vrp_dom_walker): New class.
1873 (vrp_dom_walker::before_dom_children): New member function.
1874 (vrp_dom_walker::after_dom_children): Likewise.
1875 (identify_jump_threads): Setup domwalker. Use it rather than
1876 walking edges in a random order by hand. Simplify setup/finalization.
1877 (finalize_jump_threads): Remove.
1878 (vrp_finalize): Do not call identify_jump_threads here.
1879 (execute_vrp): Do it here instead and call thread_through_all_blocks
1882 PR tree-optimization/71437
1883 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
1885 (simplify_stmt_for_jump_threading): Add basic_block argument. All
1887 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
1888 (dom_opt_dom_walker::thread_across_edge): Remove
1889 handle_dominating_asserts argument. All callers changed.
1890 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
1891 changes. Remove calls to lhs_of_dominating_assert. Other
1892 uses of handle_dominating_asserts turn into unconditional code
1893 (simplify_control_stmt_condition_1): Likewise.
1894 (simplify_control_stmt_condition): Likewise.
1895 (thread_through_normal_block, thread_across_edge): Likewise.
1896 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
1897 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
1898 object if it is not an SSA_NAME.
1899 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
1900 before calling into the VRP specific simplifiers.
1901 (identify_jump_threads): Remove handle_dominating_asserts
1904 2017-03-16 Jakub Jelinek <jakub@redhat.com>
1907 * tree-diagnostic.c (default_tree_printer): No longer static.
1908 * tree-diagnostic.h (default_tree_printer): New prototype.
1910 2017-03-16 Tamar Christina <tamar.christina@arm.com>
1912 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
1913 Change ins into fmov.
1915 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1917 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
1918 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
1919 Use h_con constraint for operand 1.
1920 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
1921 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
1923 2017-03-15 Jeff Law <law@redhat.com>
1925 PR tree-optimization/71437
1926 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
1927 (record_temporary_equivalences): Use it.
1929 PR tree-optimization/71437
1930 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
1931 tree-ssa-scopedtables.
1932 (lookup_avail_expr, build_and_record_new_cond): Likewise.
1933 (record_conditions, record_cond, vuse_eq): Likewise.
1934 (record_edge_info): Adjust to API tweak of record_conditions.
1935 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
1936 (record_temporary_equivalences, optimize_stmt): Likewise.
1937 (eliminate_redundant_computations): Likewise.
1938 (record_equivalences_from_stmt): Likewise.
1939 * tree-ssa-scopedtables.c: Include options.h and params.h.
1940 (vuse_eq): New function, moved from tree-ssa-dom.c
1941 (build_and_record_new_cond): Likewise.
1942 (record_conditions): Likewise. Accept vector of conditions rather
1943 than edge_equivalence structure for first argument.
1944 for the first argument.
1945 (avail_exprs_stack::lookup_avail_expr): New member function, moved
1946 from tree-ssa-dom.c.
1947 (avail_exprs_stack::record_cond): Likewise.
1948 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
1949 from tree-ssa-dom.c.
1950 (avail_exprs_stack): Add new member functions lookup_avail_expr
1952 (record_conditions): Declare.
1954 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
1957 * lra-constraints.c (process_alt_operands): Increase reject for
1958 reloading an input/output operand.
1960 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
1963 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
1964 insns to convert from signed/unsigned char/short to IEEE 128-bit
1966 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
1968 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
1971 * config/i386/i386.c (ix86_vector_duplicate_value): Create
1972 subreg of inner mode for values already in registers.
1974 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
1976 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
1977 iteration reg is used after the loop.
1979 2017-03-14 Martin Sebor <msebor@redhat.com>
1981 PR tree-optimization/79800
1982 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
1983 precision in negative-positive range.
1984 (format_floating): Call non-const overload with adjusted precision.
1986 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
1989 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
1992 2017-03-14 Martin Sebor <msebor@redhat.com>
1995 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
1996 * builtins.def (aligned_alloc): Use it.
1999 * Makefile.in (GTFILES): Add calls.c.
2000 * calls.c: Include "gt-calls.h".
2002 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
2004 PR rtl-optimization/79728
2005 * regs.h (struct target_regs): New field
2006 x_contains_allocatable_regs_of_mode.
2007 (contains_allocatable_regs_of_mode): New macro.
2008 * reginfo.c (init_reg_sets_1): Initialize it, and change
2009 contains_reg_of_mode so it includes global regs as well.
2010 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
2011 rather than contains_regs_of_mode.
2013 2017-03-14 Martin Liska <mliska@suse.cz>
2015 * doc/invoke.texi: Document options that can't be combined with
2016 -fcheck-pointer-bounds.
2018 2017-03-14 Martin Liska <mliska@suse.cz>
2021 * doc/invoke.texi (-Wchkp): Document the option.
2023 2017-03-14 Martin Liska <mliska@suse.cz>
2025 * Makefile.in: Install gcov-dump.
2027 2017-03-14 Martin Liska <mliska@suse.cz>
2029 * multiple_target.c (expand_target_clones): Bail out for
2030 an invalid attribute.
2032 2017-03-14 Richard Biener <rguenther@suse.de>
2034 * alias.c (struct alias_set_entry): Pack properly.
2035 * cfgloop.h (struct loop): Likewise.
2036 * cse.c (struct set): Likewise.
2037 * ipa-utils.c (struct searchc_env): Likewise.
2038 * loop-invariant.c (struct invariant): Likewise.
2039 * lra-remat.c (struct cand): Likewise.
2040 * recog.c (struct change_t): Likewise.
2041 * rtl.h (struct address_info): Likewise.
2042 * symbol-summary.h (function_summary): Likewise.
2043 * tree-loop-distribution.c (struct partition): Likewise.
2044 * tree-object-size.c (struct object_size_info): Likewise.
2045 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
2046 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
2047 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
2048 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
2049 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
2050 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
2051 (struct _stmt_vec_info): Likewise.
2053 2017-03-14 Martin Liska <mliska@suse.cz>
2056 * multiple_target.c (create_dispatcher_calls): Check that
2057 a target can create a function dispatcher.
2059 2017-03-14 Martin Liska <mliska@suse.cz>
2062 * multiple_target.c (expand_target_clones): Drop local.local
2063 flag for default implementation.
2065 2017-03-14 Richard Biener <rguenther@suse.de>
2067 PR tree-optimization/80030
2068 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
2070 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
2072 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
2073 gcc_fallthrough() instead of __attribute__((fallthrough));
2075 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
2077 * doc/gcc.texi: Remove "up" link to (DIR).
2078 * doc/gccint.texi: Ditto.
2080 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
2082 * doc/install.texi (Specific) <avr>: Remove reference to
2085 2017-03-13 Jeff Law <law@redhat.com>
2087 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
2088 attribute rather than comments.
2090 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
2091 match_scratch operand is highest.
2093 2017-03-13 Martin Liska <mliska@suse.cz>
2096 * ipa-pure-const.c (warn_function_noreturn): If the declarations
2097 is a CHKP clone, use original declaration.
2099 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2101 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
2102 (arc_conditional_register_usage): Use a different allocation order
2103 when optimizing for size.
2104 * common/config/arc/arc-common.c (arc_option_optimization_table):
2105 Section anchors default on when optimizing for size.
2107 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2109 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
2111 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2113 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
2114 * config/arc/arc.md (cpu_facility): Add cd variant.
2115 (*movqi_insn): Add code density variant.
2116 (*movhi_insn): Likewise.
2117 (*movqi_insn): Likewise.
2118 (*addsi3_mixed): Likewise.
2119 (subsi3_insn): Likewise.
2121 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2123 * config/arc/arc.md (movsi_cond_exec): Update constraint.
2125 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2127 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
2128 expressions with MINUS and UNARY ops.
2130 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2133 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
2135 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
2136 between vec_select and vector argument.
2137 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
2138 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
2139 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
2140 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
2141 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
2142 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
2144 2017-03-13 Richard Biener <rguenther@suse.de>
2147 * params.def (vect-max-peeling-for-alignment): Fix typo.
2149 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2151 * doc/install.texi (Specific) <mips-*-*>: Remove description of
2152 issue that only occurred with binutils below 2.18.
2154 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2156 * doc/install.texi (Specific) <cris-axis-elf>: No longer
2157 refer to binutils 2.11/2.12 minimum.
2159 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2161 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
2162 ftp.kernel.org and simplify binutils requirement.
2164 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
2166 * doc/invoke.texi (Warning Options): Fix spelling of link-time
2168 (Optimize Options): Ditto. Also remove redundancy.
2170 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2172 PR translation/79848
2173 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
2175 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
2176 to G_ to avoid double translation.
2178 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2180 PR translation/79923
2181 * auto-profile.c (get_combined_location): Convert leading
2182 character of diagnostics to lower case and remove trailing period.
2183 (read_profile): Likewise for various diagnostics.
2184 * config/arm/arm.c (arm_option_override): Remove trailing period
2185 from various diagnostics.
2186 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
2187 (msp430_expand_delay_cycles): Likewise.
2189 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2192 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
2193 full command-line argument, rather than just "str".
2194 (aarch64_validate_march): Likewise.
2195 (aarch64_validate_mtune): Likewise.
2197 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
2199 PR rtl-optimization/78911
2200 * lra-assigns.c (must_not_spill_p): New function.
2201 (spill_for): Use it.
2203 2017-03-10 Jakub Jelinek <jakub@redhat.com>
2205 PR tree-optimization/79981
2206 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
2207 ATOMIC_COMPARE_EXCHANGE ifn result.
2208 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
2209 IFN_ATOMIC_COMPARE_EXCHANGE.
2211 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2214 * opts.c (parse_sanitizer_options): Add missing question mark to
2215 "did you mean" message.
2217 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2219 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
2221 (VMULEUH_UNS): Likewise.
2222 (VMULOUB_UNS): Likewise.
2223 (VMULOUH_UNS): Likewise.
2224 * config/rs6000/rs6000.c (builtin_function_type): Remove
2225 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
2227 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2230 * read-rtl-function.c (function_reader::read_rtx_operand): Update
2231 x with result of extra_parsing_for_operand_code_0.
2232 (function_reader::extra_parsing_for_operand_code_0): Convert
2233 return type from void to rtx, returning x. When reading
2234 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
2235 larger size containing struct block_symbol.
2237 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
2239 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
2240 -mfloat128-hardware without -m64.
2242 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
2245 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
2246 entries to the case statement that marks unsigned arguments to
2247 overloaded functions.
2249 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
2251 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
2252 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
2254 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
2257 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
2258 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
2260 2017-03-10 Martin Liska <mliska@suse.cz>
2264 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
2265 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
2268 2017-03-10 Olivier Hainque <hainque@adacore.com>
2270 * tree-switch-conversion (array_value_type): Start by resetting
2271 candidate type to it's main variant.
2273 2017-03-10 Jakub Jelinek <jakub@redhat.com>
2275 PR rtl-optimization/79909
2276 * combine.c (try_combine): Use simplify_replace_rtx on individual
2277 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
2278 of the whole CALL_INSN_FUNCTION_USAGE.
2280 PR tree-optimization/79972
2281 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
2282 get_range_info on SSA_NAMEs. Formatting fixes.
2284 2017-03-10 Richard Biener <rguenther@suse.de>
2285 Jakub Jelinek <jakub@redhat.com>
2287 PR tree-optimization/77975
2288 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
2289 edge to be constant.
2290 (get_val_for): For constant x return it. Formatting fix.
2291 (loop_niter_by_eval): Avoid pointless looping if the next iteration
2292 would use the same bases as the current one.
2294 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2296 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
2297 instead of vec_select for V1TImode.
2298 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
2300 (VSX_LE_128): Add V1TI to this mode iterator.
2301 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
2302 (*vsx_le_perm_store_<mode>): Likewise.
2303 (pre-reload splitter for VSX stores): Likewise.
2304 (post-reload splitter for VSX stores): Likewise.
2305 (*vsx_xxpermdi2_le_<mode>): Likewise.
2306 (*vsx_lxvd2x2_le_<mode>): Likewise.
2307 (*vsx_stxvd2x2_le_<mode>): Likewise.
2309 2017-03-09 Michael Eager <eager@eagercon.com>
2311 Correct failures with --enable-checking=yes,rtl.
2313 * config/microblaze/microblaze.c (microblaze_expand_shift):
2314 Replace GET_CODE test with CONST_INT_P and INTVAL test with
2315 test for const0_rtx.
2316 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
2317 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
2319 2017-03-09 Richard Biener <rguenther@suse.de>
2321 PR tree-optimization/79977
2322 * graphite-scop-detection.c (scop_detection::merge_sese):
2323 Handle the case of extra exits to blocks dominating the entry.
2325 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
2327 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2330 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
2332 PR rtl-optimization/79949
2333 * lra-constraints.c (process_alt_operands): Check memory when
2334 trying to predict a cycle. Print about the overall increase.
2336 2017-03-09 Richard Biener <rguenther@suse.de>
2339 * gimple-expr.c (useless_type_conversion_p): Preserve
2340 TYPE_SATURATING for fixed-point types.
2342 2017-03-09 Richard Biener <rguenther@suse.de>
2345 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
2346 alignment of BLKmode params.
2348 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2351 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
2352 (VALL_NO_V2Q): Likewise.
2354 * config/aarch64/aarch64-simd.md
2355 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
2357 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
2358 VALL_NO_V2Q mode iterator.
2359 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
2361 2017-03-09 Martin Liska <mliska@suse.cz>
2363 PR tree-optimization/79631
2364 * tree-chkp-opt.c (chkp_is_constant_addr): Call
2365 tree_int_cst_sign_bit just for INTEGER constants.
2367 2017-03-09 Martin Liska <mliska@suse.cz>
2371 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
2374 2017-03-09 Marek Polacek <polacek@redhat.com>
2377 * tree.c (inchash::add_expr): Handle TREE_VEC.
2379 2017-03-09 Martin Liska <mliska@suse.cz>
2382 (chkp_narrow_size_and_offset): New function.
2383 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
2384 (void chkp_parse_bit_field_ref): New function.
2385 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
2386 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
2388 2017-03-09 Martin Liska <mliska@suse.cz>
2391 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
2392 (chkp_find_bounds_1): Remove gcc_unreachable.
2394 2017-03-09 Jakub Jelinek <jakub@redhat.com>
2397 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
2398 BUILT_IN_SYNC*, determine the access type from the size suffix and
2399 always build a MEM_REF with that type. Handle forgotten
2400 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
2403 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
2404 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
2405 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
2406 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
2407 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
2408 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
2409 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
2410 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
2411 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
2412 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
2413 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
2414 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
2415 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
2416 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
2417 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
2418 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
2419 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
2420 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
2421 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
2422 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
2423 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
2424 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
2425 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
2426 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
2427 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
2428 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
2429 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
2430 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
2431 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
2432 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
2433 definitions outside of __OPTIMIZE__ guarded section.
2436 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
2437 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
2438 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
2439 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
2442 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2444 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
2445 ("vfenez<mode>"): Add missing constraints.
2447 2017-03-08 Martin Sebor <msebor@redhat.com>
2450 * config/nds32/nds32.c (nds32_option_override):
2451 Fix misspelled diagnostic.
2453 2017-03-08 Jakub Jelinek <jakub@redhat.com>
2456 * gimplify.c (gimplify_omp_for): Replace index var in outer
2457 taskloop statement with an artificial variable and add
2458 OMP_CLAUSE_PRIVATE clause for it.
2460 2017-03-08 Richard Biener <rguenther@suse.de>
2462 PR tree-optimization/79955
2463 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
2464 for accesses that are completely outside of the variable.
2466 2017-03-08 Andrew Haley <aph@redhat.com>
2468 PR tree-optimization/79943
2469 * tree-ssa-loop-split.c (compute_new_first_bound): When
2470 calculating the new upper bound, (END-BEG) should be added, not
2473 2017-03-08 Jakub Jelinek <jakub@redhat.com>
2475 * config/avr/avr.md (setmemhi): Make sure match_dup
2476 operand number comes before match_scratch.
2478 2017-03-08 Richard Biener <rguenther@suse.de>
2480 PR tree-optimization/79920
2481 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
2482 with ncopies == 1 to ...
2483 (vect_transform_slp_perm_load): ... here. Properly compute
2484 all element loads by iterating VF times over the group. Do
2485 not handle ncopies (computed in a broken way) in
2486 vect_create_mask_and_perm.
2488 2017-03-08 Jakub Jelinek <jakub@redhat.com>
2491 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
2492 is a uniform vector, use uniform_vector_p return value instead of
2493 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
2495 2017-03-07 Marek Polacek <polacek@redhat.com>
2498 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
2499 (alloca_call_type): Likewise.
2501 2017-03-07 Martin Liska <mliska@suse.cz>
2503 * gcov.c (process_args): Put comment to correct location.
2505 2017-03-07 Martin Liska <mliska@suse.cz>
2508 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
2509 Use array_at_struct_end_p instead of DECL_CHAIN (field).
2510 (chkp_narrow_bounds_for_field): Likewise.
2511 (chkp_parse_array_and_component_ref): Pass one more argument to
2514 2017-03-07 Richard Biener <rguenther@suse.de>
2516 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
2519 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
2521 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
2522 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
2524 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2527 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
2528 to end of description.
2529 (PARAM_MAX_STORES_TO_MERGE): Likewise.
2531 2017-03-07 Jakub Jelinek <jakub@redhat.com>
2533 PR rtl-optimization/79901
2534 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
2536 (*avx512f_<code><mode>3<mask_name>): ... this.
2537 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
2538 iterator instead of VI8_AVX2_AVX512BW.
2540 PR rtl-optimization/79901
2541 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
2542 min/max expander, expand it using expand_vec_cond_expr.
2545 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
2548 2017-03-06 Jakub Jelinek <jakub@redhat.com>
2551 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
2552 to void * for PCH reasons.
2553 * dwarf2out.c (output_loc_operands, output_die): Cast
2554 v.val_vec.array to unsigned char *.
2556 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
2559 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
2562 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
2564 PR rtl-optimization/79571
2565 * lra-constraints.c (process_alt_operands): Calculate static
2566 reject and subtract it from overall when only addresses will be
2569 2017-03-06 Julia Koval <julia.koval@intel.com>
2572 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
2573 incoming stack boundary to 128 for 64-bit targets.
2575 2017-03-06 Richard Biener <rguenther@suse.de>
2577 PR tree-optimization/79894
2578 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
2579 to NULL after folding it.
2581 2017-03-06 Richard Biener <rguenther@suse.de>
2583 PR tree-optimization/79824
2584 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
2585 check disabling peeling for gaps.
2587 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
2589 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
2590 attributes): Document gettimeofday.
2592 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
2594 * config/s390/s390.c (s390_option_override_internal): Set
2595 PARAM_MIN_VECT_LOOP_BOUND
2597 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
2599 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
2600 * config/s390/s390.md: Likewise.
2602 2017-03-06 Jakub Jelinek <jakub@redhat.com>
2605 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
2606 (<avx2_avx512>_perm<mode>): Rename to ...
2607 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
2609 (<avx512>_perm<mode>_mask): Rename to ...
2610 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
2612 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
2613 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
2614 instead of VI8F_256_512.
2615 (avx512f_perm<mode>): New define_expand.
2616 (avx512f_perm<mode>_mask): Likewise.
2617 (avx512f_perm<mode>_1<mask_name>): New define_insn.
2618 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
2620 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
2622 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
2623 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
2625 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
2627 2017-03-06 Martin Liska <mliska@suse.cz>
2630 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
2631 when having a SSA NAME w/o VAR_DECL assigned to it.
2633 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
2635 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
2636 msa_dpsub_<su>_d): Fix MODE for vec_select.
2638 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
2640 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
2642 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
2644 2017-03-06 Richard Biener <rguenther@suse.de>
2646 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
2647 * plugin.c (register_plugin_info): Likewise.
2648 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
2650 2017-03-05 Jakub Jelinek <jakub@redhat.com>
2652 * config/i386/sse.md (sse_storehps, sse_storelps,
2653 avx_<castmode><avxsizesuffix>_<castmode>,
2654 avx512f_<castmode><avxsizesuffix>_<castmode>,
2655 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
2656 in condition that at least one operand is not a MEM.
2658 2017-03-03 Jakub Jelinek <jakub@redhat.com>
2661 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
2662 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
2664 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
2665 gimple_call_nothrow_p flag based on whether original builtin can throw.
2666 If it can, emit following stmts on the fallthrough edge.
2667 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
2668 don't create new bb if inserting just debug stmts on the edge, try to
2669 insert them on the fallthru bb or just reset debug stmts.
2671 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
2674 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
2675 restore recog_data (including the operand rtxes inside it) around
2676 the call to get_insn_template.
2678 2017-03-03 Martin Sebor <msebor@redhat.com>
2680 PR tree-optimization/79699
2681 * context.c (context::~context): Free MPFR caches to avoid
2682 a memory leak on program exit.
2684 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2686 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
2687 Use wide_int::ulow () instead of .elt (0).
2689 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
2691 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
2692 (*pushxf): Limit oF constraint to 32bit targets and add oC
2693 constraint for 64bit targets.
2694 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
2695 (*pushdf): Change rmF constraint to rmC.
2697 2017-03-03 Martin Liska <mliska@suse.cz>
2699 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
2700 Remove unused variable.
2702 2017-03-03 Jakub Jelinek <jakub@redhat.com>
2705 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
2706 is a memory operand, increase num_memory.
2707 (ix86_expand_args_builtin): Likewise.
2709 2017-03-03 Jan Hubicka <jh@suse.cz>
2712 * ipa-devirt.c (maybe_record_node): Properly handle
2713 __cxa_pure_virtual visibility.
2715 2017-03-03 Martin Liska <mliska@suse.cz>
2717 PR tree-optimization/79803
2718 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
2720 (pass_loop_prefetch::execute): Disabled optimization if an
2721 assumption about L1 cache size is not met.
2723 2017-03-03 Martin Liska <mliska@suse.cz>
2725 PR rtl-optimization/79574
2726 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
2727 (hash_scan_set): Likewise.
2728 (dump_hash_table): Likewise.
2729 (hoist_code): Likewise.
2731 2017-03-03 Richard Biener <rguenther@suse.de>
2733 * fixed-value.c (fixed_from_string): Restore use of elt (1)
2734 in place of uhigh ().
2735 (fixed_convert_from_real): Likewise.
2737 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
2740 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
2742 2017-03-03 Richard Biener <rguenther@suse.de>
2745 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
2746 TYPE_OVERFLOW_UNDEFINED check.
2748 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2750 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
2752 (vector_ae_<mode>_p): Likewise.
2753 (vector_nez_<mode>_p): Likewise.
2754 (vector_ne_v2di_p): Likewise.
2755 (vector_ae_v2di_p): Likewise.
2756 (vector_ne_<mode>_p): Likewise.
2757 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
2759 (vsx_tsqrt<mode>2_fe): Likewise.
2761 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
2764 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
2766 2017-03-02 Jakub Jelinek <jakub@redhat.com>
2768 PR rtl-optimization/79780
2769 * cprop.c (one_cprop_pass): When second and further conditional trap
2770 in a single basic block is turned into an unconditional trap, turn it
2771 into a deleted note to avoid RTL verification failures.
2773 2017-03-02 Richard Biener <rguenther@suse.de>
2775 * fold-const.c (const_binop): Use ulow () instead of elt (0).
2777 2017-03-02 Richard Biener <rguenther@suse.de>
2779 PR tree-optimization/79345
2781 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
2782 param and abort the walk, returning -1 if it is hit.
2783 (walk_aliased_vdefs): Take a limit param and pass it on.
2784 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
2785 defaulting to 0 and return a signed int.
2786 * tree-ssa-uninit.c (struct check_defs_data): New struct.
2787 (check_defs): New helper.
2788 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
2789 about uninitialized memory.
2790 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
2791 bogus uninitialized warning.
2792 (fixed_convert_from_real): Likewise.
2794 2017-03-02 Bin Cheng <bin.cheng@arm.com>
2796 PR tree-optimization/66768
2797 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
2798 iv_use if base object can't be determined.
2800 2017-03-02 Jakub Jelinek <jakub@redhat.com>
2802 PR tree-optimization/79345
2803 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
2804 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
2805 (get_pattern_stats): Initialize it.
2806 * genemit.c (gen_expand): Verify match_scratch numbers come after
2807 match_operand/match_dup numbers.
2808 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
2809 match_scratch numbers.
2810 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
2812 * config/s390/s390.md (trunctdsd2): Likewise.
2814 2017-03-02 Richard Biener <rguenther@suse.de>
2816 * wide-int.h (wide_int_storage::operator=): Implement in terms
2819 2017-03-02 Richard Biener <rguenther@suse.de>
2821 PR tree-optimization/79777
2822 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
2823 the to insert expression to sth existing.
2825 2017-03-01 Martin Sebor <msebor@redhat.com>
2828 * gimple-ssa-sprintf.c
2829 (directive::known_width_and_precision): New function.
2830 (format_integer): Use it.
2831 (get_mpfr_format_length): Consider the full range of precision
2832 when computing %g output with the # flag. Set the likely byte
2833 count to 3 rather than 1 when precision is indeterminate.
2834 (format_floating): Correct the lower bound of precision.
2836 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2838 * doc/invoke.texi: Document default code model for 64-bit Linux.
2840 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2843 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
2844 udiv rather than div since input pattern is unsigned.
2846 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
2848 * config/i386/i386.c (print_reg): Warn for values of
2849 unsupported size in integer register.
2851 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2854 * config/rs6000/predicates.md (current_file_function_operand): Do
2855 not allow self calls to be local if the function is replaceable.
2857 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
2860 * config/rs6000/altivec.h (vec_ctz and others): Change the
2861 preprocessor macro that controls conditional compilation from
2862 _ARCH_PWR9 to __POWER9_VECTOR__.
2863 (vec_all_ne): Change parameterization of __altivec_scalar_pred
2864 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
2865 control (instead of _ARCH_PWR9 control) so that template
2866 definition uses power9-specific function.
2867 (vec_any_eq): Likewise.
2868 (vec_all_ne): Change macro definition to use a power9-specific
2869 expansion under #ifdef __POWER9_VECTOR__ control (instead of
2870 _ARCH_PWR9 control).
2871 (vec_any_eq) Likewise.
2872 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
2873 expansion for CMPNEF to remove support for xvcmpnesp instruction.
2874 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
2875 support for xvcmpnedp instruction.
2876 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
2877 macro expansion so that Power9 implementation of vec_all_ne does
2878 not use the AltiVec predicate framework.
2879 (VCMPNEH_P): Likewise.
2880 (VCMPNEW_P): Likewise.
2881 (VCMPNED_P): Likewise.
2882 (VCMPNEFP_P): Likewise.
2883 (VCMPNEDP_P): Likewise.
2884 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
2885 implementation of vec_any_eq to not use AltiVec predicate
2887 (VCMPAEH_P): Likewise.
2888 (VCMPAEW_P): Likewise.
2889 (VCMPAED_P): Likewise.
2890 (VCMPAEFP_P): Likewise.
2891 (VCMPAEDP_P): Likewise.
2892 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
2893 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
2894 not use the AltiVec predicate framework.
2895 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
2896 of vec_any_eq to not use AltiVec predicate framework.
2897 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
2898 support for predefined __POWER9_VECTOR__ macro to indicate that
2899 Power9 instruction selection is enabled.
2900 (altivec_overloaded_builtins): Remove extraneous
2901 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
2902 function argument types RS6000_BTI_bool_V16QI and
2903 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
2904 entry for overloaded function argument types RS6000_BTI_bool_V4SI
2905 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
2906 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
2907 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
2908 Power9 for implementations of vec_cmpne. Change the signature for
2909 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
2910 (representing vec_all_ne) to remove the previously described first
2911 argument of type RS6000_BTI_INTSI, as this was an artifact of
2912 reliance on the AltiVec predicate framework, which is no longer
2913 used in the implementation of these functions. Add
2914 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
2915 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
2916 since, unlike the AltiVec predicate framework implementation, we
2917 do not share function descriptors between vec_alle and vec_anyeq.
2918 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
2919 set of modes that receive special treatment even when
2920 TARGET_P9_VECTOR is true. The special treatment emits code that
2921 does not depend on Power9 instructions.
2922 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
2923 define_expand to not rely on AltiVec predicate framework.
2924 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
2926 (vector_ne_v2di_p): Change this define_expand to not rely on
2927 AltiVec predicate framework.
2928 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
2930 (vector_ne_<mode>_p): Change this define_expand to not rely on
2931 AltiVec predicate framework.
2932 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
2934 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
2935 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
2936 define_insn pattern.
2937 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
2938 define_insn pattern because the xvcmpne<VSs>. instruction is not
2940 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
2941 instruction is not supported.
2943 2017-03-01 Jakub Jelinek <jakub@redhat.com>
2945 * config/nvptx/nvptx.c: Include intl.h.
2947 2017-03-01 Martin Jambor <mjambor@suse.cz>
2950 * ipa-prop.h (ipa_bits): Removed field known.
2951 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
2952 to pointers. Adjusted their comments to warn about their sharing.
2953 (ipcp_transformation_summary): Change bits to a vector of pointers.
2954 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
2955 (ipa_get_ipa_bits_for_value): Declare.
2956 * tree-vrp.h (value_range): Mark as GTY((for_user)).
2957 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
2958 (ipa_bits_hash_table): Likewise.
2959 (ipa_vr_ggc_hash_traits): Likewise.
2960 (ipa_vr_hash_table): Likewise.
2961 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
2962 being pointers and vr_known being removed.
2963 (ipa_set_jf_unknown): Likewise.
2964 (ipa_get_ipa_bits_for_value): New function.
2965 (ipa_set_jfunc_bits): Likewise.
2966 (ipa_get_value_range): New overloaded functions.
2967 (ipa_set_jfunc_vr): Likewise.
2968 (ipa_compute_jump_functions_for_edge): Use the above functions to
2969 construct bits and vr parts of jump functions.
2970 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
2971 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
2973 (ipcp_grow_transformations_if_necessary): Also allocate
2974 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
2976 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
2977 them. Fix too long lines.
2978 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
2979 vr_known being removed.
2980 (ipa_read_jump_function): Use new setter functions to construct bits
2981 and vr parts of jump functions or set them to NULL.
2982 (write_ipcp_transformation_info): Adjust for bits being pointers.
2983 (read_ipcp_transformation_info): Likewise.
2984 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
2986 Include gt-ipa-prop.h.
2987 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
2989 (ipcp_store_bits_results): Likewise.
2990 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
2991 Do not write to existing jump functions but use a temporary instead.
2993 2017-03-01 Jakub Jelinek <jakub@redhat.com>
2996 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
2997 attempt to use its first operand as BIT_FIELD_REF base.
2999 2017-03-01 Richard Biener <rguenther@suse.de>
3002 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
3003 interpolating formula in wrapping arithmetic.
3004 (chrec_apply): Convert chrec_evaluate return value to wanted type.
3006 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3008 PR tree-optimization/79734
3009 * tree-vect-generic.c (expand_vector_condition): Optimize
3010 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
3011 Handle VEC_COND_EXPR where comparison has different inner width from
3014 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
3016 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
3017 markup, and similar issues. Remove @opindex entries for things
3018 that aren't options. Add missing -mmpy-option entries.
3020 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3022 PR tree-optimization/79737
3023 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
3024 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
3025 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
3026 instead of byte_size. Formatting fix.
3027 (shift_bytes_in_array_right): Formatting fix.
3029 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
3032 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
3033 condition on optimize for the leaf function test.
3035 2017-02-28 Martin Liska <mliska@suse.cz>
3038 * read-rtl-function.c (function_reader::handle_unknown_directive):
3039 Bail out when one uses -flto.
3041 2017-02-28 Martin Liska <mliska@suse.cz>
3043 * common.opt: Replace space with tabular for options of <number>
3045 * config/i386/i386.opt: Show <number> value for
3046 -mlarge-data-threshold.
3047 * opts.c (print_filtered_help): Do not display number in hexadecimal
3050 2017-02-28 Martin Liska <mliska@suse.cz>
3052 * common.opt: Fix --help=option -Q for options which are of
3055 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
3057 * config/i386/i386.c (print_reg): Error out for values
3058 of 8-bit size in invalid integer register.
3060 2017-02-28 Martin Sebor <msebor@redhat.com>
3062 PR tree-optimization/79691
3063 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
3065 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3068 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
3069 gcc_unreachable with output_operand_lossage.
3071 2017-02-28 Richard Biener <rguenther@suse.de>
3073 PR tree-optimization/79740
3074 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
3076 (visit_nary_op): Insert the nary into the hashtable if we
3077 pattern-matched sth.
3078 * tree-ssa-pre.c (eliminate_insert): Robustify.
3080 2017-02-28 Richard Biener <rguenther@suse.de>
3083 * fold-const.c (decode_field_reference): Reject out-of-bound
3086 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3088 * config/i386/i386.c: Include intl.h.
3089 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
3090 instead of just cond ? "..." : "...".
3091 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
3092 * coverage.c (read_counts_file): Likewise.
3093 * omp-offload.c: Include intl.h.
3094 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
3095 of just cond ? "..." : "...".
3096 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
3097 of just cond ? "..." : "...".
3099 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
3102 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
3104 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
3105 'tune for' CPU name.
3106 * config/arm/arm-cpu-data.h: Regenerated.
3108 2017-02-28 Richard Biener <rguenther@suse.de>
3110 PR tree-optimization/79732
3111 * tree-inline.c (expand_call_inline): Do not shadow var.
3113 2017-02-28 Richard Biener <rguenther@suse.de>
3115 PR tree-optimization/79723
3116 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
3117 address-space properly.
3119 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
3121 * doc/optinfo.texi (Optimization groups): Fix option used for
3123 * doc/invoke.texi (-fopt-info): Document "omp".
3124 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
3125 (OPTGROUP_ALL): Add OPTGROUP_OMP.
3126 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
3127 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
3128 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
3130 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
3132 * dumpfile.c (optgroup_options): Instead of "openmp", associate
3133 OPTGROUP_OMP with "omp".
3135 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
3138 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
3139 for arithmetic shift of unsigned V2DI.
3141 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
3143 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
3144 arc/linux.h headers.
3145 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
3146 (LINK_SPEC): Likewise.
3147 (ARC_TLS_EXTRA_START_SPEC): Likewise.
3148 (EXTRA_SPECS): Likewise.
3149 (STARTFILE_SPEC): Likewise.
3150 (ENDFILE_SPEC): Likewise.
3151 (LIB_SPEC): Likewise.
3152 (TARGET_SDATA_DEFAULT): Likewise.
3153 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
3154 (MULTILIB_DEFAULTS): Likewise.
3155 (DWARF2_UNWIND_INFO): Likewise.
3156 * config/arc/big.h: New file.
3157 * config/arc/elf.h: Likewise.
3158 * config/arc/linux.h: Likewise.
3159 * config/arc/t-uClibc: Remove.
3161 2017-02-27 Bin Cheng <bin.cheng@arm.com>
3163 PR tree-optimization/77536
3164 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
3165 (tree_transform_and_unroll_loop): Use above function to compute the
3166 estimated niter of unrolled loop and use it when scaling profile.
3167 Also use count info rather than frequency if it's non-zero.
3168 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
3169 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
3170 (vect_transform_loop): Call above function.
3172 2017-02-27 Richard Biener <rguenther@suse.de>
3174 PR tree-optimization/45397
3175 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
3176 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
3177 (visit_nary_op): Add pattern matching for CSEing sign-changed
3178 or truncated operations with wider ones.
3180 2017-02-27 Richard Biener <rguenther@suse.de>
3182 PR tree-optimization/79690
3183 * tree-vect-stmts.c (vectorizable_store): Use vector type
3184 built from the DR with address-space.
3186 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
3188 * doc/invoke.texi (Optimize Options): Refine the description
3189 of asan-use-after-return.
3191 2017-02-25 Alan Modra <amodra@gmail.com>
3193 PR rtl-optimization/79584
3194 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
3195 base, not ad->base_term, the reg within base. Remove assertion
3196 that ad->base == ad->base_term. Replace gen_int_mode using
3197 bogus mode with const0_rtx.
3199 2017-02-25 Jakub Jelinek <jakub@redhat.com>
3202 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
3203 FMA_EXPR like tcc_binary or tcc_unary.
3205 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
3208 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
3210 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
3211 (output_loc_operands): Handle DW_OP_call_ref and
3212 DW_OP_GNU_variable_value.
3213 (struct variable_value_struct): New type.
3214 (struct variable_value_hasher): Likewise.
3215 (variable_value_hash): New variable.
3216 (string_types): Remove.
3217 (copy_loc_descr): New function.
3218 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
3219 (prepend_loc_descr_to_each): New function.
3220 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
3221 instead of add_loc_descr_to_each if the first argument is single
3222 location list and the second has multiple.
3223 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
3224 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
3225 when looking for variable value which doesn't have other location info.
3226 (loc_list_from_tree): Formatting fix.
3227 (gen_array_type_die): Simplify DW_AT_string_length handling.
3228 (adjust_string_types): Remove.
3229 (gen_subprogram_die): Don't call adjust_string_types nor test/set
3230 string_types. Call resolve_variable_values.
3231 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
3232 (resolve_addr_in_expr): Likewise. Add A argument.
3233 (copy_deref_exprloc): Remove deref argument. Adjust for the
3234 original expression being DW_OP_GNU_variable_value with optionally
3235 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
3236 optionally after it.
3237 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
3238 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
3239 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
3240 (variable_value_hasher::hash, variable_value_hasher::equal): New
3242 (resolve_variable_value_in_expr, resolve_variable_value,
3243 resolve_variable_values, note_variable_value_in_expr,
3244 note_variable_value): New functions.
3245 (dwarf2out_early_finish): Call note_variable_value on all toplevel
3248 2017-02-24 Jakub Jelinek <jakub@redhat.com>
3251 * opts.h (handle_generated_option): Add GENERATED_P argument.
3252 * opts-common.c (handle_option): Adjust function comment.
3253 (handle_generated_option): Add GENERATED_P argument, pass it to
3255 (control_warning_option): Pass false to handle_generated_option
3257 * opts.c (maybe_default_option): Pass true to handle_generated_option
3259 * optc-gen.awk: Likewise.
3261 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3263 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
3264 a REG, look at the REG it is a SUBREG of.
3265 (splitter for cmpeqsi_t): Ditto.
3267 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3269 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
3270 the special USEs with the pattern of the insn, not the insn itself.
3272 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
3275 * doc/invoke.texi: Document -mload-store-pairs.
3277 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3278 Sandra Loosemore <sandra@codesourcery.com>
3280 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
3281 argument isn't a CONST_INT.
3282 (nios2_alternate_compare_const): Assert op is a CONST_INT.
3283 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
3284 (nios2_validate_compare): Bypass alternate compare logic if *op2
3286 (ldstwm_operation_p): Return false if first_base is not a REG or
3287 if first_offset is not a CONST_INT.
3289 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3291 * config/cris/cris.md: Use correct operand in a define_peephole2.
3293 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3295 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
3297 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3299 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
3300 this_insn if it is an INSN or JUMP_INSN.
3301 (force_offsettable): Look at base, not at addr.
3302 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
3303 on things that aren't necessarily CONST_INTs.
3305 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
3307 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
3308 -mfpmath=sse is the default also for x86-32 targets with SSE2
3309 instruction set when @option{-ffast-math} is enabled
3311 2017-02-24 Jeff Law <law@redhat.com>
3313 PR rtl-optimizatoin/79286
3314 * ira.c (update_equiv_regs): Drop may_trap_p exception to
3317 2017-02-24 Richard Biener <rguenther@suse.de>
3319 PR tree-optimization/79389
3320 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
3323 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
3325 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
3326 function comment to reflect reality.
3327 (loop_exits_before_overflow): Fix typo in function description.
3329 2017-02-24 Richard Biener <rguenther@suse.de>
3331 PR tree-optimization/79389
3332 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
3333 properly that a threading opportunity exists. Detect conditional
3334 copy/constant propagation opportunities.
3336 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
3338 * config/visium/visium.md (type): Add trap.
3339 (b): New mode attribute.
3340 (*btst): Rename into...
3341 (*btst<mode>): ...this and adjust.
3342 (*cbranchsi4_btst_insn): Rename into...
3343 (*cbranch<mode>4_btst_insn): ...this and adjust.
3344 (trap): New define_insn.
3346 2017-02-23 Jakub Jelinek <jakub@redhat.com>
3348 PR tree-optimization/79389
3349 * ifcvt.c (struct noce_if_info): Add rev_cond field.
3350 (noce_reversed_cond_code): New function.
3351 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
3352 reversed_comparison_code. Formatting fix.
3353 (noce_try_store_flag): Test rev_cond != NULL in addition to
3354 reversed_comparison_code.
3355 (noce_try_store_flag_constants): Likewise.
3356 (noce_try_store_flag_mask): Likewise.
3357 (noce_try_addcc): Use rev_cond if non-NULL instead of
3358 reversed_comparison_code.
3359 (noce_try_cmove_arith): Likewise. Formatting fixes.
3360 (noce_try_minmax, noce_try_abs): Clear rev_cond.
3361 (noce_find_if_block): Initialize rev_cond.
3362 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
3363 instead of false as last argument never attempt to reverse it
3366 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3368 PR tree-optimization/79663
3369 * tree-predcom.c (combine_chains): Process refs in reverse order
3370 only for ZERO length chains, and add explaining comment.
3372 2017-02-23 Jeff Law <law@redhat.com>
3374 PR tree-optimization/79578
3375 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
3376 in call to operand_equal_p.
3378 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
3381 * config/i386/cpuid.h: Fix another undefined behavior.
3383 2017-02-23 Richard Biener <rguenther@suse.de>
3385 PR tree-optimization/79683
3386 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
3387 vector types for data-refs.
3389 2017-02-23 Martin Liska <mliska@suse.cz>
3391 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
3393 2017-02-23 Jakub Jelinek <jakub@redhat.com>
3396 * internal-fn.c (get_range_pos_neg): Moved to ...
3397 * tree.c (get_range_pos_neg): ... here. No longer static.
3398 * tree.h (get_range_pos_neg): New prototype.
3399 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
3400 are known to be in between 0 and signed maximum inclusive, try to
3401 expand both unsigned and signed divmod and use the cheaper one from
3404 2017-02-22 Jeff Law <law@redhat.com>
3406 PR tree-optimization/79578
3407 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
3408 to compare base operands.
3410 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
3413 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
3414 gpc_reg_operand instead of fpr_reg_operand.
3416 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
3418 * config/mips/mips.c (mips_return_in_memory): Force FP
3419 vector types to be returned in memory for o32 ABI.
3421 2017-02-22 Jakub Jelinek <jakub@redhat.com>
3423 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
3424 instead of DW_TAG_member for static data member declarations and don't
3425 set no_linkage_name for static inline data members.
3426 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
3429 2017-02-22 Martin Liska <mliska@suse.cz>
3431 * doc/invoke.texi: Replace inequality signs with square brackets
3434 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3436 PR tree-optimization/68644
3437 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
3439 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
3442 * lra-constraints.c (simplify_operand_subreg): Handle
3443 WORD_REGISTER_OPERATIONS targets.
3445 2017-02-22 Jakub Jelinek <jakub@redhat.com>
3448 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
3449 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
3450 elimination by swapping fld*.
3452 2017-02-22 Richard Biener <rguenther@suse.de>
3454 PR tree-optimization/79673
3455 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
3456 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
3457 irrelevant address-space qualifiers and avoiding a
3458 ADDR_SPACE_CONVERT_EXPR from fold_convert.
3460 2017-02-22 Richard Biener <rguenther@suse.de>
3462 PR tree-optimization/79666
3463 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
3464 to not symbolically negate if that may introduce undefined
3467 2017-02-22 Martin Liska <mliska@suse.cz>
3470 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
3471 * data-streamer-out.c (streamer_write_gcov_count_stream):
3473 * value-prof.c (stream_out_histogram_value): Make assert more
3474 precise based on type of counter.
3476 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
3479 * config/i386/i386.md (standard_x87sse_constant_load splitter):
3480 Use nonimmediate_operand instead of memory_operand for operand 1.
3481 (float-extend standard_x87sse_constant_load splitter): Ditto.
3483 2017-02-21 Jeff Law <law@redhat.com>
3485 PR tree-optimization/79621
3486 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
3487 blocks with edges to themselves.
3489 2017-02-21 Jakub Jelinek <jakub@redhat.com>
3492 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
3493 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
3494 Use gimple_call_builtin_p.
3497 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
3498 on temporarily removed DEBUG_INSNs.
3500 PR tree-optimization/79649
3501 * tree-loop-distribution.c (classify_partition): Give up on
3502 non-generic address space loads/stores.
3504 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
3506 * doc/loop.texi (Loop manipulation): Remove nonexistent
3507 tree_ssa_loop_version from the documentation.
3508 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
3510 2017-02-21 Jakub Jelinek <jakub@redhat.com>
3513 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
3514 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
3515 * config/rs6000/rs6000.c: Include except.h.
3516 (rs6000_expand_split_stack_prologue): Call
3517 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
3519 2017-02-21 Martin Jambor <mjambor@suse.cz>
3522 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
3525 2017-02-21 Martin Jambor <mjambor@suse.cz>
3527 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
3528 for backward compatibility only.
3529 * doc/invoke.texi (Option Summary): Remove all references to
3532 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
3536 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3538 * lra-constraints.c (curr_insn_transform): Handle
3539 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
3541 2017-02-21 Martin Liska <mliska@suse.cz>
3543 * config/i386/i386.opt: Replace -masm-dialect with -masm.
3545 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
3547 PR translation/79638
3548 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
3550 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
3553 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
3554 (arm_function_ok_for_sibcall): Return false for an indirect call by
3555 descriptor if all the argument registers are used.
3556 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
3557 alignment of the function.
3559 2017-02-21 Jakub Jelinek <jakub@redhat.com>
3561 PR tree-optimization/61441
3562 * simplify-rtx.c (simplify_const_unary_operation): For
3563 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
3564 the sNaN unmodified.
3566 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
3568 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
3569 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
3570 instead of SYSTEM_HEADER_DIR.
3572 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
3573 Martin Liška <mliska@suse.cz>
3575 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
3576 Fix typos and grammar, use active voice, and clarify.
3578 2017-02-20 Marek Polacek <polacek@redhat.com>
3581 * gimplify.c (gimplify_expr): Handle unused *&&L;.
3584 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
3586 2017-02-20 Jakub Jelinek <jakub@redhat.com>
3589 * config/i386/i386.c (ix86_expand_builtin): Handle
3590 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
3591 ix86_builtins_isa[fcode].isa as a requirement of those
3592 flags and any other flag in the bitmask.
3593 (ix86_init_mmx_sse_builtins): Use 0 instead of
3594 ~OPTION_MASK_ISA_64BIT as mask.
3595 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
3596 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
3597 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
3598 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
3600 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3603 * lra-constraints.c (split_reg): Check requested split mode
3604 is supported by the register.
3606 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3608 * lra-constraints.c (simplify_operand_subreg): Remove early
3611 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3614 * lra-constraints.c (curr_insn_transform): Tighten condition
3615 for converting SUBREG reloads from OP_OUT to OP_INOUT.
3617 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3620 * lra-constraints.c (curr_insn_transform): Handle
3621 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
3623 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
3626 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
3628 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
3630 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
3633 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
3636 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3638 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
3641 2017-02-18 Jakub Jelinek <jakub@redhat.com>
3643 * final.c (last_columnnum, override_columnnum): New variables.
3644 (final_start_function): Set last_columnnum, pass it to begin_prologue
3645 hook and pass 0 to dwarf2out_begin_prologue.
3646 (final_scan_insn): Update override_columnnum. Pass last_columnnum
3647 to source_line debug hook.
3648 (notice_source_line): Compute last_columnnum and for debug_column_info
3649 return true on column changes.
3650 * debug.h (struct gcc_debug_hooks): Add column argument to
3651 source_line and begin_prologue hooks.
3652 (debug_nothing_int_charstar_int_bool): Remove prototype.
3653 (debug_nothing_int_int_charstar,
3654 debug_nothing_int_int_charstar_int_bool): New prototypes.
3655 (dwarf2out_begin_prologue): Add column argument.
3656 * debug.c (do_nothing_debug_hooks): Adjust source_line and
3657 begin_prologue hooks.
3658 (debug_nothing_int_charstar_int_bool): Remove.
3659 (debug_nothing_int_int_charstar,
3660 debug_nothing_int_int_charstar_int_bool): New functions.
3661 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
3662 through to dwarf2out_source_line.
3663 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
3664 (dwarf2out_source_line): Add column argument, emit it if requested.
3665 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
3667 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
3668 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
3669 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
3670 through to dwarf2out_begin_prologue.
3671 (vmsdbgout_source_line): Add column argument, pass it through to
3672 dwarf2out_source_line.
3673 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
3674 dbxout_source_line caller.
3675 (dbxout_source_line): Add column argument.
3677 * common.opt (gno-column-info, gcolumn-info): New options.
3678 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
3679 (check_die): Also test for multiple DW_AT_decl_column attributes.
3680 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
3681 DW_AT_decl_column if requested.
3682 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
3684 (gen_variable_die): Likewise.
3685 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
3686 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
3689 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
3690 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
3691 (ix86_handle_option): Handle OPT_m3dnowa.
3692 * doc/invoke.texi (-m3dnowa): Document.
3693 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
3694 -m3dnowa instead of -m3dnow -march=athlon.
3697 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
3698 instead of gcc_assert for K, r and R code checks. Formatting fixes.
3700 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3703 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
3704 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
3705 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
3706 generator for vsx_xxpermdi_<mode>_be.
3707 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
3708 force big-endian semantics.
3709 (vsx_xxpermdi_<mode>_be): New define_expand with same
3710 implementation as previous version of vsx_xxpermdi_<mode>.
3712 2017-02-17 Jakub Jelinek <jakub@redhat.com>
3714 PR tree-optimization/79327
3715 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
3716 variable, its initialization and use.
3718 2017-02-17 Julia Koval <julia.koval@intel.com>
3720 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
3721 (OPTION_MASK_ISA_PKU_UNSET): New.
3722 (ix86_handle_option): Handle -mrdpid.
3723 * config/i386/cpuid.h (bit_RDPID): New.
3724 * config/i386/driver-i386.c (host_detect_local_cpu):
3725 Detect RDPID feature.
3726 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
3727 * config/i386/i386-c.c (ix86_target_macros_internal):
3729 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
3730 (ix86_valid_target_attribute_inner_p): Add "rdpid".
3731 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
3732 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
3733 * config/i386/i386.md (define_insn "rdpid"): New.
3734 * config/i386/i386.opt Add -mrdpid.
3735 * config/i386/immintrin.h (_rdpid_u32): New.
3737 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
3739 PR rtl-optimization/79541
3740 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
3741 instead of transforming it into USE.
3743 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
3745 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
3746 If HONOR_SNANS (SFmode) force the input to a register.
3747 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
3748 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
3749 an frsp or similar insn.
3751 2017-02-17 Martin Liska <mliska@suse.cz>
3753 PR rtl-optimization/79577
3754 * params.def (selsched-max-sched-times): Increase minimum to 1.
3756 2017-02-17 Martin Liska <mliska@suse.cz>
3758 PR rtl-optimization/79574
3759 * gcse.c (want_to_gcse_p): Prevent integer overflow.
3761 2017-02-17 Martin Liska <mliska@suse.cz>
3763 PR tree-optimization/79529
3764 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
3765 ssa_defined_default_def_p to handle cases which are implicitly
3767 * tree-ssa.c (ssa_defined_default_def_p): New function.
3768 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
3769 which are implicitly defined.
3770 * tree-ssa.h (ssa_defined_default_def_p): Declare.
3772 2017-02-17 Richard Biener <rguenther@suse.de>
3775 * params.def (max-ssa-name-query-depth): Limit to 10.
3777 2017-02-17 Richard Biener <rguenther@suse.de>
3779 PR tree-optimization/79552
3780 * tree-ssa-structalias.c (visit_loadstore): Properly verify
3783 2017-02-17 Richard Biener <rguenther@suse.de>
3786 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
3788 2017-02-17 Marek Polacek <polacek@redhat.com>
3791 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
3792 (fold_negate_expr): New wrapper.
3794 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
3796 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
3797 Correct terminology and de-emphasize pre-standard behavior.
3799 2017-02-16 Alan Modra <amodra@gmail.com>
3801 PR rtl-optimization/79286
3802 * ira.c (def_dominates_uses): New function.
3803 (update_equiv_regs): Don't create an equivalence for insns that
3804 may trap where the register def does not dominate the use.
3806 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
3808 PR rtl-optimization/78127
3809 * lra.c (lra): Call lra_eliminate before finish the loop after
3812 2017-02-16 Richard Biener <rguenther@suse.de>
3814 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
3816 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
3817 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
3818 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
3819 (isl_val_int_from_wi): New function.
3820 (extract_affine_gmp): Rename to ...
3821 (extract_affine_wi): ... this, take a widest_int.
3822 (extract_affine_int): Just wrap extract_affine_wi.
3823 (add_param_constraints): Use isl_val_int_from_wi.
3824 (add_loop_constraints): Likewise, and extract_affine_wi.
3826 2017-02-15 Jeff Law <law@redhat.com>
3829 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
3830 ira_init_register_move_cost_if_necessary.
3832 2017-02-15 Martin Sebor <msebor@redhat.com>
3835 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
3836 removed in a prior commit.
3838 2017-02-15 Bin Cheng <bin.cheng@arm.com>
3840 PR tree-optimization/79347
3841 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
3842 counters during peeling.
3844 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
3846 * Makefile.in (site.exp): Remove "set ISLVER".
3848 2017-02-15 Jakub Jelinek <jakub@redhat.com>
3851 * real.c (real_from_integer): Call real_convert even for decimal.
3853 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
3856 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
3858 2017-02-14 Andrew Pinski <apinski@cavium.com>
3860 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
3861 cores and change the partno/implementer to be correct.
3862 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
3863 the 'B" as the implementer.
3864 * config/aarch64/aarch64-tune.md: Regenerate.
3866 2017-02-14 Carl Love <cel@us.ibm.com>
3868 * config/rs6000/rs6000.c: Add case statement entry to make the
3869 xvcvuxdsp built-in argument unsigned.
3870 * config/rs6000/vsx.md: Fix the source and return operand types so they
3871 match the instruction definitions from the ISA document. Fix typo
3872 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
3875 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
3878 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
3879 member early_clobber_alts.
3880 * lra-lives.c (reg_early_clobber_p): New.
3881 (process_bb_lives): Use it.
3882 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
3883 (debug_operand_data): Initialize early_clobber_alts.
3884 (setup_operand_alternative): Set up early_clobber_alts.
3885 (collect_non_operand_hard_regs): Ditto. Pass early clobber
3886 alternatives to new_insn_reg.
3887 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
3889 (lra_update_insn_regno_info): Pass the new arg.
3891 2017-02-14 Jakub Jelinek <jakub@redhat.com>
3894 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
3895 (new_oacc_loop_raw): Don't clear already cleared fields.
3898 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
3899 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
3900 _mm512_prefetch_i64gather_ps): New inline functions and macros.
3902 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
3905 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
3907 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
3910 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
3911 the extra instruction to the right place to store 128-bit constant
3914 2017-02-14 Martin Sebor <msebor@redhat.com>
3917 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
3918 warning for strings of unknown length.
3920 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
3922 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
3924 2017-02-14 Jeff Law <law@redhat.com>
3927 * ira-costs.c (scan_one_insn): Initialize register move costs
3928 for pseudos seen in USE/CLOBBER insns.
3930 PR tree-optimization/79095
3931 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
3932 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
3933 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
3934 if the operands are known to be not equal, then the resulting range
3936 (intersect_ranges): If the new range is ~[0,0] and the old range is
3937 wide, then prefer ~[0,0].
3938 * tree-vrp.c (overflow_comparison_p_1): New function.
3939 (overflow_comparison_p): New function.
3940 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
3941 if NAME is used in an overflow test.
3942 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
3943 overflow check that can be expressed as an equality test, then adjust
3944 ops to be that equality test.
3946 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3948 * config/s390/s390-builtin-types.def: Remove flags argument.
3949 * config/s390/s390.c (s390_init_builtins): Likewise.
3951 2017-02-14 Martin Liska <mliska@suse.cz>
3953 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
3954 vector. Fix trailing white spaces.
3956 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
3958 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
3961 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3963 PR rtl-optimization/68664
3964 * config/arm/arm.c (arm_sched_can_speculate_insn):
3965 New function. Declare prototype.
3966 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
3968 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3970 PR rtl-optimization/68664
3971 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
3973 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
3975 2017-02-14 Amit Pawar <amit.pawar@amd.com>
3977 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
3978 max skip bytes for function, loop and jump.
3980 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3982 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
3983 ABS_EXPR for gimple dump.
3985 2017-02-14 Jakub Jelinek <jakub@redhat.com>
3988 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
3990 PR tree-optimization/79408
3991 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
3992 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
3993 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
3994 also if rhs1 is INTEGER_CST.
3996 2017-02-14 Richard Biener <rguenther@suse.de>
3999 * tree-into-ssa.c (insert_phi_nodes): When the function can
4000 have abnormal edges rewrite SSA names with broken use-def
4001 dominance out of SSA and register them for PHI insertion.
4003 2017-02-13 Martin Sebor <msebor@redhat.com>
4006 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
4007 clearing info.nowrite flag when snprintf size argument is a range.
4009 2017-02-13 Jakub Jelinek <jakub@redhat.com>
4011 * cprop.c (cprop_jump): Add missing space in string literal.
4012 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
4013 (get_constraint_for_component_ref): Likewise.
4014 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
4015 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
4016 * lra-constraints.c (process_alt_operands): Likewise.
4017 * ipa-inline.c (inline_small_functions): Likewise.
4018 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
4019 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
4020 * trans-mem.c (diagnose_tm_1_op): Likewise.
4021 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
4022 (grid_parallel_clauses_gridifiable): Likewise.
4024 * config/nvptx/mkoffload.c (process): Add space in between
4027 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
4028 "MOD4_SSE_REGS" and "ALL_REGS".
4030 * spellcheck.c (test_data): Add , in between "foo" and "food".
4032 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4035 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
4036 boundary crossing check and subsequent code generation agree.
4038 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4040 * config/aarch64/aarch64.c (has_memory_op): Delete.
4041 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
4044 2017-02-13 Jakub Jelinek <jakub@redhat.com>
4046 PR rtl-optimization/79388
4047 PR rtl-optimization/79450
4048 * combine.c (distribute_notes): When removing TEM_INSN for which
4049 corresponding dest has last value recorded, invalidate that last
4052 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4054 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
4055 of explicit '@'. Add missing assembly comment marker on branch costs
4058 2017-02-13 Nathan Sidwell <nathan@acm.org>
4060 * gengtype-lex.l (<in_struct>): Add '/'.
4062 2017-02-13 Martin Liska <mliska@suse.cz>
4065 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
4067 2017-02-13 Richard Biener <rguenther@suse.de>
4069 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
4071 * configure: Re-generate.
4072 * config.in: Likewise.
4073 * graphite-dependences.c: Simplify as if
4074 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
4075 * graphite-isl-ast-to-gimple.c: Likewise.
4076 * graphite-optimize-isl.c: Likewise.
4077 * graphite-poly.c: Likewise.
4078 * graphite-sese-to-poly.c: Likewise.
4079 * graphite.h: Likewise.
4080 * toplev.c: Include isl/version.h and use isl_version () for
4081 printing the ISL version.
4082 * doc/install.texi: Update ISL requirement.
4084 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
4086 * doc/standards.texi (Standards): Update reference to
4089 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
4091 * doc/extend.texi (Named Address Spaces): sourceware.org now
4093 * doc/install.texi (Binaries): Ditto.
4096 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
4098 * doc/cpp.texi: Replace "stringify"/"stringification" with C
4099 standard terminology "stringize"/"stringizing" throughout.
4100 * doc/cppinternals.texi: Likewise.
4102 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
4104 * doc/extend.texi: Fix some spelling mistakes and typos.
4105 * doc/invoke.texi: Likewise.
4107 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
4110 * params.def (inline-min-speedup) Change from 10 to 8.
4112 2017-02-11 Jakub Jelinek <jakub@redhat.com>
4114 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
4117 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
4120 * ipa-inline-analysis.c (get_minimal_bb): New function.
4121 (record_modified): Use it.
4122 (remap_edge_change_prob): Handle also ancestor functions.
4124 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
4126 * doc/contrib.texi (Contributors): Remove broken link into
4127 the Mauve CVS repository.
4129 2017-02-11 Jakub Jelinek <jakub@redhat.com>
4132 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
4133 result computation whenever lhs doesn't have vector mode, not
4134 just when it has BLKmode.
4136 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
4138 * doc/makefile.texi (profiledbootstrap): Refer to the
4139 installation instructions only in textual form.
4141 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4144 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
4146 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
4148 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
4149 (Specific): Update mingw-w64 reference.
4151 (Specific): Remove broken link to Renesas RX processor.
4153 2017-02-10 Richard Biener <rguenther@suse.de>
4155 * toplev.c (process_options): Do not mention obsolete graphite
4156 options when printing sorry message about missing graphite support.
4157 Mention -floop-nest-optimize.
4159 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
4161 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
4162 (vtst_p16): Likewise.
4163 (vtstq_p8): Likewise.
4164 (vtstq_p16): Likewise.
4166 (vtstq_p64): Likewise.
4167 * config/arm/arm_neon.h (vgetq_lane_p64): New.
4168 (vset_lane_p64): New.
4169 (vsetq_lane_p64): New.
4171 2017-02-10 Jakub Jelinek <jakub@redhat.com>
4173 PR tree-optimization/79411
4174 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
4175 stmt operands are SSA_NAMEs used in abnormal phis.
4176 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
4179 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
4182 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
4185 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
4187 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
4189 2017-02-09 Jakub Jelinek <jakub@redhat.com>
4191 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
4195 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
4196 not arbitrary TREE_CONSTANT.
4199 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
4200 "omp declare target link" attribute unless is_global_var.
4201 * omp-offload.c (find_link_var_op): Likewise.
4203 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
4204 Chung-Lin Tang <cltang@codesourcery.com>
4206 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
4208 (gimplify_adjust_omp_clauses): Don't delete TILE.
4209 (gimplify_omp_for): Deal with TILE.
4210 * internal-fn.c (expand_GOACC_TILE): New function.
4211 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
4213 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
4214 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
4216 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
4217 avoid DIV for outermost collapse var.
4218 (expand_oacc_for): Insert tile element loop as needed. Adjust.
4219 Remove out of date comments, fix whitespace.
4220 * omp-general.c (omp_extract_for_data): Deal with tiling.
4221 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
4222 adjust OLF_DIM_BASE value.
4223 (struct omp_for_data): Add tiling field.
4224 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
4225 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
4226 for auto loops. Remove default auto determining, moved to
4227 oacc_loop_fixed_partitions.
4228 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
4229 stmts, add e_mask field.
4230 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
4231 (oacc_thread_numbers): Use oacc_dim_call.
4232 (oacc_xform_tile): New.
4233 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
4234 (finish_oacc_loop): Adjust for ifns vector.
4235 (oacc_loop_discover_walk): Append loop abstraction sites to list,
4236 add case for GOACC_TILE fns.
4237 (oacc_loop_xform_loop): Delete.
4238 (oacc_loop_process): Iterate over call list directly, and add
4239 handling for GOACC_TILE fns.
4240 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
4242 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
4243 vector partitioning to outer loops. Assign 2 partitions to loops
4244 when available. Add TILE handling.
4245 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
4246 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
4247 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
4248 * tree.c (omp_clause_num_ops): Adjust TILE ops.
4249 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
4251 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
4253 * configure.ac (ACX_BUGURL): Update.
4254 * configure: Regenerate.
4256 2017-02-09 Richard Biener <rguenther@suse.de>
4258 PR tree-optimization/69823
4259 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
4260 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
4262 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
4264 * config/arc/arc-c.def: Add __NPS400__ definition.
4265 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
4266 (TARGET_NPS400): Define.
4268 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
4270 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
4272 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
4274 (arc_cpu_types): Fill the arch_info field with a pointer into the
4275 arc_arch_types table.
4276 (arc_selected_cpu): Declare.
4277 * config/arc/arc.c (arc_selected_cpu): Make global.
4278 (arc_selected_arch): Delete.
4279 (arc_base_cpu): Delete.
4280 (arc_override_options): Remove references to deleted variables,
4281 update access to arch information.
4282 (ARC_OPT): Update access to arch information.
4283 (ARC_OPTX): Likewise.
4284 * config/arc/arc.h (arc_base_cpu): Remove declaration.
4285 (TARGET_ARC600): Update access to arch information.
4286 (TARGET_ARC601): Likewise.
4287 (TARGET_ARC700): Likewise.
4288 (TARGET_EM): Likewise.
4289 (TARGET_HS): Likewise.
4290 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
4293 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
4296 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
4297 condition/operands for integer GE/LE/GEU/LEU operations.
4299 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
4301 PR translation/79397
4302 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
4305 2017-02-08 Martin Jambor <mjambor@suse.cz>
4308 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
4309 whether allocation happened.
4310 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
4311 nothing was allocated.
4313 2017-02-08 Jakub Jelinek <jakub@redhat.com>
4315 PR tree-optimization/79408
4316 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
4317 constant, but SSA_NAME with a known integer range, use the minimum
4318 of that range instead of op1 to determine if modulo can be replaced
4319 with its first operand.
4321 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4323 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
4325 2017-02-08 Richard Biener <rguenther@suse.de>
4327 PR tree-optimization/71824
4328 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
4329 Check all loops contained in the merged region.
4331 2017-02-07 Andrew Pinski <apinski@cavium.com>
4333 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
4335 2017-02-07 Andrew Pinski <apinski@cavium.com>
4337 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
4338 (thunderxt88): Likewise.
4339 (thunderxt81): Disable LSE and change v8.1 to v8.
4340 (thunderxt83): Likewise.
4342 2017-02-07 Jakub Jelinek <jakub@redhat.com>
4343 Richard Biener <rguenther@suse.de>
4346 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
4347 type from int to size_t.
4348 * ira-costs.c (struct_costs_size): Change type from int to size_t.
4350 2017-02-07 Jakub Jelinek <jakub@redhat.com>
4352 PR rtl-optimization/79386
4353 * cprop.c (bypass_conditional_jumps): Initialize
4354 bypass_last_basic_block already before splitting bbs after
4355 unconditional traps...
4356 (bypass_conditional_jumps): ... rather than here.
4359 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
4360 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
4361 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
4362 fix -masm=intel patterns.
4364 2017-02-07 Richard Biener <rguenther@suse.de>
4366 PR tree-optimization/79256
4368 * builtins.c (get_object_alignment_2): Use min_align_of_type
4369 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
4370 and ADJUST_FIELD_ALIGN.
4372 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
4374 * doc/tm.texi: Regenerate.
4375 * stor-layout.c (layout_decl): Adjust.
4376 (update_alignment_for_field): Likewise.
4377 (place_field): Likewise.
4378 (min_align_of_type): Likewise.
4379 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
4380 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
4381 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
4382 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
4383 * config/frv/frv.c (frv_adjust_field_align): Likewise.
4384 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
4385 * config/i386/i386.c (x86_field_alignment): Likewise.
4386 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
4387 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
4388 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
4389 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
4390 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
4391 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
4395 2017-01-30 Richard Biener <rguenther@suse.de>
4397 PR tree-optimization/79256
4398 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
4399 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
4402 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
4404 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
4405 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
4406 builtins to SImode and emit a zero-extend, if necessary.
4408 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
4410 * docs/invoke.texi (RISC-V Options): Alphabetize.
4412 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
4414 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
4417 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
4419 * config/riscv/riscv.c: New file.
4420 * gcc/common/config/riscv/riscv-common.c: Likewise.
4421 * config.gcc: Likewise.
4422 * config/riscv/constraints.md: Likewise.
4423 * config/riscv/elf.h: Likewise.
4424 * config/riscv/generic.md: Likewise.
4425 * config/riscv/linux.h: Likewise.
4426 * config/riscv/multilib-generator: Likewise.
4427 * config/riscv/peephole.md: Likewise.
4428 * config/riscv/pic.md: Likewise.
4429 * config/riscv/predicates.md: Likewise.
4430 * config/riscv/riscv-builtins.c: Likewise.
4431 * config/riscv/riscv-c.c: Likewise.
4432 * config/riscv/riscv-ftypes.def: Likewise.
4433 * config/riscv/riscv-modes.def: Likewise.
4434 * config/riscv/riscv-opts.h: Likewise.
4435 * config/riscv/riscv-protos.h: Likewise.
4436 * config/riscv/riscv.h: Likewise.
4437 * config/riscv/riscv.md: Likewise.
4438 * config/riscv/riscv.opt: Likewise.
4439 * config/riscv/sync.md: Likewise.
4440 * config/riscv/t-elf-multilib: Likewise.
4441 * config/riscv/t-linux: Likewise.
4442 * config/riscv/t-linux-multilib: Likewise.
4443 * config/riscv/t-riscv: Likewise.
4444 * configure.ac: Likewise.
4445 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
4446 Waterman as RISC-V maintainers.
4447 * doc/install.texi: Add RISC-V entries.
4448 * doc/invoke.texi: Add RISC-V options section.
4449 * doc/md.texi: Add RISC-V constraints section.
4450 * configure: Regenerated.
4452 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4455 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
4456 false values to be constant vectors with all 0 or all 1 bits set.
4457 (vcondu<mode><mode>): Likewise.
4458 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
4460 (fpmask_comparison_operator): Update comment.
4461 (vecint_comparison_operator): New predicate.
4462 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
4463 vector conditionals when the true and false values are constant
4464 vectors with all 0 bits or all 1 bits set.
4466 2017-02-06 Martin Sebor <msebor@redhat.com>
4468 PR tree-optimization/79376
4469 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
4471 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
4473 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
4474 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
4475 to simplify split condition.
4477 2017-02-06 Jakub Jelinek <jakub@redhat.com>
4479 * omp-expand.c (oxpand_omp_atomic_fetch_op,
4480 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
4483 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
4485 PR rtl-optimization/68664
4486 * target.def (can_speculate_insn): New hook.
4487 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
4488 * doc/tm.texi: Regenerate.
4489 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
4490 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
4491 (rs6000_sched_can_speculate_insn): New function.
4493 2017-02-06 Jakub Jelinek <jakub@redhat.com>
4495 PR tree-optimization/79284
4496 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
4497 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
4498 vectorizable_mask_load_store, vectorizable_operation,
4499 vect_is_simple_cond, get_same_sized_vectype): Use it instead
4500 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
4501 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
4502 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
4503 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4504 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
4505 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
4506 is_gimple_assign (stmt). Replace another such test with
4507 is_gimple_assign (stmt).
4509 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
4512 * config/avr/avr.c (rtl-iter.h): Include it.
4513 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
4514 (avr_legitimate_combined_insn): ...and implementation.
4516 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
4518 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
4519 * config/s390/s390.c (s390_const_operand_ok)
4520 (s390_canonicalize_comparison, s390_extract_part)
4521 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
4522 (s390_contiguous_bitmask_p, s390_rtx_costs)
4523 (legitimize_pic_address): Likewise.
4524 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
4525 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
4526 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
4527 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
4528 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
4530 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
4532 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
4533 REGNO($0) == REGNO($1).
4535 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4537 * config/s390/linux.h(SIZE_TYPE): Add comment.
4539 2017-02-06 Julian Brown <julian@codesourcery.com>
4540 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4541 Virendra Pathak <virendra.pathak@broadcom.com>
4543 * config/aarch64/aarch64-cores.def: Change the scheduler
4545 * config/aarch64/aarch64.md: Include thunderx2t99.md.
4546 * config/aarch64/thunderx2t99.md: New file.
4548 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
4550 * doc/standards.texi (Go Language): Update link to language
4553 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
4555 * tree-eh.c (lower_resx): Sanitize profile.
4556 (cleanup_empty_eh_move_lp): Likewise.
4558 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
4561 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
4563 * cfgloopmanip.h (loop_version): Update prototype.
4564 * modulo-sched.c (sms_schedule): Update call of loop_version.
4565 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
4566 * tree-parloops.c (gen_parallel_loop): Likewise.
4567 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
4568 * tree-ssa-loop-split.c (split_loop): Likewise.
4569 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
4570 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
4572 2017-02-05 Martin Liska <mliska@suse.cz>
4575 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
4577 (print_operand_address): Initialize a struct to zero.
4579 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
4581 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
4582 garbage collector only in textual form.
4584 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
4586 * doc/extend.texi (x86 specific memory model extensions for
4587 transactional memory): Simplify a phrase.
4589 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
4592 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
4593 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
4594 (atomic_storedi_1): Likewise.
4596 2017-02-04 Jakub Jelinek <jakub@redhat.com>
4598 PR tree-optimization/79338
4599 * tree-parloops.c (gather_scalar_reductions): Don't call
4600 vect_analyze_loop_form for loop->inner before destroying loop's
4603 2017-02-03 Martin Sebor <msebor@redhat.com>
4605 PR tree-optimization/79327
4606 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
4607 when precision has resulted in leading zeros.
4608 (format_integer): Adjust the likely counter to assume an unknown
4609 argument that may be zero is non-zero.
4611 2017-02-03 Jason Merrill <jason@redhat.com>
4614 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
4615 avoid copying non-taken branch.
4617 2017-02-03 Jakub Jelinek <jakub@redhat.com>
4619 PR tree-optimization/79340
4620 * tree-vect-loop.c (vectorizable_reduction): Release
4621 vec_defs elements after safe_splicing them into other vectors.
4624 PR tree-optimization/79327
4625 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
4626 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
4628 (format_integer): Use wide_int_to_tree instead of build_int_cst
4629 + to_?hwi. If argmin is NULL, just set argmin and argmax to
4630 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
4631 of shortest and longest sequence.
4633 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
4635 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
4636 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
4638 2017-02-03 Walter Lee <walt@tilera.com>
4641 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
4642 after initial stackframe link reg save.
4643 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
4645 2017-02-03 Jakub Jelinek <jakub@redhat.com>
4648 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
4649 wu for stxssp alternative.
4651 2017-02-03 Martin Sebor <msebor@redhat.com>
4653 PR tree-optimization/79352
4654 * gimple-fold.c (get_range_strlen): Add argument.
4655 (get_range_strlen): Change return type to bool.
4656 (get_maxval_strlen): Pass in a dummy argument.
4657 * gimple-fold.h (get_range_strlen): Change return type to bool.
4658 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
4659 * tree.h (array_at_struct_end_p): Add argument.
4660 * tree.c (array_at_struct_end_p): Handle it.
4662 2017-02-03 Martin Liska <mliska@suse.cz>
4665 * multiple_target.c (create_dispatcher_calls): Redirect edge
4666 from a caller of a dispatcher.
4667 (expand_target_clones): Make the clones local.
4668 (ipa_target_clone): Do both target clones and resolvers.
4669 (ipa_dispatcher_calls): Remove the pass.
4670 (pass_dispatcher_calls::gate): Likewise.
4671 (make_pass_dispatcher_calls): Likewise.
4672 * passes.def (pass_target_clone): Put as very first IPA early
4675 2017-02-03 Martin Liska <mliska@suse.cz>
4677 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
4678 in case of a function with ifunc attribute.
4680 2017-02-03 Martin Liska <mliska@suse.cz>
4682 * cgraph.c (cgraph_node::dump): Dump function version info.
4683 * symtab.c (symtab_node::dump_base): Add missing new line.
4685 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
4687 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
4688 (ifcombine_ifandif): Use it.
4690 2017-02-03 Martin Liska <mliska@suse.cz>
4692 * doc/invoke.texi: Document default value for
4693 use-after-scope-direct-emission-threshold.
4695 2017-02-03 Martin Liska <mliska@suse.cz>
4697 PR tree-optimization/79339
4698 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
4699 (format_floating): Likewise.
4701 2017-02-03 Martin Liska <mliska@suse.cz>
4704 * ipa-prop.c (ipa_node_params_t::insert): Remove current
4706 (ipa_node_params_t::remove): Likewise.
4707 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
4708 initialization from removed ipa_node_params_t::insert.
4709 (ipa_node_params::~ipa_node_params): Move from removed
4710 ipa_node_params_t::release.
4711 * symbol-summary.h (symbol_summary::m_released): New member.
4712 Do not release a summary twice. Do not allow to call finalizer
4713 for types of a summary that live in GGC memory.
4715 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4717 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
4720 2017-02-02 Martin Sebor <msebor@redhat.com>
4723 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
4724 (format_string): Tighten up the range of output for non-constant
4725 strings and correct the expected range for wide non-constant strings.
4727 2017-02-02 Martin Sebor <msebor@redhat.com>
4729 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
4732 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
4734 (-fdump-tree-@var): Add to index and document how to come up
4735 with pass-specific option and dump file names.
4736 (-fdump-passes): Clarify where to look for output.
4738 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
4741 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
4742 statistics of the analyzed path; allow threading for speed when
4743 any of BBs along the path are optimized for speed.
4745 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
4748 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
4749 settings of the virtual registers.
4752 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
4754 * explow.c (get_dynamic_stack_size): Take known alignment of stack
4755 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
4758 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4760 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
4761 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
4763 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4765 * config/s390/s390.md: Add missing comments with the expanded
4767 * config/s390/vector.md: Likewise.
4768 * config/s390/vx-builtins.md: Likewise.
4770 2017-02-02 Jakub Jelinek <jakub@redhat.com>
4773 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
4774 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
4775 conditions on a single line.
4777 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4779 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
4780 __S390_VX__ to __VX__.
4782 2017-02-01 Andrew Pinski <apinski@cavium.com>
4784 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
4785 stmt_info to record_stmt_cost.
4786 (vect_get_known_peeling_cost): Pass stmt_info if known to
4788 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
4789 cpu_vector_cost field into
4790 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
4791 field into vec_int_stmt_cost and vec_fp_stmt_cost.
4792 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
4793 splitting of scalar_stmt_cost and vec_stmt_cost.
4794 (thunderx_vector_cost): Likewise.
4795 (cortexa57_vector_cost): LIkewise.
4796 (exynosm1_vector_cost): Likewise.
4797 (xgene1_vector_cost): Likewise.
4798 (thunderx2t99_vector_cost): Improve after the splitting of the two
4800 (aarch64_builtin_vectorization_cost): Update for the splitting of
4801 scalar_stmt_cost and vec_stmt_cost.
4803 2017-02-01 Torvald Riegel <triegel@redhat.com>
4804 Richard Henderson <rth@redhat.com>
4806 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
4807 conditional on existance of a fast atomic load.
4808 * optabs-query.c (can_atomic_load_p): New function.
4809 * optabs-query.h (can_atomic_load_p): Declare it.
4810 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
4811 no fast atomic load is available for the particular size of access.
4812 (expand_atomic_compare_and_swap): Likewise.
4813 (expand_atomic_load): Likewise.
4814 (expand_atomic_store): Likewise.
4815 (expand_atomic_fetch_op): Likewise.
4816 * testsuite/lib/target-supports.exp
4817 (check_effective_target_sync_int_128): Remove x86 because it provides
4818 no fast atomic load.
4819 (check_effective_target_sync_int_128_runtime): Likewise.
4821 2017-02-01 Richard Biener <rguenther@suse.de>
4823 * graphite.c: Include tree-vectorizer.h for find_loop_location.
4824 (graphite_transform_loops): Provide opt-info for optimized nests.
4825 * tree-parloop.c (parallelize_loops): Provide opt-info for
4828 2017-02-01 Richard Biener <rguenther@suse.de>
4831 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
4834 2017-02-01 Richard Biener <rguenther@suse.de>
4836 PR tree-optimization/71824
4837 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
4838 Verify the loops are valid in the merged SESE region.
4839 (scop_detection::can_represent_loop_1): Check analyzing the
4840 evolution of the number of iterations in the region succeeds.
4842 2017-01-31 Ian Lance Taylor <iant@golang.org>
4844 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
4845 REG_ARGS_SIZE note to 32-bit push insns and call insn.
4847 2017-01-31 David Malcolm <dmalcolm@redhat.com>
4849 PR preprocessor/79210
4850 * input.c (get_substring_ranges_for_loc): Replace line_width
4851 assertion with error-handling.
4853 2017-01-31 Richard Biener <rguenther@suse.de>
4855 PR tree-optimization/77318
4856 * graphite-sese-to-poly.c (extract_affine): Fix assert.
4857 (create_pw_aff_from_tree): Take loop parameter.
4858 (add_condition_to_pbb): Pass loop of the condition to
4859 create_pw_aff_from_tree.
4861 2017-01-31 Jakub Jelinek <jakub@redhat.com>
4863 * config/s390/s390.c (s390_asan_shadow_offset): New function.
4864 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
4866 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4870 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
4872 (convert_int_to_float128): Likewise.
4873 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
4874 (convert_int_to_float128): Likewise.
4875 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
4876 (UNSPEC_IEEE128_CONVERT): Likewise.
4877 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
4878 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
4879 Use local variables for IBM extended format.
4880 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
4881 (fix_trunc<mode>si2_fprs): Likewise.
4882 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
4883 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
4884 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
4885 to know that we can now have integers of all sizes in vector
4887 (fix<uns>_<mode>di2_hw): Likewise.
4888 (float<uns>_<mode>si2_hw): Likewise.
4889 (fix_<mode>si2_hw): Likewise.
4890 (fixuns_<mode>si2_hw): Likewise.
4891 (float<uns>_<mode>di2_hw): Likewise.
4892 (float_<mode>di2_hw): Likewise.
4893 (float_<mode>si2_hw): Likewise.
4894 (floatuns_<mode>di2_hw): Likewise.
4895 (floatuns_<mode>si2_hw): Likewise.
4896 (xscvqp<su>wz_<mode>): Delete, no longer used.
4897 (xscvqp<su>dz_<mode>): Likewise.
4898 (xscv<su>dqp_<mode>): Likewise.
4899 (ieee128_mfvsrd_64bit): Likewise.
4900 (ieee128_mfvsrd_32bit): Likewise.
4901 (ieee128_mfvsrwz): Likewise.
4902 (ieee128_mtvsrw): Likewise.
4903 (ieee128_mtvsrd_64bit): Likewise.
4904 (ieee128_mtvsrd_32bit): Likewise.
4906 2017-01-31 Martin Liska <mliska@suse.cz>
4909 * ipa-prop.c (ipa_free_all_node_params): Call release method
4910 instead of ~sumbol_summary to not to trigger double times
4913 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
4915 PR tree-optimization/71691
4916 * bitmap.h (class auto_bitmap): New.
4917 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
4918 is_maybe_undefined instead of ssa_undefined_value_p.
4920 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4922 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
4923 __S390_ARCH_LEVEL__ to __ARCH__.
4925 2017-01-31 Jakub Jelinek <jakub@redhat.com>
4927 PR tree-optimization/79267
4928 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
4929 if should_remove_lhs_p is true.
4931 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
4934 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
4935 (add_alignment_attribute): New.
4936 (base_type_die): Add alignment attribute.
4937 (subrange_type_die): Likewise.
4938 (modified_type_die): Likewise.
4939 (gen_array_type_die): Likewise.
4940 (gen_descr_array_type_die: Likewise.
4941 (gen_enumeration_type_die): Likewise.
4942 (gen_subprogram_die): Likewise.
4943 (gen_variable_die): Likewise.
4944 (gen_field_die): Likewise.
4945 (gen_ptr_to_mbr_type_die): Likewise.
4946 (gen_struct_or_union_type_die): Likewise.
4947 (gen_subroutine_type_die): Likewise.
4948 (gen_typedef_die): Likewise.
4949 (base_type_cmp): Compare alignment attribute.
4951 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4954 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
4955 (setb_unsigned) New pattern for setb with CCUNS.
4956 * config/rs6000/rs6000.c (expand_block_compare): Use a different
4957 subfc./subfe sequence to avoid overflow problems. Generate a
4958 shorter sequence with cmpld/setb for power9.
4959 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
4960 for generating subfc. instruction.
4961 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
4962 now uses this instruction.
4964 2017-01-30 Ian Lance Taylor <iant@google.com>
4967 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
4968 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
4970 2017-01-30 Martin Sebor <msebor@redhat.com>
4972 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
4973 Move constant to the right of a relational operator.
4974 (get_mpfr_format_length, format_character, format_string): Ditto.
4975 (should_warn_p, maybe_warn): Same.
4977 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
4979 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
4982 * asan.c (get_translation_unit_decl): Remove function.
4983 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
4985 2017-01-30 Martin Liska <mliska@suse.cz>
4987 PR gcov-profile/79259
4988 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
4991 2017-01-30 Martin Liska <mliska@suse.cz>
4994 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
4995 Initialize variables with NULL value.
4997 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
5000 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
5002 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
5004 2017-01-30 Richard Biener <rguenther@suse.de>
5006 PR tree-optimization/79276
5007 * tree-vrp.c (process_assert_insertions): Properly adjust common
5008 when removing a duplicate.
5010 * gcc.dg/torture/pr79276.c: New testcase.
5012 2017-01-30 Richard Biener <rguenther@suse.de>
5014 PR tree-optimization/79256
5015 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
5016 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
5018 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
5020 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
5023 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
5024 ("*r<noxa>sbg_<mode>_sll_bitmask")
5025 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
5026 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
5027 Use contiguous_bitmask_nowrap_operand.
5029 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5032 * config/rs6000/altivec.h (vec_xl): Revise #define.
5033 (vec_xst): Likewise.
5035 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
5037 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
5039 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
5041 PR rtl-optimization/79194
5042 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
5043 traps before call to bypass_conditional_jumps.
5045 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
5047 PR tree-optimization/71374
5048 * lra-constraints.c (check_conflict_input_operands): New.
5049 (match_reload): Use it.
5051 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
5054 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
5055 account to calculate conflict_set.
5057 2017-01-27 Bin Cheng <bin.cheng@arm.com>
5059 PR rtl-optimization/78559
5060 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
5061 other_insn in combine.
5063 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
5065 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
5066 uint16_type_node for BT_UINT16.
5068 2017-01-27 David Malcolm <dmalcolm@redhat.com>
5070 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
5071 "RTL Tests" to menu.
5072 (GIMPLE Tests): New node.
5073 (RTL Tests): New node.
5075 2017-01-27 Richard Biener <rguenther@suse.de>
5077 PR tree-optimization/79245
5078 * tree-loop-distribution.c (distribute_loop): Apply cost
5079 modeling also to detected patterns.
5081 2017-01-27 Richard Biener <rguenther@suse.de>
5083 PR tree-optimization/71433
5084 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
5085 (compare_assert_loc): New function.
5086 (process_assert_insertions): Sort and optimize assert locations
5087 to remove duplicates and push down identical assertions on
5088 edges to their destination block.
5090 2017-01-27 Richard Biener <rguenther@suse.de>
5092 PR tree-optimization/79244
5093 * tree-vrp.c (remove_range_assertions): Forcefully propagate
5094 out SSA names even if abnormal.
5096 2017-01-27 Jakub Jelinek <jakub@redhat.com>
5098 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
5099 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
5100 instead of MPFR_RNDN.
5102 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
5105 * arm.c (arm_option_override): Don't call build_target_option_node
5106 until after doing all option overrides.
5107 (arm_valid_target_attribute_tree): Likewise.
5109 2017-01-27 Martin Liska <mliska@suse.cz>
5111 * doc/invoke.texi (-fprofile-arcs): Document profiling support
5112 for {cd}tors and C++ {cd}tors.
5114 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
5116 * config/s390/s390.md ("*setmem_long_and")
5117 ("*setmem_long_and_31z"): Use zero_extend instead of and.
5119 2017-01-26 Martin Sebor <msebor@redhat.com>
5121 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
5124 2017-01-26 Martin Sebor <msebor@redhat.com>
5126 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
5127 HAVE_DFmode before using XFmode or DFmode.
5128 (parse_directive): Avoid using the z length modifier to avoid
5129 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
5132 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
5133 to accept adjustment as an array.
5134 (get_int_range): New function.
5135 (struct directive): Make width and prec arrays.
5136 (directive::set_width, directive::set_precision): Call get_int_range.
5137 (format_integer, format_floating): Handle width and precision ranges.
5138 (format_string, parse_directive): Same.
5140 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5143 * dwarf2out.c (generate_skeleton_bottom_up): For children with
5144 comdat_type_p set, just clone them, but keep the children in the
5148 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
5149 which have direct callers with -fvar-tracking-assignments enabled
5151 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
5152 inside of type units.
5154 2017-01-26 Martin Sebor <msebor@redhat.com>
5157 * gimple-ssa-sprintf.c (struct result_range): Add likely and
5159 (struct format_result): Replace number_chars, number_chars_min,
5160 and number_chars_max with a single member of struct result_range.
5162 (format_result::operator+=): Adjust.
5163 (struct fmtresult): Remove bounded. Handle likely and unlikely
5165 (fmtresult::adjust_for_width_or_precision): New function.
5166 (fmtresult:type_max_digits): New function.
5167 (bytes_remaining): Handle likely and unlikely counters.
5168 (min_bytes_remaining): Remove.
5169 (format_percent): Simplify.
5170 (format_integer, format_floating): Set likely and unlikely counters.
5171 (get_string_length, format_character, format_string): Same.
5172 (format_plain, should_warn_p): New function.
5173 (maybe_warn): Call should_warn_p. Update diagnostic messages
5174 and handle those for all directives, including plain strings.
5175 (format_directive): Handle likely and unlikely counters.
5176 Remove unnecessary quoting from diagnostics. Add an informational
5178 (add_bytes): Remove.
5179 (pass_sprintf_length::compute_format_length): Simplify.
5180 (try_substitute_return_value): Handle likely and unlikely counters.
5182 2017-01-26 Carl Love <cel@us.ibm.com>
5184 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
5185 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
5187 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
5190 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
5191 endianess for subregs into account.
5192 * lra-constraints.c (lra_constraints): Do risky transformations
5193 always on the first iteration.
5194 * lra-lives.c (check_pseudos_live_through_calls): Add arg
5195 last_call_used_reg_set.
5196 (process_bb_lives): Define and use last_call_used_reg_set.
5197 * lra.c (lra): Always continue after lra_constraints on the first
5200 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
5202 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
5204 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
5206 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5208 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
5209 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
5210 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
5211 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
5212 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
5213 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
5214 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
5215 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
5216 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
5218 2017-01-26 Marek Polacek <polacek@redhat.com>
5221 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
5222 for the third operand.
5224 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5227 * omp-low.c (struct omp_context): Add simt_stmt field.
5228 (scan_omp_for): Return omp_context *.
5229 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
5230 context to the _simt_ SIMD stmt.
5231 (lower_omp_for): For combined SIMD with sibling _simt_
5232 SIMD, make sure to use the same decls in _looptemp_
5233 clauses as in the sibling.
5235 2017-01-26 David Sherwood <david.sherwood@arm.com>
5238 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
5241 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5244 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
5245 emit fld b; fld a; if possible.
5247 * brig-builtins.def: Update copyright years.
5248 * config/arm/arm_acle_builtins.def: Update copyright years.
5250 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5253 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
5254 constraint instead of o for the stxsd instruction.
5256 2017-01-25 Carl Love <cel@us.ibm.com>
5258 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
5259 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
5261 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
5263 * doc/invoke.texi (C++ Dialect Options): Fix typo.
5265 2017-01-25 Richard Biener <rguenther@suse.de>
5267 PR tree-optimization/69264
5268 * target.def (vector_alignment_reachable): Improve documentation.
5269 * doc/tm.texi: Regenerate.
5270 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
5272 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
5273 earlier changes with respect to TYPE_USER_ALIGN.
5274 (vector_alignment_reachable_p): Likewise. Improve dumping.
5276 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5279 * config/arm/arm.md (xordi3): Force constant operand into a register
5282 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5284 * doc/invoke.texi (-fstore-merging): Correct default optimization
5285 levels at which it is enabled.
5286 (-O): Move -fstore-merging from list to...
5289 2017-01-25 Richard Biener <rguenther@suse.de>
5292 * omp-expand.c: Include debug.h.
5293 (expand_omp_taskreg): Make sure to generate early debug before
5294 outlining anything from a function.
5295 (expand_omp_target): Likewise.
5296 (grid_expand_target_grid_body): Likewise.
5298 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
5301 * asan.c (get_translation_unit_decl): New function.
5302 (asan_add_global): Extract modules file name from globals
5303 TRANSLATION_UNIT_DECL name.
5305 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
5308 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
5309 for long calls with APCS frame and VFP.
5311 2017-01-24 David Malcolm <dmalcolm@redhat.com>
5313 * cfg.c (original_copy_tables_initialized_p): New function.
5314 * cfg.h (original_copy_tables_initialized_p): New decl.
5315 * cfgrtl.c (relink_block_chain): Guard the call to
5316 free_original_copy_tables with a call to
5317 original_copy_tables_initialized_p.
5318 * cgraph.h (symtab_node::native_rtl_p): New decl.
5319 * cgraphunit.c (symtab_node::native_rtl_p): New function.
5320 (symtab_node::needed_p): Don't assert for early assembly output
5321 for __RTL functions.
5322 (cgraph_node::finalize_function): Set "force_output" for __RTL
5324 (cgraph_node::analyze): Bail out early for __RTL functions.
5325 (analyze_functions): Update assertion to support __RTL functions.
5326 (cgraph_node::expand): Bail out early for __RTL functions.
5327 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
5329 * function.h (struct function): Update comment for field
5331 * gimple-expr.c: Include "tree-pass.h".
5332 (gimple_has_body_p): Return false for __RTL functions.
5333 * Makefile.in (OBJS): Add run-rtl-passes.o.
5334 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
5336 (gcc::pass_manager::get_clean_slate): New accessor.
5337 * passes.c: Include "insn-addr.h".
5338 (should_skip_pass_p): Add logging. Update logic for running
5339 "expand" to be compatible with both __GIMPLE and __RTL. Guard
5340 property-provider override so it is only done for gimple passes.
5342 (skip_pass): New function.
5343 (execute_one_pass): Call skip_pass when skipping passes.
5344 * read-md.c (md_reader::read_char): Support filtering
5345 the input to a subset of line numbers.
5346 (md_reader::md_reader): Initialize fields
5347 m_first_line and m_last_line.
5348 (md_reader::read_file_fragment): New function.
5349 * read-md.h (md_reader::read_file_fragment): New decl.
5350 (md_reader::m_first_line): New field.
5351 (md_reader::m_last_line): New field.
5352 * read-rtl-function.c (function_reader::create_function): Only
5353 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
5354 curr_properties. Set DECL_INITIAL to a dummy block.
5355 (read_rtl_function_body_from_file_range): New function.
5356 * read-rtl-function.h (read_rtl_function_body_from_file_range):
5358 * run-rtl-passes.c: New file.
5359 * run-rtl-passes.h: New file.
5361 2017-01-24 Jeff Law <law@redhat.com>
5363 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
5366 2017-01-24 Bin Cheng <bin.cheng@arm.com>
5368 PR tree-optimization/79159
5369 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
5370 (record_nonwrapping_iv): Improve boundary using above function if no
5371 value range information.
5373 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
5374 Martin Jambor <mjambor@suse.cz>
5376 * brig-builtins.def: New file.
5377 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
5378 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
5379 (DEF_HSAIL_SAT_BUILTIN): Likewise.
5380 (DEF_HSAIL_INTR_BUILTIN): Likewise.
5381 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
5382 * builtin-types.def (BT_INT8): New.
5383 (BT_INT16): Likewise.
5384 (BT_UINT8): Likewise.
5385 (BT_UINT16): Likewise.
5386 (BT_FN_ULONG): Likewise.
5387 (BT_FN_UINT_INT): Likewise.
5388 (BT_FN_UINT_ULONG): Likewise.
5389 (BT_FN_UINT_LONG): Likewise.
5390 (BT_FN_UINT_PTR): Likewise.
5391 (BT_FN_ULONG_PTR): Likewise.
5392 (BT_FN_INT8_FLOAT): Likewise.
5393 (BT_FN_INT16_FLOAT): Likewise.
5394 (BT_FN_UINT32_FLOAT): Likewise.
5395 (BT_FN_UINT16_FLOAT): Likewise.
5396 (BT_FN_UINT8_FLOAT): Likewise.
5397 (BT_FN_UINT64_FLOAT): Likewise.
5398 (BT_FN_UINT16_UINT32): Likewise.
5399 (BT_FN_UINT32_UINT16): Likewise.
5400 (BT_FN_UINT16_UINT16_UINT16): Likewise.
5401 (BT_FN_INT_PTR_INT): Likewise.
5402 (BT_FN_UINT_PTR_UINT): Likewise.
5403 (BT_FN_LONG_PTR_LONG): Likewise.
5404 (BT_FN_ULONG_PTR_ULONG): Likewise.
5405 (BT_FN_VOID_UINT64_UINT64): Likewise.
5406 (BT_FN_UINT8_UINT8_UINT8): Likewise.
5407 (BT_FN_INT8_INT8_INT8): Likewise.
5408 (BT_FN_INT16_INT16_INT16): Likewise.
5409 (BT_FN_INT_INT_INT): Likewise.
5410 (BT_FN_UINT_FLOAT_UINT): Likewise.
5411 (BT_FN_FLOAT_UINT_UINT): Likewise.
5412 (BT_FN_ULONG_UINT_UINT): Likewise.
5413 (BT_FN_ULONG_UINT_PTR): Likewise.
5414 (BT_FN_ULONG_ULONG_ULONG): Likewise.
5415 (BT_FN_UINT_UINT_UINT): Likewise.
5416 (BT_FN_VOID_UINT_PTR): Likewise.
5417 (BT_FN_UINT_UINT_PTR: Likewise.
5418 (BT_FN_UINT32_UINT64_PTR): Likewise.
5419 (BT_FN_INT_INT_UINT_UINT): Likewise.
5420 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
5421 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
5422 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
5423 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
5424 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
5425 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
5426 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
5427 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
5428 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
5429 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
5430 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
5431 * doc/frontends.texi: List BRIG FE.
5432 * doc/install.texi (Testing): Add BRIG tesring requirements.
5433 * doc/invoke.texi (Overall Options): Mention BRIG.
5434 * doc/standards.texi (Standards): Doucment BRIG HSA version.
5436 2017-01-24 Richard Biener <rguenther@suse.de>
5438 PR translation/79208
5439 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
5441 2017-01-24 Martin Jambor <mjambor@suse.cz>
5444 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
5445 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
5448 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
5451 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
5452 casts from signed to unsigned really don't have a range.
5454 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
5456 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
5457 GMP_RNDx for compatiblity.
5459 2017-01-24 Martin Liska <mliska@suse.cz>
5462 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
5463 that would prevent us to call alloca with -1 as argument.
5465 2017-01-24 Jakub Jelinek <jakub@redhat.com>
5467 * dwarf2out.c (output_compilation_unit_header, output_file_names):
5468 Avoid -Wformat-security warning.
5470 2017-01-23 Andrew Pinski <apinski@cavium.com>
5472 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
5475 2017-01-23 Martin Sebor <msebor@redhat.com>
5478 * gimple-ssa-sprintf.c (warn_level): New global.
5479 (format_integer): Use it here and throughout the rest of the file.
5480 Use the same switch to compute sign as base.
5481 (maybe_warn): New function.
5482 (format_directive): Factor out warnings into maybe_warn.
5483 Add debugging output. Use warn_level.
5484 (add_bytes): Use warn_level.
5485 (pass_sprintf_length::compute_format_length): Add debugging output.
5486 (try_substitute_return_value): Same.
5487 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
5490 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
5491 (struct fmtresult, format_integer, format_floating): Adjust.
5492 (fmtresult::fmtresult): Set max correctly in two argument ctor.
5493 (get_string_length, format_string,format_directive): Same.
5494 (pass_sprintf_length::compute_format_length): Same.
5495 (try_substitute_return_value): Simplify slightly.
5498 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
5499 (fmtresult::operator+=): Outlined.
5500 (struct fmtresult): Add ctors.
5501 (struct conversion_spec): Rename...
5502 (struct directive): ...to this. Add and remove data members.
5503 (directive::set_width, directive::set_precision): New functions.
5504 (format_percent): Use fmtresult ctor.
5505 (get_width_and_precision): Remove.
5506 (format_integer): Make naming changes. Avoid computing width and
5508 (format_floating): Same. Adjust indentation.
5509 (format_character, format_none): New functions.
5510 (format_string): Moved character handling to format_character.
5511 (format_directive): Remove arguments, change return type.
5512 (parse_directive): New function.
5513 (pass_sprintf_length::compute_format_length): Move directive
5514 parsing to parse_directive.
5516 2017-01-23 Jakub Jelinek <jakub@redhat.com>
5518 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
5519 (assign_assembler_name_if_needed): ... this.
5520 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
5521 (assign_assembler_name_if_needed): ... this.
5522 (free_lang_data_in_cgraph): Adjust callers.
5523 * cgraphunit.c (cgraph_node::analyze): Likewise.
5524 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
5527 2017-01-23 Richard Biener <rguenther@suse.de>
5529 PR tree-optimization/79088
5530 PR tree-optimization/79188
5531 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
5532 resetting loop bounds after last path deletion. Reset loop
5533 bounds of the target loop, make code match the comments.
5534 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
5535 Make sure loops need no fixups.
5537 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
5539 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
5540 exponent support with double type for first argument.
5541 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
5542 type returned by __builtin_vec_extract_sig,
5543 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
5544 functions from "vector int" to "vector unsigned int" or from
5545 "vector long long int" to "vector unsigned long long int".
5546 Changed type returned by __builtin_vec_extract_exp,
5547 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
5548 functions from "vector int" to "vector unsigned int" or from
5549 "vector long long int" to "vector unsigned long long int".
5550 Changed return type of __builtin_vec_test_data_class,
5551 __builtin_vec_test_data_class_sp, and
5552 __builtin_vec_test_data_class_dp from "vector int" to
5553 "vector bool int" or from "vector long long int" to "vector bool
5554 long long int" and changed second argument type from "unsigned
5555 int" to "int". Added new overloaded function forms "vector float
5556 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
5557 "vector float __builtin_vec_insert_exp_sp (vector float, vector
5558 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
5559 double, vector unsigned long long int)" and "vector double
5560 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
5561 long int)". Changed return type of
5562 __builtin_scalar_test_data_class and
5563 __builtin_scalar_test_data_class_sp and
5564 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
5565 int" and changed second argument from "unsigned int" to "int".
5566 Changed type returned by __builtin_scalar_test_neg,
5567 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
5568 from "int" to "bool int". Added new overloaded function form
5569 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
5570 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
5571 exponent double-precision with floating point first argument.
5572 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
5573 documentation of scalar_test_data_class, scalar_test_neg,
5574 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
5575 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
5576 vec_test_data_class built-in functions to reflect refinements in
5577 their type signatures.
5579 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
5581 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
5583 (aarch64_elf_asm_destructor): Likewise.
5585 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
5587 PR rtl-optimization/78634
5588 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
5589 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
5590 * ifcvt.c (noce_try_cmove): Add missing cost check.
5592 PR rtl-optimization/71724
5593 * combine.c (if_then_else_cond): Look for situations where it is
5594 beneficial to undo the work of one of the recursive calls.
5596 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5598 PR tree-optimization/70754
5599 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5600 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5601 combined stmt before it if not NULL.
5602 (combine_chains): Process refs reversely and compute dominance point
5605 2017-01-23 Martin Liska <mliska@suse.cz>
5607 PR tree-optimization/79196
5608 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
5609 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
5611 (strlen_optimize_stmt): Call the renamed function.
5613 2017-01-23 Michael Matz <matz@suse.de>
5615 PR tree-optimization/78384
5616 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
5618 2017-01-23 Richard Biener <rguenther@suse.de>
5620 PR tree-optimization/79186
5621 * tree-vrp.c (register_new_assert_for): Make sure we've seen
5622 both incoming edges before moving an assert.
5624 2017-01-23 Martin Jambor <mjambor@suse.cz>
5626 * ipa-prop.c (load_from_param_1): Removed.
5627 (load_from_unmodified_param): Bits from load_from_param_1 put back
5629 (load_from_param): Removed.
5630 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
5631 with stmt. Reverted back to use of load_from_unmodified_param.
5633 2017-01-23 Martin Jambor <mjambor@suse.cz>
5636 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
5637 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
5638 field a pointer to garbage collected vector, mark lattices and
5639 ipcp_orig_node with GTY((skip)).
5640 (ipa_get_param_count): Adjust to descriptors being a pointer.
5641 (ipa_get_param): Likewise.
5642 (ipa_get_type): Likewise.
5643 (ipa_get_param_move_cost): Likewise.
5644 (ipa_set_param_used): Likewise.
5645 (ipa_get_controlled_uses): Likewise.
5646 (ipa_set_controlled_uses): Likewise.
5647 (ipa_is_param_used): Likewise.
5648 (ipa_node_params_t): Move into garbage collector. New methods insert
5650 (ipa_node_params_sum): Annotate wth GTY(()).
5651 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
5653 (ipa_load_from_parm_agg): Adjust declaration.
5654 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
5655 * ipa-profile.c (ipa_profile): Likewise.
5656 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
5657 (ipa_populate_param_decls): Make descriptors parameter garbage
5659 (ipa_dump_param): Adjust to descriptors being a pointer.
5660 (ipa_alloc_node_params): Likewise.
5661 (ipa_initialize_node_params): Likewise.
5662 (load_from_param_1): Make descriptors parameter garbage collected.
5663 (load_from_unmodified_param): Likewise.
5664 (load_from_param): Likewise.
5665 (ipa_load_from_parm_agg): Likewise.
5666 (ipa_node_params::~ipa_node_params): Removed.
5667 (ipa_free_all_node_params): Remove call to delete operator.
5668 (ipa_node_params_t::insert): New.
5669 (ipa_node_params_t::remove): Likewise.
5670 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
5671 copy known_csts and known_contexts vectors.
5672 (ipa_read_node_info): Adjust to descriptors being a pointer.
5673 (ipcp_modif_dom_walker): Make m_descriptors field garbage
5675 (ipcp_transform_function): Make descriptors variable garbage
5678 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
5680 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
5681 * config/i386/avx512dqintrin.h: Ditto.
5682 * config/i386/avx512fintrin.h: Ditto.
5683 * gcc/config/i386/i386.c: Handle new builtins.
5684 * config/i386/i386-builtin.def: Add new builtins.
5685 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
5686 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
5688 2017-01-23 Jakub Jelinek <jakub@redhat.com>
5689 Martin Liska <mliska@suse.cz>
5691 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
5692 * asan.c (asan_expand_poison_ifn): Support stores and use
5693 appropriate ASAN report function.
5694 * internal-fn.c (expand_ASAN_POISON_USE): New function.
5695 * internal-fn.def (ASAN_POISON_USE): Declare.
5696 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
5697 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
5698 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
5699 ASAN_POISON calls w/o LHS.
5700 * tree-ssa.c (execute_update_addresses_taken): Create clobber
5701 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
5702 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
5703 * gimplify.c (asan_poison_variables): Add attribute
5704 use_after_scope_memory to variables that really needs to live
5706 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
5707 having the attribute.
5709 2017-01-23 Martin Liska <mliska@suse.cz>
5711 * asan.c (create_asan_shadow_var): New function.
5712 (asan_expand_poison_ifn): Likewise.
5713 * asan.h (asan_expand_poison_ifn): New declaration.
5714 * internal-fn.c (expand_ASAN_POISON): Likewise.
5715 * internal-fn.def (ASAN_POISON): New builtin.
5716 * sanopt.c (pass_sanopt::execute): Expand
5717 asan_expand_poison_ifn.
5718 * tree-inline.c (copy_decl_for_dup_finish): Make function
5720 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
5721 * tree-ssa.c (is_asan_mark_p): New function.
5722 (execute_update_addresses_taken): Rewrite local variables
5723 (identified just by use-after-scope as addressable) into SSA.
5725 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
5727 * doc/install.texi (Specific): opensource.apple.com uses https
5728 now. Remove trailing slash.
5730 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
5732 * README.Portability: Remove note on an Irix compatibility issue.
5734 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
5736 * gcov.c (INCLUDE_ALGORITHM): Define.
5737 (INCLUDE_VECTOR): Define.
5738 No longer include <vector> and <algorithm> directly.
5740 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
5742 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
5744 * doc/invoke.texi (Code Gen Options): Ditto.
5746 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
5749 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
5751 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
5753 rtl-optimization/79125
5754 * cprop.c (local_cprop_pass): Handle cases where we make an
5757 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
5761 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
5762 read from, for big endian.
5764 2017-01-20 Jiong Wang <jiong.wang@arm.com>
5766 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
5767 register pauth builtins for LP64 only.
5769 2017-01-20 Marek Polacek <polacek@redhat.com>
5772 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
5775 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
5777 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
5779 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
5780 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
5781 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
5783 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5786 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
5787 in big-endian mode when they are not a single duplicated value.
5789 2017-01-20 Richard Biener <rguenther@suse.de>
5791 * BASE-VER: Bump to 7.0.1.
5793 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
5795 * omp-low.c (omplow_simd_context): New struct. Use it...
5796 (lower_rec_simd_input_clauses): ...here and...
5797 (lower_rec_input_clauses): ...here to hold common data. Adjust all
5798 references to idx, lane, max_vf, is_simt.
5800 2017-01-20 Graham Markall <graham.markall@embecosm.com>
5802 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
5805 2017-01-20 Martin Jambor <mjambor@suse.cz>
5807 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
5808 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
5810 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
5811 (GTFILES): Rename hsa.c to hsa-common.c.
5812 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
5813 * hsa-dump.c: Likewise.
5814 * hsa-gen.c: Likewise.
5815 * hsa-regalloc.c: Likewise.
5816 * ipa-hsa.c: Likewise.
5817 * omp-expand.c: Likewise.
5818 * omp-low.c: Likewise.
5819 * toplev.c: Likewise.
5821 2017-01-20 Marek Polacek <polacek@redhat.com>
5824 * doc/invoke.texi: Document -Wduplicated-branches.
5825 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
5826 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
5827 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
5828 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
5829 return 0 only when not OEP_LEXICOGRAPHIC.
5830 (fold_build_cleanup_point_expr): Use the expression
5831 location when building CLEANUP_POINT_EXPR.
5832 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
5833 * tree.c (add_expr): Handle error_mark_node.
5835 2017-01-20 Martin Liska <mliska@suse.cz>
5838 * tree-profile.c (init_ic_make_global_vars): Do not call
5840 (gimple_init_gcov_profiler): Likewise.
5842 2017-01-20 Martin Liska <mliska@suse.cz>
5845 * cgraph.h (maybe_create_reference): Remove argument and
5847 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
5849 * ipa-cp.c (create_specialized_node): Likewise.
5850 * symtab.c (symtab_node::maybe_create_reference): Handle
5851 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
5853 2017-01-20 Martin Liska <mliska@suse.cz>
5855 * read-rtl-function.c (function_reader::create_function): Use
5856 build_decl instread of build_decl_stat.
5858 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
5860 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
5861 * config/i386/avx512dqintrin.h: Ditto.
5862 * config/i386/avx512fintrin.h: Ditto.
5863 * config/i386/i386-builtin-types.def: Add new types.
5864 * gcc/config/i386/i386.c: Handle new types.
5865 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
5866 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
5867 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
5868 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
5869 (__builtin_ia32_kshiftridi): New.
5870 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
5872 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
5876 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
5877 define to rs6000_init_stack_protect_guard.
5878 (rs6000_init_stack_protect_guard): New function.
5880 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
5881 Yunqiang Su <yunqiang.su@imgtec.com>
5883 * config.gcc (supported_defaults): Add madd4.
5884 (with_madd4): Add validation.
5885 (all_defaults): Add madd4.
5886 * config/mips/mips.opt (mmadd4): New option.
5887 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
5889 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
5891 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
5892 (ISA_HAS_FUSED_MADD4): Likewise.
5893 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
5894 * gcc/doc/install.texi (--with-madd4): Document the new option.
5896 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5898 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
5899 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
5900 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
5901 (aarch64_init_pauth_hint_builtins): New.
5902 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
5903 (aarch64_expand_builtin): Expand new builtins.
5905 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5907 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
5908 * combine-stack-adj.c (no_unhandled_cfa): Handle
5909 REG_CFA_TOGGLE_RA_MANGLE.
5910 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
5911 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
5912 info for return address signing.
5913 (aarch64_expand_epilogue): Likewise.
5915 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5917 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
5918 * config/aarch64/aarch64-protos.h
5919 (aarch64_return_address_signing_enabled): New declaration.
5920 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
5922 (aarch64_expand_prologue): Sign return address before it's pushed onto
5924 (aarch64_expand_epilogue): Authenticate return address fetched from
5926 (aarch64_override_options): Sanity check for ILP32 and ISA level.
5927 (aarch64_attributes): New function attributes for "sign-return-address".
5928 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
5929 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
5930 ("*do_return"): Generate combined instructions according to key index.
5931 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
5932 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
5934 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
5935 * config/aarch64/aarch64.opt (msign-return-address=): New.
5936 * doc/extend.texi (AArch64 Function Attributes): Documents
5937 "sign-return-address=".
5938 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
5940 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
5942 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
5943 overall option summary.
5945 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5947 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
5948 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
5949 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
5950 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
5952 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
5954 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
5955 -mpower9-minmax by default for -mcpu=power9.
5956 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
5957 128-bit floating point.
5959 2017-01-20 Alan Modra <amodra@gmail.com>
5961 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
5962 optimizing for size.
5964 2017-01-20 Alan Modra <amodra@gmail.com>
5967 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
5968 for strcmp and strncmp from corresponding builtin decl.
5970 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
5972 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
5973 instead of i386/rtems-64.h.
5974 * config/i386/rtems-64.h: Remove.
5976 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
5980 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
5982 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
5984 2017-01-19 Tamar Christina <tamar.christina@arm.com>
5986 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
5987 Change int to HOST_WIDE_INT.
5988 * config/aarch64/aarch64-protos.h
5989 (aarch64_simd_gen_const_vector_dup): Likewise.
5990 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
5992 2017-01-19 David Malcolm <dmalcolm@redhat.com>
5994 * langhooks-def.h (lhd_type_for_size): New decl.
5995 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
5996 * langhooks.c (lhd_type_for_size): New function, taken from
5999 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
6001 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
6002 define_bypass for CR latency.
6003 (power9-cracked-alu): Update bypass latency and remove power9-branch.
6004 (power9-alu2): Add define_bypass for CR latency.
6006 (power9-mul): Update insn latency.
6007 (power9-mul-compare): Update insn latency, bypass latency and remove
6010 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6012 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
6014 * config/aarch64/aarch64.md
6015 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
6016 aarch64_nopcrelative_literal_loads.
6017 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
6019 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
6021 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
6023 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
6025 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
6028 * config.gcc (supported_defaults): Add lxc1-sxc1.
6029 (with_lxc1_sxc1): Add validation.
6030 (all_defaults): Add lxc1-sxc1.
6031 * config/mips/mips.opt (mlxc1-sxc1): New option.
6032 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
6034 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
6035 __mips_no_lxc1_sxc1.
6036 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
6037 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
6038 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
6040 2017-01-19 Richard Biener <rguenther@suse.de>
6042 PR tree-optimization/72488
6043 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
6044 sure to restore SSA info.
6045 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
6047 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
6049 PR rtl-optimization/79121
6050 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
6051 of the inner type when shifting an extended value.
6053 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
6056 * symtab.c (symtab_node::equal_address_to): Fix comparing of
6057 interposable aliases.
6059 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
6062 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
6063 Use the evmergelohi instruction.
6064 (mov_si<mode>_e500_subreg4_2_le): Likewise.
6065 (mov_sitf_e500_subreg8_2_be): Likewise.
6066 (mov_sitf_e500_subreg12_2_le): Likewise.
6067 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
6068 (mov_si<mode>_e500_subreg4_2_be): Likewise.
6069 (mov_sitf_e500_subreg8_2_le): Likewise.
6070 (mov_sitf_e500_subreg12_2_be): Likewise.
6072 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6074 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
6075 attribute from vecsimple to vecperm.
6076 (altivec_vbpermq2): Likewise.
6078 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6081 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
6083 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6084 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
6085 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
6086 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
6087 case where N arg is SIZE_MAX.
6088 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
6089 (cmpstrsi): Add pattern.
6091 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
6093 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6094 __builtin_vec_revb builtins.
6095 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
6096 built-in functions to support generation of the ISA 3.0 XXBR<x>
6097 vector byte reverse instructions.
6098 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
6099 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
6100 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
6101 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
6102 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
6103 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
6104 (P9V_BUILTIN_VEC_REVB): Likewise.
6105 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
6106 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
6107 (p9_xxbrq_v16qi): Likewise.
6108 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
6109 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
6110 (p9_xxbrh_v8hi): Likewise.
6111 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
6112 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
6113 vec_revb built-in functions.
6115 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
6117 PR rtl-optimization/78952
6118 * config/i386/i386.md (any_extract): New code iterator.
6119 (*insvqi_2): Use any_extract for source operand.
6120 (*insvqi_3): Use any_shiftrt for source operand.
6122 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
6124 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
6126 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
6128 2017-01-18 Matthias Klose <doko@ubuntu.com>
6130 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
6132 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6134 * config/rs6000/altivec.h (vec_bperm): Change #define.
6135 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
6136 (altivec_vbpermq2): New define_insn.
6137 (altivec_vbpermd): Likewise.
6138 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
6140 (VBPERMD): Likewise.
6141 (VBPERM): New polymorphic function interface.
6142 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
6143 Add entries for P9V_BUILTIN_VEC_VBPERM.
6144 * doc/extend.texi: Add interfaces for vec_bperm.
6146 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6148 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
6149 first letter of error messages.
6150 (s390_resolve_overloaded_builtin): Likewise.
6151 * config/s390/s390.c (s390_expand_builtin): Likewise.
6152 (s390_invalid_arg_for_unprototyped_fn): Likewise.
6153 (s390_valid_target_attribute_inner_p): Likewise.
6154 * config/s390/s390.md ("tabort"): Likewise.
6156 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
6158 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
6159 (ISA_AVOID_DIV_HILO): New macro.
6160 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
6161 (ISA_HAS_DDIV): Likewise.
6163 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6165 * doc/invoke.texi (fabi-version): Correct number of occurrences.
6167 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6169 * doc/invoke.texi (fabi-version): Spelling fix.
6171 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6174 * doc/invoke.texi (fabi-version): Mention mangling fix for
6177 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6180 * doc/invoke.texi (fabi-version): Document discriminator mangling.
6182 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
6185 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
6186 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
6188 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
6190 (stack_protect_test): Ditto.
6191 * config/rs6000/rs6000.opt (mstack-protector-guard=,
6192 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
6194 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
6195 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
6196 -mstack-protector-guard-offset=.
6197 (RS/6000 and PowerPC Options): Ditto.
6199 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
6201 * config/i386/i386.h (MASK_CLASS_P): New define.
6202 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
6203 there are no registers from different register sets also when
6204 mask registers are used. Update function comment.
6205 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
6206 to (*k/*r) and (*k/*km) alternatives.
6208 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
6210 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
6211 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
6212 (EH_RETURN_HANDLER_RTX): New define.
6213 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
6214 Force frame pointer in EH return functions.
6215 (aarch64_expand_epilogue): Add barrier for eh_return.
6216 (aarch64_final_eh_return_addr): Remove.
6217 (aarch64_eh_return_handler_rtx): New function.
6218 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
6220 (aarch64_eh_return_handler_rtx): New prototype.
6222 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6224 * config/rs6000/altivec.h (vec_rlmi): New #define.
6225 (vec_vrlnm): Likewise.
6226 (vec_rlnm): Likewise.
6227 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
6228 (UNSPEC_VRLNM): Likewise.
6229 (VIlong): New mode iterator.
6230 (altivec_vrl<VI_char>mi): New define_insn.
6231 (altivec_vrl<VI_char>nm): Likewise.
6232 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
6235 (RLNM): New polymorphic function entry.
6236 (VRLWMI): New monomorphic function entry.
6238 (RLMI): New polymorphic function entry.
6239 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
6240 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
6241 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
6244 2017-01-17 Jakub Jelinek <jakub@redhat.com>
6247 * dwarf2out.c (field_byte_offset): Restore the
6248 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
6249 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
6250 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
6253 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
6256 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
6258 2017-01-17 Jakub Jelinek <jakub@redhat.com>
6261 * dwarf2out.c (add_data_member_location_attribute): For constant
6262 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
6263 instead of DW_AT_data_member_location, DW_AT_bit_offset and
6264 DW_AT_byte_size attributes.
6266 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
6268 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
6269 after forcing to constant memory when the code model is medium.
6271 2017-01-17 Julia Koval <julia.koval@intel.com>
6274 * config/i386/avx512fintrin.h
6275 (_mm512_i32gather_ps): Change __addr type to void const*.
6276 (_mm512_mask_i32gather_ps): Ditto.
6277 (_mm512_i32gather_pd): Ditto.
6278 (_mm512_mask_i32gather_pd): Ditto.
6279 (_mm512_i64gather_ps): Ditto.
6280 (_mm512_mask_i64gather_ps): Ditto.
6281 (_mm512_i64gather_pd): Ditto.
6282 (_mm512_mask_i64gather_pd): Ditto.
6283 (_mm512_i32gather_epi32): Ditto.
6284 (_mm512_mask_i32gather_epi32): Ditto.
6285 (_mm512_i32gather_epi64): Ditto.
6286 (_mm512_mask_i32gather_epi64): Ditto.
6287 (_mm512_i64gather_epi32): Ditto.
6288 (_mm512_mask_i64gather_epi32): Ditto.
6289 (_mm512_i64gather_epi64): Ditto.
6290 (_mm512_mask_i64gather_epi64): Ditto.
6291 (_mm512_i32scatter_ps): Change __addr type to void*.
6292 (_mm512_mask_i32scatter_ps): Ditto.
6293 (_mm512_i32scatter_pd): Ditto.
6294 (_mm512_mask_i32scatter_pd): Ditto.
6295 (_mm512_i64scatter_ps): Ditto.
6296 (_mm512_mask_i64scatter_ps): Ditto.
6297 (_mm512_i64scatter_pd): Ditto.
6298 (_mm512_mask_i64scatter_pd): Ditto.
6299 (_mm512_i32scatter_epi32): Ditto.
6300 (_mm512_mask_i32scatter_epi32): Ditto.
6301 (_mm512_i32scatter_epi64): Ditto.
6302 (_mm512_mask_i32scatter_epi64): Ditto.
6303 (_mm512_i64scatter_epi32): Ditto.
6304 (_mm512_mask_i64scatter_epi32): Ditto.
6305 (_mm512_i64scatter_epi64): Ditto.
6306 (_mm512_mask_i64scatter_epi64): Ditto.
6307 * config/i386/avx512pfintrin.h
6308 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
6309 (_mm512_mask_prefetch_i32gather_ps): Ditto.
6310 (_mm512_mask_prefetch_i64gather_pd): Ditto.
6311 (_mm512_mask_prefetch_i64gather_ps): Ditto.
6312 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
6313 (_mm512_prefetch_i32scatter_ps): Ditto.
6314 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
6315 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
6316 (_mm512_prefetch_i64scatter_pd): Ditto.
6317 (_mm512_prefetch_i64scatter_ps): Ditto.
6318 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
6319 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
6320 * config/i386/avx512vlintrin.h
6321 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
6322 (_mm_mmask_i32gather_ps): Ditto.
6323 (_mm256_mmask_i32gather_pd): Ditto.
6324 (_mm_mmask_i32gather_pd): Ditto.
6325 (_mm256_mmask_i64gather_ps): Ditto.
6326 (_mm_mmask_i64gather_ps): Ditto.
6327 (_mm256_mmask_i64gather_pd): Ditto.
6328 (_mm_mmask_i64gather_pd): Ditto.
6329 (_mm256_mmask_i32gather_epi32): Ditto.
6330 (_mm_mmask_i32gather_epi32): Ditto.
6331 (_mm256_mmask_i32gather_epi64): Ditto.
6332 (_mm_mmask_i32gather_epi64): Ditto.
6333 (_mm256_mmask_i64gather_epi32): Ditto.
6334 (_mm_mmask_i64gather_epi32): Ditto.
6335 (_mm256_mmask_i64gather_epi64): Ditto.
6336 (_mm_mmask_i64gather_epi64): Ditto.
6337 (_mm256_i32scatter_ps): Change __addr type to void*.
6338 (_mm256_mask_i32scatter_ps): Ditto.
6339 (_mm_i32scatter_ps): Ditto.
6340 (_mm_mask_i32scatter_ps): Ditto.
6341 (_mm256_i32scatter_pd): Ditto.
6342 (_mm256_mask_i32scatter_pd): Ditto.
6343 (_mm_i32scatter_pd): Ditto.
6344 (_mm_mask_i32scatter_pd): Ditto.
6345 (_mm256_i64scatter_ps): Ditto.
6346 (_mm256_mask_i64scatter_ps): Ditto.
6347 (_mm_i64scatter_ps): Ditto.
6348 (_mm_mask_i64scatter_ps): Ditto.
6349 (_mm256_i64scatter_pd): Ditto.
6350 (_mm256_mask_i64scatter_pd): Ditto.
6351 (_mm_i64scatter_pd): Ditto.
6352 (_mm_mask_i64scatter_pd): Ditto.
6353 (_mm256_i32scatter_epi32): Ditto.
6354 (_mm256_mask_i32scatter_epi32): Ditto.
6355 (_mm_i32scatter_epi32): Ditto.
6356 (_mm_mask_i32scatter_epi32): Ditto.
6357 (_mm256_i32scatter_epi64): Ditto.
6358 (_mm256_mask_i32scatter_epi64): Ditto.
6359 (_mm_i32scatter_epi64): Ditto.
6360 (_mm_mask_i32scatter_epi64): Ditto.
6361 (_mm256_i64scatter_epi32): Ditto.
6362 (_mm256_mask_i64scatter_epi32): Ditto.
6363 (_mm_i64scatter_epi32): Ditto.
6364 (_mm_mask_i64scatter_epi32): Ditto.
6365 (_mm256_i64scatter_epi64): Ditto.
6366 (_mm256_mask_i64scatter_epi64): Ditto.
6367 (_mm_i64scatter_epi64): Ditto.
6368 (_mm_mask_i64scatter_epi64): Ditto.
6369 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
6370 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
6371 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
6372 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
6373 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
6374 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
6375 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
6376 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
6377 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
6378 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
6379 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
6380 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
6381 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
6382 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
6383 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
6384 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
6385 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
6386 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
6387 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
6388 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
6389 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
6390 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
6391 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
6392 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
6393 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
6394 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
6395 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
6396 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
6397 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
6398 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
6399 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
6400 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
6401 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
6402 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
6403 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
6404 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
6405 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
6406 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
6407 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
6408 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
6409 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
6410 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
6411 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
6412 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
6413 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
6414 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
6415 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
6416 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
6417 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
6418 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
6419 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
6420 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
6421 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
6422 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
6423 definitions accordingly.
6425 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
6426 Kuan-Lin Chen <kuanlinchentw@gmail.com>
6429 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
6432 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
6435 * ira-conflicts.c (ira_build_conflicts): Update total conflict
6436 hard regs for inner regno.
6438 2017-01-17 Martin Liska <mliska@suse.cz>
6441 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
6442 assumption and add comment.
6444 2017-01-17 Nathan Sidwell <nathan@acm.org>
6446 * ipa-visibility.c (localize_node): New function, broken out of ...
6447 (function_and_variable_visibility): ... here. Call it.
6449 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
6452 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
6453 correctly set frequency of oudgoing edge.
6454 (duplicate_thread_path): Fix profile updating.
6456 2017-01-17 Jakub Jelinek <jakub@redhat.com>
6459 * configure.ac: Add GCC_BASE_VER.
6460 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
6461 version from BASE-VER file.
6462 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
6463 (gcc.o): Depend on $(BASEVER).
6464 * common.opt (dumpfullversion): New option.
6465 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
6466 * doc/invoke.texi: Document -dumpfullversion.
6467 * doc/install.texi: Document --with-gcc-major-version-only.
6468 * configure: Regenerated.
6470 2017-01-17 Richard Biener <rguenther@suse.de>
6472 PR tree-optimization/71433
6473 * tree-vrp.c (register_new_assert_for): Merge same asserts
6474 on all incoming edges.
6475 (process_assert_insertions_for): Handle insertions at the
6478 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
6480 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
6481 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
6483 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
6486 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
6489 2017-01-17 Alan Modra <amodra@gmail.com>
6492 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
6493 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
6494 symbolic stack limit when pic.
6496 2017-01-16 Martin Sebor <msebor@redhat.com>
6498 PR tree-optimization/78608
6499 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
6501 2017-01-16 Jeff Law <law@redhat.com>
6504 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
6505 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
6506 for several include directories that may be relative to sysroot.
6507 * config/i386/x-mingw32 (gplus_includedir): Define.
6508 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
6509 (native_system_includedir): Likewise.
6510 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
6511 override if TARGET_SYSTEM_ROOT is defined.
6512 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
6514 PR tree-optimization/79090
6515 PR tree-optimization/33562
6516 PR tree-optimization/61912
6517 PR tree-optimization/77485
6518 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
6519 and computed trims into the dump file.
6521 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
6523 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
6525 2017-01-16 Jakub Jelinek <jakub@redhat.com>
6528 * gimplify.c (gimplify_init_constructor): If want_value and
6529 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
6533 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
6534 sequence. Formatting fixes.
6535 (doloop_optimize): Formatting fixes.
6538 * gcc.c (debug_level_greater_than_spec_func): New function.
6539 (static_spec_functions): Add debug-level-gt spec function.
6540 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
6542 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
6543 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
6544 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
6545 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
6546 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
6547 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
6549 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
6551 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
6552 QImode fixups to general and mask registers only.
6554 2017-01-16 Carl Love <cel@us.ibm.com>
6556 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
6557 for built-in functions
6558 vector signed char vec_nabs (vector signed char)
6559 vector signed short vec_nabs (vector signed short)
6560 vector signed int vec_nabs (vector signed int)
6561 vector signed long long vec_nabs (vector signed long long)
6562 vector float vec_nabs (vector float)
6563 vector double vec_nabs (vector double)
6564 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
6566 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
6567 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
6568 * doc/extend.texi: Update the documentation file for the new built-in
6571 2017-01-16 Martin Sebor <msebor@redhat.com>
6573 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
6576 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6578 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
6579 UNSPEC_VSX__XXSPLTD to require special splat handling.
6581 2017-01-16 David Malcolm <dmalcolm@redhat.com>
6584 * system.h: Poison strndup.
6586 2017-01-16 Alan Modra <amodra@gmail.com>
6589 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
6592 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
6594 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
6596 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
6598 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
6599 call recog here. Assert that INSN_CODE (insn) is non-negative.
6601 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
6604 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
6606 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
6607 in the currently scheduled RTL fragment.
6609 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
6611 PR rtl-optimization/78751
6612 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
6615 2017-01-14 Jeff Law <law@redhat.com>
6617 PR tree-optimization/79090
6618 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
6619 variable length stores.
6620 (compute_trims): Delete dead assignment to *trim_tail.
6621 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
6624 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
6626 PR rtl-optimization/78626
6627 PR rtl-optimization/78727
6628 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
6629 of a block, and split such blocks after everything else is finished.
6631 2017-01-14 Alan Modra <amodra@gmail.com>
6634 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
6635 target legitimate_combined_insn.
6636 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
6637 (rs6000_legitimate_combined_insn): New function.
6638 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
6640 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
6641 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
6642 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
6644 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
6646 * doc/frontends.texi (G++ and GCC): Remove references to Java.
6648 2017-01-13 Jeff Law <law@redhat.com>
6650 PR tree-optimization/33562
6651 PR tree-optimization/61912
6652 PR tree-optimization/77485
6653 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
6655 (delete_dead_assignment): Likewise.
6656 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
6657 statement to delete_dead_call and delete_dead_assignment.
6659 2017-01-13 David Malcolm <dmalcolm@redhat.com>
6662 * substring-locations.c (format_warning_va): Strengthen case 1 so
6663 that both endpoints of the substring must be within the format
6664 range for just the substring to be printed.
6666 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
6668 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
6669 * config/i386/i386.c (ix86_target_string): Add missing options
6670 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
6671 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
6672 flags_other and ix86_target_other to flags2_other. Display unknown
6674 (ix86_valid_target_attribute_inner_p): Add missing options and
6675 reorder options by implied ISAs, as in ix86_target_string.
6677 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
6679 * hash-table.h (hash_table::too_empty_p): New function.
6680 (hash_table::expand): Use it.
6681 (hash_table::traverse): Likewise.
6682 (hash_table::empty_slot): Use sizeof (value_type) instead of
6683 sizeof (PTR) to convert bytes to elements. Shrink the table
6684 if the current size is excessive for the current number of
6687 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
6689 * ira-costs.c (record_reg_classes): Break from the inner loop
6690 early once alt_fail is known to be true. Update outer loop
6691 handling accordingly.
6693 2017-01-13 Jeff Law <law@redhat.com>
6695 * tree-ssa-dse.c (decrement_count): New function.
6696 (increment_start_addr, maybe_trim_memstar_call): Likewise.
6697 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
6698 when we know the partially dead statement is a mem* function.
6700 PR tree-optimization/61912
6701 PR tree-optimization/77485
6702 * tree-ssa-dse.c: Include expr.h.
6703 (maybe_trim_constructor_store): New function.
6704 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
6706 PR tree-optimization/33562
6707 PR tree-optimization/61912
6708 PR tree-optimization/77485
6709 * doc/invoke.texi: Document new dse-max-object-size param.
6710 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
6711 * tree-ssa-dse.c: Include params.h.
6712 (dse_store_status): New enum.
6713 (initialize_ao_ref_for_dse): New, partially extracted from
6715 (valid_ao_ref_for_dse, normalize_ref): New.
6716 (setup_live_bytes_from_ref, compute_trims): Likewise.
6717 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
6718 (maybe_trim_partially_dead_store): Likewise.
6719 (maybe_trim_complex_store): Likewise.
6720 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
6721 Track what bytes live from the original store. Return tri-state
6722 for dead, partially dead or live.
6723 (dse_dom_walker): Add constructor, destructor and new private members.
6724 (delete_dead_call, delete_dead_assignment): New extracted from
6726 (dse_optimize_stmt): Make a member of dse_dom_walker.
6727 Use initialize_ao_ref_for_dse.
6729 PR tree-optimization/33562
6730 PR tree-optimization/61912
6731 PR tree-optimization/77485
6732 * sbitmap.h (bitmap_count_bits): Prototype.
6733 (bitmap_clear_range, bitmap_set_range): Likewise.
6734 * sbitmap.c (bitmap_clear_range): New function.
6735 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
6737 2017-01-13 Martin Liska <mliska@suse.cz>
6740 * function.c (set_cfun): Add new argument force.
6741 * function.h (set_cfun): Likewise.
6742 * ipa-inline-transform.c (inline_call): Use the function when
6743 strict alising from is dropped for function we inline to.
6745 2017-01-13 Richard Biener <rguenther@suse.de>
6747 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
6748 for dumping GIMPLE INTEGER_CSTs.
6750 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6752 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
6753 to 201112L since C++17.
6755 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
6758 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
6759 if -fsanitize=kernel-address is present.
6761 2017-01-13 Richard Biener <rguenther@suse.de>
6763 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
6764 as _Literal ( type ) number in case usual suffixes do not
6765 preserve all information.
6767 2017-01-13 Richard Biener <rguenther@suse.de>
6769 PR tree-optimization/77283
6770 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
6771 and ssa-iterators.h.
6772 (is_feasible_trace): Implement a cost model based on joiner
6775 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6778 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
6779 char or short to __float128/_Float128 directly.
6781 2017-01-12 Martin Sebor <msebor@redhat.com>
6783 to -Wformat-overflow.
6784 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
6785 (min_bytes_remaining): Same.
6786 (get_string_length): Same.
6787 (format_string): Same.
6788 (format_directive): Same.
6790 (pass_sprintf_length::handle_gimple_call): Same.
6792 2017-01-12 Jakub Jelinek <jakub@redhat.com>
6794 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
6795 info.nowrite calls with no lhs that can't throw. Return bool
6796 whether gsi_remove has been called or not.
6797 (pass_sprintf_length::handle_gimple_call): Return bool whether
6798 try_substitute_return_value called gsi_remove. Formatting fix.
6799 (pass_sprintf_length::execute): Don't use gsi_remove if
6800 handle_gimple_call returned true.
6803 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
6804 be removed due to side-effects, don't remove following barrier nor
6805 turn the successor edge into fallthru edge.
6807 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6810 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
6811 element-reversing loads and stores as not swappable.
6813 2017-01-12 Nathan Sidwell <nathan@acm.org>
6814 Nicolai Stange <nicstange@gmail.com>
6816 * combine.c (try_combine): Don't ignore result of overlap checking
6817 loop. Combine overlap & asm check into single loop.
6819 2017-01-12 Richard Biener <rguenther@suse.de>
6821 * tree-pretty-print.c (dump_generic_node): Provide -gimple
6822 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
6824 2017-01-12 Richard Biener <rguenther@suse.de>
6826 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
6827 and TS_TARGET_OPTION directly derive from TS_BASE.
6828 * tree-core.h (tree_optimization_option): Derive from tree_base.
6829 (tree_target_option): Likewise.
6831 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
6833 * config/i386/i386.c (memory_address_length): Increase len
6834 only when rip_relative_addr_p returns false.
6836 2017-01-11 Julia Koval <julia.koval@intel.com>
6838 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
6839 (OPTION_MASK_ISA_SGX_SET): New.
6840 (ix86_handle_option): Handle OPT_msgx.
6841 * config.gcc: Added sgxintrin.h.
6842 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
6843 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
6844 * config/i386/i386.c (ix86_target_string): Add -msgx.
6846 (ix86_option_override_internal): Handle new options.
6847 (ix86_valid_target_attribute_inner_p): Add sgx.
6848 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
6849 * config/i386/i386.opt: Add msgx.
6850 * config/i386/sgxintrin.h: New file.
6851 * config/i386/x86intrin.h: Add sgxintrin.h.
6853 2017-01-11 Jakub Jelinek <jakub@redhat.com>
6856 * fold-const.c (maybe_nonzero_address): Return 1 for function
6858 (tree_single_nonzero_warnv_p): Don't handle function local objects
6862 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
6865 2017-01-11 David Malcolm <dmalcolm@redhat.com>
6868 * opts.c: Include "spellcheck.h"
6869 (struct string_fragment): New struct.
6870 (struct edit_distance_traits<const string_fragment &>): New
6872 (get_closest_sanitizer_option): New function.
6873 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
6875 2017-01-11 Jakub Jelinek <jakub@redhat.com>
6877 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
6879 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
6880 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
6881 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
6882 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
6883 for initial die_offset if dwarf_split_debug_info.
6884 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
6885 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
6887 (output_skeleton_debug_sections): Formatting fix. Use
6888 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
6889 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
6891 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
6893 * config/arm/cortex-a53.md: Add bypasses for
6895 (cortex_a53_r2f): Only use for transfers.
6896 (cortex_a53_f2r): Likewise.
6897 (cortex_a53_r2f_cvt): Add reservation for conversions.
6898 (cortex_a53_f2r_cvt): Likewise.
6900 2017-01-11 Tamar Christina <tamar.christina@arm.com>
6902 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
6903 to all inlined functions, change static to extern.
6905 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
6908 * config/arm/arm.c (legitimize_pic_address): Handle reference to
6910 (arm_assemble_integer): Likewise.
6912 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
6914 * config.gcc: Use new awk script to check CPU, FPU and architecture
6915 parameters for --with-... options.
6916 * config/arm/parsecpu.awk: New file
6917 * config/arm/arm-cpus.in: New file.
6918 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
6920 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
6922 * config/arm/t-arm: Update dependency rules.
6923 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
6924 of processing .def files.
6925 * config/arm/genopt.sh: Deleted.
6926 * config/arm/gentune.sh: Deleted.
6927 * config/arm/arm-cores.def: Deleted.
6928 * config/arm/arm-arches.def: Deleted.
6929 * config/arm/arm-fpus.def: Deleted.
6930 * config/arm/arm-tune.md: Regenerated.
6931 * config/arm/arm-tables.opt: Regenerated.
6932 * config/arm/arm-cpu.h: New generated file.
6933 * config/arm/arm-cpu-data.h: New generated file.
6934 * config/arm/arm-cpu-cdata.h: New generated file.
6936 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
6939 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
6941 (input_varpool_node): Unpack dynamically_initialized bit.
6943 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
6945 PR rtl-optimization/79032
6946 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
6947 the alignment of the adjusted memory reference against that of MODE,
6948 instead of the alignment of the original memory reference.
6950 2017-01-11 Martin Jambor <mjambor@suse.cz>
6952 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
6954 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
6955 decorated functions.
6957 2017-01-11 Richard Biener <rguenther@suse.de>
6959 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
6960 set range/nonnull info for PHI results. Do not set it on
6961 stmts marked for removal.
6963 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
6965 * expr.c (store_field): In the bitfield case, fetch the return value
6966 from the registers before applying a single big-endian adjustment.
6967 Always do a final load for a BLKmode value not larger than a word.
6969 2017-01-10 David Malcolm <dmalcolm@redhat.com>
6972 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
6973 that we correctly handle column numbers greater than
6974 LINE_MAP_MAX_COLUMN_NUMBER.
6976 2017-01-10 Martin Sebor <msebor@redhat.com>
6979 * gimple-ssa-sprintf.c (get_destination_size): Call
6980 {init,fini}object_sizes.
6981 * tree-object-size.c (addr_object_size): Adjust.
6982 (pass_through_call): Adjust.
6983 (pass_object_sizes::execute): Adjust.
6984 * tree-object-size.h (fini_object_sizes): Declare.
6986 2017-01-10 Martin Sebor <msebor@redhat.com>
6988 PR tree-optimization/78775
6989 * builtins.c (get_size_range): Move...
6990 * calls.c: ...to here.
6991 (alloc_max_size): Accept zero argument.
6992 (operand_signed_p): Remove.
6993 (maybe_warn_alloc_args_overflow): Call get_size_range.
6994 * calls.h (get_size_range): Declare.
6996 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
6998 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
6999 from TI's devices.csv file as of September 2016.
7000 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
7002 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
7004 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
7005 * doc/invoke.texi: Likewise.
7006 * doc/md.texi: Likewise.
7007 * doc/objc.texi: Likewise.
7009 2017-01-10 Joshua Conner <joshconner@google.com>
7011 * config/arm/fuchsia-elf.h: New file.
7012 * config/fuchsia.h: New file.
7013 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
7014 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
7016 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
7018 2016-01-10 Richard Biener <rguenther@suse.de>
7020 PR tree-optimization/79034
7021 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
7022 Propagate out degenerate PHIs in the joiner.
7024 2017-01-10 Martin Liska <mliska@suse.cz>
7026 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
7027 (sort_congruence_classes_by_decl_uid): Likewise.
7028 (sort_congruence_class_groups_by_decl_uid): Likewise.
7029 (sem_item_optimizer::merge_classes): Sort class, groups in these
7030 classes and members in the groups by DECL_UID of declarations.
7031 This would make merge operations stable.
7033 2017-01-10 Martin Liska <mliska@suse.cz>
7035 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
7036 usage of m_classes_vec.
7037 (sem_item_optimizer::~sem_item_optimizer): Likewise.
7038 (sem_item_optimizer::get_group_by_hash): Likewise.
7039 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
7040 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
7041 (sem_item_optimizer::verify_classes): Likewise.
7042 (sem_item_optimizer::process_cong_reduction): Likewise.
7043 (sem_item_optimizer::dump_cong_classes): Likewise.
7044 (sem_item_optimizer::merge_classes): Likewise.
7045 * ipa-icf.h (congruence_class_hash): Rename from
7046 congruence_class_group_hash. Remove declaration of m_classes_vec.
7048 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
7050 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
7051 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
7052 * config.gcc: Add avx512vpopcntdqintrin.h.
7053 * config/i386/avx512vpopcntdqintrin.h: New.
7054 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
7055 * config/i386/i386-builtin-types.def: Add new types.
7056 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
7057 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
7058 __builtin_ia32_vpopcountq_v8di_mask): New.
7059 * config/i386/i386-c.c (ix86_target_macros_internal): Define
7060 __AVX512VPOPCNTDQ__.
7061 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
7062 (PTA_AVX512VPOPCNTDQ): Define.
7063 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
7064 TARGET_AVX512VPOPCNTDQ_P): Define.
7065 * config/i386/i386.opt: Add mavx512vpopcntdq.
7066 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
7067 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
7069 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7072 * predict.def (PRED_CALL): Set to 67.
7074 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
7076 * expr.c (store_field): In the bitfield case, if the value comes from
7077 a function call and is of an aggregate type returned in registers, do
7078 not modify the field mode; extract the value in all cases if the mode
7079 is BLKmode and the size is not larger than a word.
7081 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
7084 * config/i386/cpuid.h: Fix undefined behavior.
7086 2017-01-04 Jeff Law <law@redhat.com>
7088 PR tree-optimization/79007
7089 PR tree-optimization/67955
7090 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
7091 conservative for pt.null when flag_non_call_exceptions is on.
7093 2017-01-09 Jakub Jelinek <jakub@redhat.com>
7095 PR translation/79019
7096 PR translation/79020
7097 * params.def (PARAM_INLINE_MIN_SPEEDUP,
7098 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
7099 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
7101 * config/avr/avr.opt (maccumulate-args): Likewise.
7102 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
7103 * common.opt (freport-bug): Likewise.
7104 * cif-code.def (CIF_FINAL_ERROR): Likewise.
7105 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
7106 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
7107 translatable string.
7108 * config/i386/i386.c (function_value_32): Likewise.
7109 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
7110 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
7112 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
7113 * common/config/msp430/msp430-common.c (msp430_handle_option):
7115 * symtab.c (symtab_node::verify_base): Likewise.
7116 * opts.c (set_debug_level): Likewise.
7117 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
7118 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
7119 missing whitespace to translatable strings.
7120 * config/avr/avr.md (bswapsi2): Fix typo in comment.
7121 * config/sh/superh.h: Likewise.
7122 * config/i386/xopintrin.h: Likewise.
7123 * config/i386/znver1.md: Likewise.
7124 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
7125 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
7126 * double-int.h (struct double_int): Likewise.
7127 * double-int.c (div_and_round_double): Likewise.
7128 * wide-int.cc: Likewise.
7129 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
7130 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
7131 * cfgcleanup.c (crossjumps_occured): Renamed to ...
7132 (crossjumps_occurred): ... this.
7133 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
7136 PR tree-optimization/78899
7137 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
7138 returning bool return struct loop *, NULL for failure and the new
7140 (versionable_outer_loop_p): Don't version outer loop if it has
7141 dont_vectorized bit set.
7142 (tree_if_conversion): When versioning outer loop, ensure
7143 tree_if_conversion is performed also on the inner loop of the
7144 non-vectorizable outer loop copy.
7145 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
7146 LOOP_VECTORIZED in inner loop of the scalar outer loop and
7147 prevent vectorization of it.
7148 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
7149 the outer loop vectorization of the non-scalar version is attempted
7150 before vectorization of the inner loop in scalar version. If
7151 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
7152 vectorization of its inner loop.
7153 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
7154 has 2 inner loops, rename also on edges from bb whose single pred
7155 is outer_loop->header. Fix typo in function comment.
7157 2017-01-09 Martin Sebor <msebor@redhat.com>
7160 * asan.c (asan_emit_stack_protection): Increase local buffer size
7161 to avoid snprintf truncation warning.
7163 2017-01-09 Andrew Pinski <apinski@cavium.com>
7165 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
7166 to reference thunderx2t99 for the tuning structure
7167 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
7169 (thunderx2t99_extra_costs): This.
7170 * config/aarch64/aarch64-tune.md: Regenerate.
7171 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
7172 (vulcan_addrcost_table): This.
7173 (vulcan_regmove_cost): Rename to ...
7174 (thunderx2t99_regmove_cost): This.
7175 (vulcan_vector_cost): Rename to ...
7176 (thunderx2t99_vector_cost): this.
7177 (vulcan_branch_cost): Rename to ...
7178 (thunderx2t99_branch_cost): This.
7179 (vulcan_tunings): Rename to ...
7180 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
7181 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
7183 2017-01-09 Martin Jambor <mjambor@suse.cz>
7187 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
7188 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
7189 (propagate_vr_accross_jump_function): Use the above function for all
7190 value range computations for pass-through jump functions and type
7191 converasion from explicit value range values.
7192 (ipcp_propagate_stage): Do not attempt to deduce types of formal
7193 parameters from TYPE_ARG_TYPES.
7194 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
7195 (ipa_write_node_info): Stream type of the actual argument.
7196 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
7198 2017-01-09 Martin Liska <mliska@suse.cz>
7201 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
7202 (lookup_compiler): Do not show error message with have_E.
7204 2017-01-09 Jakub Jelinek <jakub@redhat.com>
7206 PR tree-optimization/78938
7207 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
7208 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
7209 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
7210 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
7213 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7215 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
7218 2017-01-09 Richard Biener <rguenther@suse.de>
7220 PR tree-optimization/78997
7221 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
7222 name condition properly.
7224 2017-01-09 Richard Biener <rguenther@suse.de>
7227 * dwarf2out.c (is_cxx): New overload with context.
7228 (is_naming_typedef_decl): Use it.
7230 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7232 * invoke.texi (Option Summary): Correct spacing in option lists
7233 and add line breaks to fix over-long lines.
7235 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7239 * extend.texi (Common Variable Attributes): Add xref to GCC
7240 Internals manual to explain mode attribute keywords.
7242 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7245 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
7246 and Preprocessor Options.
7247 (Options for Linking): Document -pthread here....
7248 (RS/6000 and PowerPC Options): ...not here.
7249 (Solaris 2 Options): ...or here.
7250 * doc/cppopts.texi: Document -pthread.
7252 2017-01-08 Martin Sebor <msebor@redhat.com>
7255 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
7256 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
7257 New member functions.
7258 (format_directive): Used them.
7260 (pass_sprintf_length::handle_gimple_call): Same.
7261 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
7262 to avoid truncation for any argument.
7263 (extract_affine_mul): Same.
7264 * tree.c (get_file_function_name): Same.
7266 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7269 * predict.def (PRED_INDIR_CALL): Set to 86.
7271 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
7273 PR preprocessor/54124
7274 * doc/cppopts.texi: Reformat -d subtable to list the full name
7275 of the options. Add cross-reference to the docs for the general
7276 compiler -d options.
7277 * doc/invoke.texi (Developer Options): Add cross-reference to the
7278 preprocessor-specific -d option documentation.
7280 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
7282 PR preprocessor/13498
7283 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
7284 redudant material, and reflect new command-line options.
7285 (System Headers): Likewise.
7287 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
7289 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
7290 -isystem, and -idirafter. Copy-edit.
7291 * doc/cppopts.texi: Copy-edit. Remove contradiction about
7292 default for -ftrack-macro-expansion. Delete obsolete and
7293 badly-formatted implementation details about -fdebug-cpp output.
7294 * doc/cppwarnopts.texi: Copy-edit.
7296 2017-01-07 David Malcolm <dmalcolm@redhat.com>
7299 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
7300 that the transition from a max line width >= 1<<10 to narrower
7301 lines works correctly.
7303 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
7305 * doc/options.texi (PerFunction): New.
7306 * opt-functions.awk (switch_flags): Map both Optimization and
7307 PerFunction to CL_OPTIMIZATION.
7308 * opth-gen.awk: Test for PerFunction flag along with
7310 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
7311 it only when the latter is present. Skip those that don't in
7312 the hash function generator.
7313 * common.opt (fvar-tracking): Mark as PerFunction instead of
7315 (fvar-tracking-assignments): Likewise.
7316 (fvar-tracking-assignments-toggle): Likewise.
7317 (fvar-tracking-uninit): Likewise.
7319 2017-01-07 Jakub Jelinek <jakub@redhat.com>
7321 PR translation/79018
7322 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
7325 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
7328 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
7329 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
7330 TARGET_LONG_JUMP_TABLE_OFFSETS.
7331 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
7332 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
7333 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
7334 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
7335 * config/m68k/m68k.md (tablejump expander): Likewise.
7336 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
7337 TARGET_LONG_JUMP_TABLE_OFFSETS.
7338 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
7339 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
7341 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
7342 David Holsgrove <david.holsgrove@xilinx.com>
7344 * common/config/microblaze/microblaze-common.c
7345 (TARGET_EXCEPT_UNWIND_INFO): Remove.
7346 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
7348 * config/microblaze/microblaze.c (microblaze_must_save_register)
7349 (microblaze_expand_epilogue, microblaze_return_addr): Handle
7351 (microblaze_eh_return): New function.
7352 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
7353 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
7354 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
7355 * config/microblaze/microblaze.md (eh_return): New pattern.
7357 2017-01-06 Jakub Jelinek <jakub@redhat.com>
7359 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
7360 GCC_DIAGNOSTIC_STRINGIFY): Define.
7362 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
7364 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7366 * config/arm/arm.md (<mcrr>): New.
7368 * config/arm/arm.c (arm_arch5te): New.
7369 (arm_option_override): Set arm_arch5te.
7370 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
7372 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
7373 (arm_mcrr_qualifiers): ... this. New.
7374 (MRRC_QUALIFIERS): Define to...
7375 (arm_mrrc_qualifiers): ... this. New.
7376 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
7378 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
7379 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
7380 (MRRCI, mrrc, MRRC): New.
7381 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
7382 VUNSPEC_MRRC2): New.
7384 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7386 * config/arm/arm.md (<mcr>): New.
7388 * config/arm/arm.c (arm_coproc_builtin_available): Add
7389 support for mcr, mrc, mcr2 and mrc2.
7390 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
7391 (arm_mcr_qualifiers): ... this. New.
7392 (MRC_QUALIFIERS): Define to ...
7393 (arm_mrc_qualifiers): ... this. New.
7394 (MCR_QUALIFIERS): Define to ...
7395 (arm_mcr_qualifiers): ... this. New.
7396 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
7398 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
7399 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
7400 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
7403 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7405 * config/arm/arm.md (*ldc): New.
7409 * config/arm/arm.c (arm_coproc_builtin_available): Add
7410 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
7411 (arm_coproc_ldc_stc_legitimate_address): New.
7412 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
7413 'qualifier_const_pointer'.
7414 (LDC_QUALIFIERS): Define to...
7415 (arm_ldc_qualifiers): ... this. New.
7416 (STC_QUALIFIERS): Define to...
7417 (arm_stc_qualifiers): ... this. New.
7418 * config/arm/arm-protos.h
7419 (arm_coproc_ldc_stc_legitimate_address): New.
7420 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
7421 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
7422 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
7423 stc2, stcl, stc2l): New.
7424 * config/arm/constraints.md (Uz): New.
7425 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
7426 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
7427 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
7428 VUNSPEC_STC2L): New.
7430 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7432 * config/arm/arm.md (<cdp>): New.
7433 * config/arm/arm.c (neon_const_bounds): Rename this ...
7434 (arm_const_bounds): ... this.
7435 (arm_coproc_builtin_available): New.
7436 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
7437 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
7438 (CDP_QUALIFIERS): Define to...
7439 (arm_cdp_qualifiers): ... this. New.
7441 (arm_expand_builtin_args): Add case for 6 arguments.
7442 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
7443 (arm_const_bounds): ... this.
7444 (arm_coproc_builtin_available): New.
7445 * config/arm/arm_acle.h (__arm_cdp): New.
7447 * config/arm/arm_acle_builtins.def (cdp): New.
7449 * config/arm/iterators.md (CDPI,CDP,cdp): New.
7450 * config/arm/neon.md: Rename all 'neon_const_bounds' to
7452 * config/arm/types.md (coproc): New.
7453 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
7454 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
7455 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
7456 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
7458 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7460 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
7461 (UBINOP_QUALIFIERS): New.
7463 (acle_builtin_data): New. Change comment.
7464 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
7465 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
7466 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
7467 arm_acle_builtins.def.
7468 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
7469 (arm_init_acle_builtins): New.
7470 (CRC32_BUILTIN): Remove.
7471 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
7472 crc32cb, crc32ch and crc32cw.
7473 (arm_init_crc32_builtins): Remove.
7474 (arm_init_builtins): Use arm_init_acle_builtins rather
7475 than arm_init_crc32_builtins.
7476 (arm_expand_acle_builtin): New.
7477 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
7478 * config/arm/arm_acle_builtins.def: New.
7480 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7482 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
7483 (arm_builtin_datum): ... this.
7484 (arm_init_neon_builtin): Rename to ...
7485 (arm_init_builtin): ... this. Add a new parameters PREFIX
7486 and USE_SIG_IN_NAME.
7487 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
7488 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
7489 'arm_builtin_datum'.
7490 (arm_init_vfp_builtins): Likewise.
7491 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
7492 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
7493 (arm_expand_neon_args): Rename to ...
7494 (arm_expand_builtin_args): ... this. Rename builtin_arg
7495 enum values and differentiate between ARG_BUILTIN_MEMORY
7496 and ARG_BUILTIN_NEON_MEMORY.
7497 (arm_expand_neon_builtin_1): Rename to ...
7498 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
7499 values, arm_expand_builtin_args and add bool parameter NEON.
7500 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
7501 (arm_expand_vfp_builtin): Likewise.
7502 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
7504 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7507 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
7508 * predict.c (tree_estimate_probability_bb): Reverse direction of
7509 polymorphic call predictor.
7511 2017-01-06 David Malcolm <dmalcolm@redhat.com>
7513 * passes.c (execute_one_pass): Split out pass-skipping logic into...
7514 (determine_pass_name_match): ...this new function and...
7515 (should_skip_pass_p): ...this new function.
7517 2017-01-06 Nathan Sidwell <nathan@acm.org>
7519 * ipa-visibility.c (function_and_variable_visibility): Reformat
7520 comments and long lines. Remove extrneous if.
7521 * symtab.c (symtab_node::make_decl_local): Fix code format.
7522 (symtab_node::set_section_for_node): Fix comment typo.
7524 2017-01-06 Martin Liska <mliska@suse.cz>
7527 * lra-constraints.c: Rename invariant to lra_invariant.
7528 * predict.c (set_even_probabilities): Initialize e to NULL.
7530 2017-01-05 Martin Sebor <msebor@redhat.com>
7532 PR tree-optimization/78910
7533 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
7534 (format_integer): Correct off-by-one error in the handling
7535 of precision with negative numbers in signed conversions..
7537 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
7539 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
7541 2017-01-05 Jakub Jelinek <jakub@redhat.com>
7543 PR tree-optimization/71016
7544 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
7545 factor_out_conditional_conversion. Formatting fix.
7546 (factor_out_conditional_conversion): Add cond_stmt argument.
7547 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
7548 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
7551 2017-01-05 David Malcolm <dmalcolm@redhat.com>
7553 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
7554 read-rtl-function.o, and selftest-rtl.o.
7555 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
7556 (selftest::aarch64_test_loading_full_dump): New function.
7557 (selftest::aarch64_run_selftests): New function.
7558 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
7559 selftest::aarch64_run_selftests.
7560 * config/i386/i386.c
7561 (selftest::ix86_test_loading_dump_fragment_1): New function.
7562 (selftest::ix86_test_loading_call_insn): New function.
7563 (selftest::ix86_test_loading_full_dump): New function.
7564 (selftest::ix86_test_loading_unspec): New function.
7565 (selftest::ix86_run_selftests): Call the new functions.
7566 * emit-rtl.c (maybe_set_max_label_num): New function.
7567 * emit-rtl.h (maybe_set_max_label_num): New decl.
7568 * function.c (instantiate_decls): Guard call to
7569 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
7570 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
7572 * gensupport.c (gen_reader::gen_reader): Pass "false"
7573 for new "compact" param of rtx_reader.
7574 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
7575 rather than an empty string for NULL strings.
7576 * read-md.c: Potentially include config.h rather than bconfig.h.
7577 Wrap include of errors.h with #ifdef GENERATOR_FILE.
7578 (have_error): New global, copied from errors.c.
7579 (md_reader::read_name): Rename to...
7580 (md_reader::read_name_1): ...this, adding "out_loc" param,
7581 and converting "missing name or number" to returning false, rather
7583 (md_reader::read_name): Reimplement in terms of read_name_1.
7584 (md_reader::read_name_or_nil): New function.
7585 (md_reader::read_string): Handle "(nil)" by returning NULL.
7586 (md_reader::md_reader): Add new param "compact".
7587 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
7588 (md_reader::read_file): New method.
7589 * read-md.h (md_reader::md_reader): Add new param "compact".
7590 (md_reader::read_file): New method.
7591 (md_reader::is_compact): New accessor.
7592 (md_reader::read_name): Convert return type from void to file_location.
7593 (md_reader::read_name_or_nil): New decl.
7594 (md_reader::read_name_1): New decl.
7595 (md_reader::m_compact): New field.
7596 (noop_reader::noop_reader): Pass "false" for new "compact" param
7598 (rtx_reader::rtx_reader): Add new "compact" param.
7599 (rtx_reader::read_rtx_operand): Make virtual and convert return
7600 type from void to rtx.
7601 (rtx_reader::read_until): New decl.
7602 (rtx_reader::handle_any_trailing_information): New virtual function.
7603 (rtx_reader::postprocess): New virtual function.
7604 (rtx_reader::finalize_string): New virtual function.
7605 (rtx_reader::m_in_call_function_usage): New field.
7606 (rtx_reader::m_reuse_rtx_by_id): New field.
7607 * read-rtl-function.c: New file.
7608 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
7609 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
7610 (selftest::verify_three_block_rtl_cfg): New decl.
7611 * read-rtl-function.h: New file.
7612 * read-rtl.c: Potentially include config.h rather than bconfig.h.
7613 For host, include function.h, memmodel.h, and emit-rtl.h.
7614 (one_time_initialization): New function.
7615 (struct compact_insn_name): New struct.
7616 (compact_insn_names): New array.
7617 (find_code): Handle insn codes in compact dumps.
7618 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
7619 (bind_subst_iter_and_attr): Likewise.
7620 (add_condition_to_string): Likewise.
7621 (add_condition_to_rtx): Likewise.
7622 (apply_attribute_uses): Likewise.
7623 (add_current_iterators): Likewise.
7624 (apply_iterators): Likewise.
7625 (initialize_iterators): Guard usage of apply_subst_iterator with
7626 #ifdef GENERATOR_FILE.
7627 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
7628 (md_reader::read_mapping): Likewise.
7629 (add_define_attr_for_define_subst): Likewise.
7630 (add_define_subst_attr): Likewise.
7631 (read_subst_mapping): Likewise.
7632 (check_code_iterator): Likewise.
7633 (rtx_reader::read_rtx): Likewise. Move one-time initialization
7635 (one_time_initialization): New function.
7636 (rtx_reader::read_until): New method.
7637 (read_flags): New function.
7638 (parse_reg_note_name): New function.
7639 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
7640 Handle reuse_rtx ids.
7641 Wrap iterator lookup within #ifdef GENERATOR_FILE.
7642 Add parsing support for RTL dumps, mirroring the special-cases in
7643 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
7644 values, and calling handle_any_trailing_information.
7645 (rtx_reader::read_rtx_operand): Convert return type from void
7646 to rtx, returning return_rtx. Handle case 'e'. Call
7647 finalize_string on XSTR and XTMPL fields.
7648 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
7649 "(nil)" values were omitted. Call the postprocess vfunc on the
7651 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
7652 class ctor. Initialize m_in_call_function_usage. Call
7653 one_time_initialization.
7654 * rtl-tests.c (selftest::test_uncond_jump): Call
7655 set_new_first_and_last_insn.
7656 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
7657 * selftest-rtl.c: New file.
7658 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
7659 (selftest::get_insn_by_uid): New decl.
7660 * selftest-run-tests.c (selftest::run_tests): Call
7661 read_rtl_function_c_tests.
7662 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
7663 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
7666 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
7668 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
7669 operands in a special way. Assert that pos+len <= mode precision.
7671 2017-01-05 Jakub Jelinek <jakub@redhat.com>
7673 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
7674 3 argument Alias with unlimited for the negative form.
7675 (fno-vect-cost-model): Removed.
7677 2017-01-05 Martin Liska <mliska@suse.cz>
7679 * hsa-gen.c (gen_hsa_divmod): New function.
7680 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
7682 2017-01-05 Martin Liska <mliska@suse.cz>
7685 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
7688 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7690 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
7691 small constant length operands.
7693 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7695 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
7696 between loop iterations.
7698 2017-01-05 Martin Liska <mliska@suse.cz>
7701 * gimplify.c (gimplify_decl_expr): Compare to
7702 asan_poisoned_variables instread of checking flags.
7703 (gimplify_target_expr): Likewise.
7704 (gimplify_expr): Likewise.
7705 (gimplify_function_tree): Conditionally initialize
7706 asan_poisoned_variables.
7708 2017-01-04 Jeff Law <law@redhat.com>
7710 PR tree-optimizatin/78812
7711 * rtl.h (contains_mem_rtx_p): Prototype.
7712 * ifcvt.c (containts_mem_rtx_p): Move from here to...
7713 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
7714 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
7715 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
7716 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
7718 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7720 * input.c (assert_char_at_range): Default-initialize actual_range.
7722 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7724 * df-scan.c (df_ref_create_structure): Make regno unsigned,
7725 to match the caller.
7727 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7729 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
7730 insns after final jump in test to emit dummy move.
7732 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7734 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
7735 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
7737 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7739 * multiple_target.c (create_dispatcher_calls): Init e_next.
7740 * tree-ssa-loop-split.c (split_loop): Init border.
7741 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
7744 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
7749 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
7750 (altivec_register_operand): Do not return true if the operand
7751 contains a SUBREG mixing SImode and SFmode.
7752 (vsx_register_operand): Likewise.
7753 (vsx_reg_sfsubreg_ok): New predicate.
7754 (vfloat_operand): Do not return true if the operand contains a
7755 SUBREG mixing SImode and SFmode.
7756 (vint_operand): Likewise.
7757 (vlogical_operand): Likewise.
7758 (gpc_reg_operand): Likewise.
7759 (int_reg_operand): Likewise.
7760 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
7761 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
7762 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
7764 (rs6000_emit_move_si_sf_subreg): New helper function.
7765 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
7766 fixup SUBREGs involving SImode and SFmode.
7767 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
7768 numbers for the new peephole2 optimization.
7769 (peephole2 for SFmode unions): New peephole2 to optimize cases in
7770 the GLIBC math library that do AND/IOR/XOR operations on single
7771 precision floating point.
7772 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
7773 target macros to say whether we need to avoid SUBREGs mixing
7775 (TARGET_ALLOW_SF_SUBREG): Likewise.
7776 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
7777 (UNSPEC_SI_FROM_SF): Likewise.
7778 (iorxor): Change spacing.
7779 (and_ior_xor): New iterator for AND, IOR, and XOR.
7780 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
7781 (movdi_from_sf_zero_ext): Likewise.
7782 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
7783 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
7784 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
7785 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
7786 (fms<mode>4): Likewise.
7787 (fnma<mode>4): Likewise.
7788 (fnms<mode>4): Likewise.
7789 (nfma<mode>4): Likewise.
7790 (nfms<mode>4): Likewise.
7792 2017-01-04 Marek Polacek <polacek@redhat.com>
7795 * doc/invoke.texi: Document -Wpointer-compare.
7797 2017-01-04 Jakub Jelinek <jakub@redhat.com>
7799 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
7802 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
7803 descriptions for -gdwarf-5 and emit them as uleb128 instead of
7806 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
7809 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
7810 documentation of the powerpc_popcntb_ok attribute.
7811 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
7812 code to issue warning messages if a requested CPU configuration is
7813 not supported by the binary (assembler and loader) toolchain.
7814 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
7815 made to define a built-in function that has been disabled.
7816 (paired_init_builtins): Add assertion to prevent ICE if attempt is
7817 made to define a built-in function that has been disabled.
7818 (altivec_init_builtins): Add comment explaining why definition
7819 of the DST built-in functions is not preceded by an assertion
7820 check. Add assertions to prevent ICE if attempts are made to
7821 define an altivec predicate or an abs* built-in function that has
7823 (htm_init_builtins): Add comment explaining why definition of the
7824 htm built-in functions is not preceded by an assertion check.
7826 2017-01-04 Jeff Law <law@redhat.com>
7828 PR tree-optimizatin/67955
7829 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
7830 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
7831 the points-to solution does not include pt_null. Use DECL_PT_UID
7834 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
7836 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
7837 Use gen_int_mode instead of gen_lopwart for const_int operands.
7839 2017-01-04 Jakub Jelinek <jakub@redhat.com>
7841 PR tree-optimization/71563
7842 * match.pd: Simplify X << Y into X if Y is known to be 0 or
7843 out of range value - has low bits known to be zero.
7845 2017-01-04 Alan Modra <amodra@gmail.com>
7847 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
7848 * configure: Regenerate.
7849 * config.in: Regenerate.
7851 2017-01-04 Jakub Jelinek <jakub@redhat.com>
7854 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
7855 a substring of the message, but strcmp with the whole message. Ifdef
7856 ENABLE_NLS, translate the message first using dgettext.
7858 2017-01-03 Jeff Law <law@redhat.com>
7860 PR tree-optimizatin/78856
7861 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
7862 (mark_threaded_blocks): Remove code to truncate thread paths that
7863 cross multiple loop headers. Instead invalidate the cached loop
7864 iteration information and handle case of a thread path walking
7865 into an irreducible region.
7867 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7870 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
7871 assertions. Add support for doing the signbit if the IEEE 128-bit
7872 floating point value is in a GPR.
7873 * config/rs6000/rs6000.md (Fsignbit): Delete.
7874 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
7875 Update the length attribute if the value is in a GPR.
7876 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
7877 the sign or zero extension instruction, since the value is always 0/1.
7878 (signbit<mode>2_dm2): Delete using <Fsignbit>.
7881 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
7882 extracting SImode to a GPR register so that we can generate a
7883 store, limit the vector to be in a traditional Altivec register
7884 for the vextuwrx instruction.
7886 2017-01-03 Ian Lance Taylor <iant@google.com>
7888 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
7890 2017-01-03 Martin Sebor <msebor@redhat.com>
7892 PR tree-optimization/78696
7893 * gimple-ssa-sprintf.c (format_floating): Correct handling of
7894 precision. Use MPFR for %f for greater fidelity. Correct handling
7896 (pass_sprintf_length::compute_format_length): Set width and precision
7897 specified by asrerisk to void_node for vararg functions.
7898 (try_substitute_return_value): Adjust dump output.
7900 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
7902 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
7904 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
7906 * doc/invoke.texi (SPARC options): Document -mlra as the default.
7907 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
7908 -mlra/-mno-lra was passed to the compiler.
7910 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
7912 PR rtl-optimization/65618
7913 * emit-rtl.c (try_split): Move initialization of "before" and
7914 "after" to just before the call to emit_insn_after_setloc.
7916 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
7918 * doc/md.texi (Standard Names): Remove reference to Java frontend.
7920 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
7922 * dwarf2out.c (gen_enumeration_type_die): When
7923 -gno-strict-dwarf, add a DW_AT_encoding attribute.
7925 2017-01-03 Jakub Jelinek <jakub@redhat.com>
7927 PR tree-optimization/78965
7928 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
7929 Change first argument from const call_info & to call_info &. For %n
7930 set info.nowrite to false.
7933 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
7934 possibly throwing calls.
7936 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
7937 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
7938 and fns handling, rather than in a separate case SSA_NAME.
7940 2017-01-02 Jeff Law <law@redhat.com>
7942 * config/darwin-driver.c (darwin_driver_init): Const-correctness
7943 fixes for first_period and second_period variables.
7945 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
7948 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
7949 (*insvqi_1): New insn pattern.
7950 (*insvqi_1_mem_rex64): Ditto.
7952 (*insvqi_3): Rename from *insvqi.
7954 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
7956 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
7958 * doc/cfg.texi (Edges): Remove reference to Java.
7959 (Maintaining the CFG): Ditto.
7961 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7964 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
7965 transparent aliases.
7967 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7970 * predict.def (PRED_CALL): Update hitrate.
7971 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
7972 * predict.c (tree_estimate_probability_bb): Split CALL predictor
7973 into direct/indirect/polymorphic variants.
7975 2017-01-01 Jakub Jelinek <jakub@redhat.com>
7977 Update copyright years.
7979 * gcc.c (process_command): Update copyright notice dates.
7980 * gcov-dump.c (print_version): Ditto.
7981 * gcov.c (print_version): Ditto.
7982 * gcov-tool.c (print_version): Ditto.
7983 * gengtype.c (create_file): Ditto.
7984 * doc/cpp.texi: Bump @copying's copyright year.
7985 * doc/cppinternals.texi: Ditto.
7986 * doc/gcc.texi: Ditto.
7987 * doc/gccint.texi: Ditto.
7988 * doc/gcov.texi: Ditto.
7989 * doc/install.texi: Ditto.
7990 * doc/invoke.texi: Ditto.
7992 Copyright (C) 2017 Free Software Foundation, Inc.
7994 Copying and distribution of this file, with or without modification,
7995 are permitted in any medium without royalty provided the copyright
7996 notice and this notice are preserved.