1 2017-04-18 Jeff Law <law@redhat.com>
3 * regcprop.c (maybe_mode_change): Avoid creating copies of the
7 2017-04-13 Jeff Law <law@redhat.com>
8 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
9 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
11 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
14 * config/avr/avr.c (intl.h): Include it.
15 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17 2017-04-18 Martin Liska <mliska@suse.cz>
20 * gcov-tool.c (gcov_output_files): Validate that destination
21 file is either removed by the tool or by a user.
23 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24 Guy Benyei <guybe@mellanox.com>
26 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27 block, and do not negate it, the stored id is already negative.
29 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
31 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
33 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
37 masks of options that should be turned off if the VSX vector
38 options are turned off.
39 (OTHER_P8_VECTOR_MASKS): Likewise.
40 (OTHER_VSX_VECTOR_MASKS): Likewise.
41 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
42 rs6000_disable_incompatible_switches to validate no type switches
44 (rs6000_incompatible_switch): New function to disallow turning on
45 other vector options if -mno-vsx, -mno-power8-vector, or
46 -mno-power9-vector are specified.
48 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
50 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
52 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
54 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
55 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
56 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
57 (ARG_POINTER_CFA_OFFSET): Likewise.
59 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
61 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
62 conditions to take advantage of various optimizations.
64 2017-04-13 Jeff Law <law@redhat.com>
66 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
67 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
68 (zero_extendsidi2_dext): Likewise.
70 2017-04-13 Jakub Jelinek <jakub@redhat.com>
73 * fold-const.c (fold_ternary_loc): Revert
74 use op0 instead of fold_convert_loc (loc, type, arg0) part of
77 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
79 PR rtl-optimization/80343
80 * lra-remat.c (update_scratch_ops): Assign original hard reg to
83 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
86 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
87 to ubsan_encode_value.
89 2017-04-13 Jeff Law <law@redhat.com>
91 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
92 appearing in DEBUG_INSNs.
94 2017-04-13 Martin Liska <mliska@suse.cz>
97 * gcov-io.c (gcov_write_string): Copy to buffer just when
98 allocated size is greater than zero.
100 2017-04-13 Jakub Jelinek <jakub@redhat.com>
103 * dwarf2out.c (decls_for_scope): Ignore declarations of
104 current_function_decl in BLOCK_NONLOCALIZED_VARS.
106 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
109 * ipa-visibility.c (non_local_p): Fix typos.
110 (localize_node): When localizing symbol in same comdat group,
111 dissolve the group only when we know external symbols are going
113 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
115 2017-04-12 Jakub Jelinek <jakub@redhat.com>
117 PR tree-optimization/79390
118 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
119 order does not result in usable sequence, retry with reversed operand
125 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
126 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
127 op0 instead of fold_convert_loc (loc, type, arg0).
129 2017-04-12 Jeff Law <law@redhat.com>
131 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
132 has a delay slot in the generated code.
134 * config/cris/cris.md (cris_preferred_reload_class): Return
135 GENNONACR_REGS rather than GENERAL_REGS.
137 2017-04-12 Jakub Jelinek <jakub@redhat.com>
140 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
141 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
142 signedness of the result type.
144 2017-04-12 Richard Biener <rguenther@suse.de>
145 Jeff Law <law@redhat.com>
147 PR tree-optimization/80359
148 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
149 trim stores to TARGET_MEM_REFs.
151 2017-04-12 Richard Biener <rguenther@suse.de>
153 PR tree-optimization/79390
154 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
155 threading case even more.
157 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
160 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
161 for quad_address_p for TImode, instead of just not indexed_address.
163 2017-04-12 Richard Biener <rguenther@suse.de>
164 Bernd Edlinger <bernd.edlinger@hotmail.de>
167 * alias.c (component_uses_parent_alias_set_from): Handle
168 TYPE_TYPELESS_STORAGE.
169 (get_alias_set): Likewise.
170 * tree-core.h (tree_type_common): Add typeless_storage flag.
171 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
172 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
173 for types containing members with TYPE_TYPELESS_STORAGE.
174 (place_field): Likewise.
175 (layout_type): Likewise for ARRAY_TYPE.
176 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
177 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
178 TYPE_TYPELESS_STORAGE.
179 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
181 2017-04-12 Jakub Jelinek <jakub@redhat.com>
184 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
185 first argument to type.
187 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
191 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
192 CONST0_RTX (mode) rather than const0_rtx where appropriate.
193 (rs6000_expand_binop_builtin): Likewise.
194 (rs6000_expand_ternop_builtin): Likewise; also add missing
195 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
197 * doc/extend.texi: Document that vec_xxpermdi's third argument
200 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
202 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
203 Use shift_const cost parameter when calculating gain of STV shifts.
205 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
207 PR rtl-optimization/70478
208 * lra-constraints.c (process_alt_operands): Check memory for
209 disfavoring memory insn operand.
211 2017-04-11 Jakub Jelinek <jakub@redhat.com>
214 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
215 left shift in unsigned HOST_WIDE_INT type.
217 PR rtl-optimization/80385
218 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
219 (not (neg X)) into (plus X -1) for complex or non-integral modes.
222 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
223 if they have any depend clauses.
225 2017-04-11 Martin Liska <mliska@suse.cz>
228 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
229 * ipa-split.c (split_function): Create a local comdat symbol
230 if caller is in a comdat group.
232 2017-04-11 Martin Liska <mliska@suse.cz>
235 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
238 2017-04-11 Martin Sebor <msebor@redhat.com>
241 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
242 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
244 (directive::set_width, directive::set_precision, format_character):
246 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
249 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
252 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
253 conflict, set target->arch_name instead of target->cpu_name.
255 2017-04-11 Richard Biener <rguenther@suse.de>
257 PR tree-optimization/80374
258 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
259 build_zero_cst, remove fold_convertible_p check again.
261 2017-04-11 Martin Liska <mliska@suse.cz>
264 * ubsan.c (instrument_object_size): Do not instrument register
267 2017-04-11 Jakub Jelinek <jakub@redhat.com>
270 * config/i386/i386-builtin-types.def
271 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
272 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
273 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
274 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
275 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
276 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
277 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
278 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
279 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
280 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
281 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
282 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
283 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
284 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
285 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
286 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
287 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
288 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
289 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
290 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
291 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
292 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
293 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
294 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
295 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
296 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
297 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
298 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
299 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
300 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
301 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
302 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
303 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
304 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
305 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
306 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
307 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
308 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
309 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
310 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
311 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
312 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
313 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
314 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
315 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
316 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
318 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
319 flag to second_arg_count, handle 4 argument function type _COUNT
320 aliases, handle second_arg_count on second argument rather than last.
322 2017-04-10 Jeff Law <law@redhat.com>
324 PR tree-optimization/80374
325 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
326 record anything if we can not convert integer_zero_node to the
329 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
332 * config/rs6000/rs6000.c (rs6000_option_override_internal):
333 Enhance special handling given to the TARGET_P9_MINMAX option in
334 relation to certain other options.
336 2017-04-10 Bin Cheng <bin.cheng@arm.com>
338 PR tree-optimization/80153
339 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
340 remove POINTER_PLUS_EXPR's base part directly, rather than through
343 2017-04-10 Richard Biener <rguenther@suse.de>
344 Bin Cheng <bin.cheng@arm.com>
346 PR tree-optimization/80153
347 * tree-affine.c (aff_combination_to_tree): Get base pointer from
348 the first element of pointer type aff_tree. Build result expr in
350 (add_elt_to_tree): Convert to type unconditionally. Remove other
352 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
353 (rewrite_use_nonlinear_expr): Check invariant using iv information.
355 2017-04-10 Richard Biener <rguenther@suse.de>
357 * tree-ssa-structalias.c (find_func_aliases): Properly handle
360 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
362 PR rtl-optimization/70478
363 * lra-constraints.c (curr_small_class_check): New.
364 (update_and_check_small_class_inputs): New.
365 (process_alt_operands): Update curr_small_class_check. Disfavor
366 alternative insn memory operands. Check available regs for small
369 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
372 * config/mips/mips.opt (-mvirt): Update description.
373 * doc/invoke.texi (-mvirt): Likewise.
375 2017-04-10 Richard Biener <rguenther@suse.de>
378 * fold-const.c (fold_binary_loc): Look at unstripped ops when
379 looking for NEGATE_EXPR in -A / -B to A / B folding.
381 2017-04-10 Martin Liska <mliska@suse.cz>
383 PR gcov-profile/80224
384 * gcov.c (print_usage): Fix usage string.
385 (get_gcov_intermediate_filename): Remove.
386 (output_gcov_file): Use both for normal and intermediate format.
387 (generate_results): Do not initialize special file for
390 2017-04-10 Richard Biener <rguenther@suse.de>
392 PR tree-optimization/80304
393 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
396 2017-04-10 Nathan Sidwell <nathan@acm.org>
399 * config/rs6000/rs6000.c (rs6000_vector_type): New.
400 (rs6000_init_builtins): Use it.
402 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
404 * config/arm/arm.md (<mrc>): Add mode to SET source.
407 2017-04-10 Richard Biener <rguenther@suse.de>
410 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
412 2017-04-10 Jakub Jelinek <jakub@redhat.com>
415 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
416 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
417 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
418 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
419 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
420 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
421 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
422 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
423 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
424 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
425 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
426 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
427 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
428 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
429 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
430 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
431 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
432 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
433 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
434 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
435 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
436 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
437 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
439 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
441 PR rtl-optimization/70478
442 * lra-constraints.c: Reverse the last patch.
444 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
446 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
447 Add comment for WCHAR_T.
449 2017-04-08 Martin Liska <mliska@suse.cz>
452 2017-04-07 Martin Liska <mliska@suse.cz>
455 * ipa-split.c (split_function): Add function part to a same comdat
458 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
461 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
463 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
465 * rs6000/rs6000.c (vec_load_pendulum): Rename...
466 (vec_pairing): ...to this.
467 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
468 (rs6000_sched_init): Adjust for name change.
469 (struct rs6000_sched_context): Likewise.
470 (rs6000_init_sched_context): Likewise.
471 (rs6000_set_sched_context): Likewise.
473 2017-04-07 Jakub Jelinek <jakub@redhat.com>
479 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
481 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
482 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
483 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
485 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
487 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
489 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
491 PR rtl-optimization/70703
492 * ira-color.c (update_conflict_hard_regno_costs): Use
493 int64_t instead of HOST_WIDE_INT.
495 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
497 PR rtl-optimization/70478
498 * lra-constraints.c (process_alt_operands): Disfavor alternative
499 insn memory operands.
501 2017-04-07 Jeff Law <law@redhat.com>
503 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
504 CALL and NOTE_INSN_CALL_ARG_LOCATION.
506 2017-04-07 Martin Liska <mliska@suse.cz>
509 * config/aarch64/aarch64.c (aarch64_process_target_attr):
510 Show error message instead of an ICE.
512 2017-04-07 Martin Liska <mliska@suse.cz>
515 * ipa-split.c (split_function): Add function part to a same comdat
518 2017-04-07 Richard Biener <rguenther@suse.de>
521 * tree.c (get_unwidened): Also handle ! for_type case for
523 * convert.c (do_narrow): Split out from ...
524 (convert_to_integer_1): ... here. Do not pass final truncation
525 type to get_unwidened for TRUNC_DIV_EXPR.
527 2017-04-07 Richard Biener <rguenther@suse.de>
529 * tree-affine.c (wide_int_ext_for_comb): Take type rather
531 (aff_combination_const): Adjust.
532 (aff_combination_scale): Likewise.
533 (aff_combination_add_elt): Likewise.
534 (aff_combination_add_cst): Likewise.
535 (aff_combination_convert): Likewise.
536 (add_elt_to_tree): Likewise. Remove unused argument.
537 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
539 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
541 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
543 * config/arm/arm.c (arm_default_short_enums): Use
544 ARM_DEFAULT_SHORT_ENUMS.
545 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
547 2017-04-06 Jakub Jelinek <jakub@redhat.com>
550 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
551 members with redundant out-of-class redeclaration.
553 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
556 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
557 * config/i386/i386.md (*zero_extendsidi2):
558 Add (?*x,*x) and (?*v,*v) alternatives.
560 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
563 * config/i386/i386.c (ix86_expand_builtin)
564 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
565 mode from insn data. Convert operands to insn operand mode.
566 Copy operands that don't satisfy insn predicate to a register.
568 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
570 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
573 2017-04-06 Richard Biener <rguenther@suse.de>
575 PR tree-optimization/80334
576 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
577 preserve alignment of accesses.
579 2017-04-06 Richard Biener <rguenther@suse.de>
581 PR tree-optimization/80262
582 * tree-sra.c (build_ref_for_offset): Preserve address-space
584 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
585 Drop useless address-space information on MEM_REF offsets.
587 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
589 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
591 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
593 PR rtl-optimization/70703
594 * ira-color.c (update_conflict_hard_regno_costs): Use
595 HOST_WIDE_INT instead of long.
597 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
600 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
601 not defined for x86_64 target. Add -mmmx target option when __SSE2__
603 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
604 for x86_64 target. Handle -m3dnowa option.
606 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
608 PR rtl-optimization/70703
609 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
610 (update_conflict_hard_regno_costs): Use long instead of unsigned
611 arithmetic for cost calculation.
613 2017-04-05 Jakub Jelinek <jakub@redhat.com>
614 Bernd Edlinger <bernd.edlinger@hotmail.de>
617 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
620 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
623 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
624 ptr_mode with Pmode throughout.
625 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
626 into probe_stack_range and use DImode.
628 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
631 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
632 call_eh_return is true.
634 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
636 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
637 Initialize last_match_fntype_index.
639 2017-04-05 Jakub Jelinek <jakub@redhat.com>
642 * tree-nvr.c: Include internal-fn.h.
643 (pass_return_slot::execute): Ignore internal calls without
646 2017-04-04 Jakub Jelinek <jakub@redhat.com>
647 Richard Biener <rguenther@suse.de>
650 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
651 captures used multiple times, except for the last use.
652 * generic-match-head.c: Include gimplify.h.
654 2017-04-04 Jakub Jelinek <jakub@redhat.com>
656 PR tree-optimization/79390
657 * target.h (struct noce_if_info): Declare.
658 * targhooks.h (default_noce_conversion_profitable_p): Declare.
659 * target.def (noce_conversion_profitable_p): New target hook.
660 * ifcvt.h (struct noce_if_info): New type, moved from ...
661 * ifcvt.c (struct noce_if_info): ... here.
662 (noce_conversion_profitable_p): Renamed to ...
663 (default_noce_conversion_profitable_p): ... this. No longer
665 (noce_try_store_flag_constants, noce_try_addcc,
666 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
667 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
668 instead of noce_conversion_profitable_p.
669 * config/i386/i386.c: Include ifcvt.h.
670 (ix86_option_override_internal): Don't override
671 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
672 (ix86_noce_conversion_profitable_p): New function.
673 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
674 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
675 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
676 * doc/tm.texi: Regenerated.
678 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
680 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
683 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
686 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
687 instructions for small multiply cores.
689 2017-04-04 Jeff Law <law@redhat.com>
691 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
693 (mips_expand_vec_perm_const): Initialize elements in orig_perm
694 that are not set by the loop over the elements.
696 2017-04-04 Jakub Jelinek <jakub@redhat.com>
699 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
700 int mode, convert_modes it to mode as unsigned, otherwise use
701 lowpart_subreg to mode rather than SImode.
702 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
703 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
704 Use DImode instead of SImode for the shift count operand.
705 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
708 2017-04-04 Richard Biener <rguenther@suse.de>
711 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
712 arithmetic done for the negate or the plus. Simplify.
713 (A - (-B) -> A + B): Likewise.
714 * fold-const.c (split_tree): Make sure to not negate pointers.
716 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
718 PR rtl-optimization/60818
719 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
720 a compare of comparisons with the thing compared if this results
721 in a different machine mode.
723 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
725 * alias.c (base_alias_check): Fix typo in comment.
726 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
727 * cgraphunit.c (symbol_table::compile): Likewise.
728 * collect2.c (maybe_run_lto_and_relink): Likewise.
729 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
730 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
731 * config/avr/avr.c (avr_map_op_t): Likewise.
732 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
733 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
734 * config/epiphany/epiphany.md (movcc): Likewise.
735 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
736 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
738 * config/mips/mips.c (mips_save_restore_reg): Likewise.
739 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
740 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
741 * config/sh/sh.c (sh_rtx_costs): Likewise.
742 * fold-const.c (fold_truth_andor): Likewise.
743 * genautomata.c (collapse_flag): Likewise.
744 * gengtype.h (struct type::u::s): Likewise.
745 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
746 * input.c (FORMAT_AMOUNT): Likewise.
747 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
748 (known_aggs_to_agg_replacement_list): Likewise.
749 * ipa-inline-analysis.c: Likewise.
750 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
751 * ipa-polymorphic-call.c
752 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
753 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
754 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
756 * modulo-sched.c (apply_reg_moves): Likewise.
757 * omp-expand.c (build_omp_regions_1): Likewise.
758 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
759 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
760 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
761 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
762 * value-prof.c: Likewise.
763 * var-tracking.c (val_reset): Likewise.
765 2017-04-03 Richard Biener <rguenther@suse.de>
767 PR tree-optimization/80275
768 * fold-const.c (split_address_to_core_and_offset): Handle
771 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
773 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
774 descriptors is at least equal to that of functions.
776 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
778 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
780 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
783 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
784 (mov<IMOD4:mode>): New expander.
785 (*mov<IMOD4:mode>_internal): New insn and split pattern.
787 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
789 PR rtl-optimization/79405
790 * fwprop.c (propagations_left): New variable.
791 (forward_propagate_into): Decrement it.
792 (fwprop_init): Initialize it.
793 (fw_prop): If the variable has reached zero, stop propagating.
794 (fwprop_addr): Ditto.
796 2017-03-31 Jakub Jelinek <jakub@redhat.com>
799 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
800 a FUNCTION_DECL, pass it as decl instead of origin to
803 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
805 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
808 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
811 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
812 TARGET_VSX_SMALL_INTEGER.
814 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
816 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
817 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
819 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
821 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
822 extraction from odd-numbered MSA register.
824 2017-03-31 Jakub Jelinek <jakub@redhat.com>
827 * expmed.c (store_bit_field_1): Don't attempt to create
828 a word subreg out of hard registers wider than word if they
829 have HARD_REGNO_NREGS of 1 for their mode.
832 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
833 conversions to integer types wider than word and pointer.
836 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
837 (rtx_equal_for_cselib_p): Pass 0 to it.
838 * cselib.c (cselib_hasher::equal): Likewise.
839 (rtx_equal_for_cselib_1): Add depth argument. If depth
840 is 128, don't look up VALUE locs and punt. Increment
841 depth in recursive calls when walking VALUE locs.
843 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
845 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
846 (make_gcov_file_name): Use the canonical path name for generating
848 (read_line): Fix handling of files with ascii null bytes.
850 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
852 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
853 to initialise a vector register instead
854 of using a const_int.
856 2017-03-30 Jakub Jelinek <jakub@redhat.com>
859 * gimplify.c (omp_default_clause): Use %qs instead of %s in
862 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
865 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
866 (dfp_diex_<mode>): Update mode of operand 1.
867 * doc/extend.texi (dxex, dxexq): Document change to return type.
868 (diex, diexq): Document change to argument type.
870 2017-03-30 Martin Jambor <mjambor@suse.cz>
873 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
874 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
875 it reflects the signature changes performed at the callee side.
876 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
877 to cgraph_build_function_type_skip_args.
878 (build_function_decl_skip_args): Adjust call to the above function.
880 2017-03-30 Jakub Jelinek <jakub@redhat.com>
884 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
885 register as dest whenever it is a MEM not rtx_equal_p to the
886 corresponding dup operand, and when forcing into reg move the
887 reg into the memory afterwards.
888 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
889 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
890 for the force_reg mode.
891 (avx512vl_vextractf128<mode>): Use register as dest either
892 always when a MEM, or when it is a MEM not rtx_equal_p to the
893 corresponding dup operand, or even not when it is a CONST_VECTOR
894 depending on the mode and lo vs. hi.
895 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
897 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
898 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
899 Likewise. Require that operands[2] is even.
900 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
901 Remove extraneous parens. Require that operands[2] is a multiple
903 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
904 operands[0] is a MEM if <mask_applied>, the predicates/constraints
905 disallow memory then.
907 2017-03-30 Richard Biener <rguenther@suse.de>
909 PR tree-optimization/77498
910 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
911 to non-constants over backedges.
913 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
915 PR rtl-optimization/80233
916 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
917 as last_combined_insn. Do not test for BARRIER_P separately.
919 2017-03-29 Andreas Schwab <schwab@suse.de>
922 * calls.c (prepare_call_address): Convert funexp to Pmode before
925 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
927 PR tree-optimization/80158
928 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
929 Handle possible future case of more than one alternate
931 (replace_rhs_if_not_dup): Likewise.
932 (replace_one_candidate): Likewise.
934 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
936 PR rtl-optimization/80193
937 * ira.c (ira): Do not check allocation for LRA.
939 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
941 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
942 (nvptx_output_simt_exit): Declare.
943 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
944 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
945 (init_softstack_frame): Move initialization of crtl->is_leaf to...
946 (nvptx_declare_function_name): ...here. Emit declaration of local
947 memory space buffer for omp_simt_enter insn.
948 (nvptx_output_unisimt_switch): New.
949 (nvptx_output_softstack_switch): New.
950 (nvptx_output_simt_enter): New.
951 (nvptx_output_simt_exit): New.
952 * config/nvptx/nvptx.h (struct machine_function): New fields
953 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
954 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
955 (UNSPECV_SIMT_EXIT): Ditto.
956 (omp_simt_enter_insn): New insn.
957 (omp_simt_enter): New expansion.
958 (omp_simt_exit): New insn.
959 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
961 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
962 (expand_GOMP_SIMT_ENTER_ALLOC): New.
963 (expand_GOMP_SIMT_EXIT): New.
964 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
965 (GOMP_SIMT_ENTER_ALLOC): Ditto.
966 (GOMP_SIMT_EXIT): Ditto.
967 * target-insns.def (omp_simt_enter): New insn.
968 (omp_simt_exit): Ditto.
969 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
971 (lower_rec_simd_input_clauses): Implement SIMT privatization.
972 (lower_rec_input_clauses): Likewise.
973 (lower_lastprivate_clauses): Handle SIMT privatization.
975 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
976 (ompdevlow_adjust_simt_enter): New.
977 (find_simtpriv_var_op): New.
978 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
979 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
981 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
982 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
983 (copy_decl_for_dup_finish): Ditto.
985 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
987 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
990 * config/i386/i386.c (ix86_option_override_internal): Always
991 allow -mpreferred-stack-boundary=3 for 64-bit targets.
993 2017-03-28 Bin Cheng <bin.cheng@arm.com>
995 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
997 2017-03-28 Bin Cheng <bin.cheng@arm.com>
999 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
1000 mark new edge's irreducible flag accordign to it.
1001 (vect_do_peeling): Check loop preheader edge's irreducible flag
1002 and pass it to function slpeel_add_loop_guard.
1004 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
1006 PR tree-optimization/80218
1007 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
1008 Update block frequencies and counts.
1010 2017-03-28 Richard Biener <rguenther@suse.de>
1012 PR tree-optimization/78644
1013 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
1014 of a simplification result we may not use it at all.
1016 2017-03-28 Richard Biener <rguenther@suse.de>
1019 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
1020 without arguments, generate default definition of a SSA name.
1022 2017-03-28 Richard Biener <rguenther@suse.de>
1025 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
1026 TYPE_REF_CAN_ALIAS_ALL references.
1027 * fold-const.c (fold_indirect_ref_1): Likewise.
1029 2017-03-28 Martin Liska <mliska@suse.cz>
1032 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
1033 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
1035 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
1036 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
1038 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
1039 (EXTRA_SPECS): Define.
1040 (SUBTARGET_EXTRA_SPECS): Likewise.
1041 (SUBTARGET_CPP_SPEC): Likewise.
1042 * config/arc/elf.h (EXTRA_SPECS): Renamed to
1043 SUBTARGET_EXTRA_SPECS.
1044 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
1046 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
1048 * config/arc/simdext.md (vst64_insn): Update pattern.
1049 (vld32wh_insn): Likewise.
1050 (vld32wl_insn): Likewise.
1051 (vld64_insn): Likewise.
1052 (vld32_insn): Likewise.
1054 2017-03-28 Marek Polacek <polacek@redhat.com>
1057 * fold-const.c (fold_comparison): Use protected_set_expr_location
1058 instead of SET_EXPR_LOCATION.
1060 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
1062 * tree.c (add_expr): Avoid name lookup warning.
1064 2017-03-27 Jeff Law <law@redhat.com>
1066 PR tree-optimization/80216
1067 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
1068 function name. Limit recursion depth.
1069 (record_temporary_equivalences): Corresponding changes.
1071 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
1073 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
1076 2017-03-27 Jakub Jelinek <jakub@redhat.com>
1079 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
1081 * cfgcleanup.c (reg_note_cfa_p): New array.
1082 (insns_have_identical_cfa_notes): New function.
1083 (old_insns_match_p): Don't cross-jump in between /f
1084 and non-/f instructions. If both i1 and i2 are frame related,
1085 verify all CFA notes, their order and content.
1087 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1090 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
1091 HImode and SImode with zero extend to DImode to one insn.
1092 (bswap<mode>2_extenddi): Likewise.
1093 (bswapsi2_extenddi): Likewise.
1094 (bswaphi2_extendsi): Likewise.
1095 (bswaphi2): Combine bswap HImode and SImode into one insn.
1096 Separate memory insns from swapping register.
1097 (bswapsi2): Likewise.
1098 (bswap<mode>2): Likewise.
1099 (bswaphi2_internal): Delete, no longer used.
1100 (bswapsi2_internal): Likewise.
1101 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
1102 store, and gpr<-gpr swap insns.
1103 (bswap<mode>2_store): Likewise.
1104 (bswaphi2_reg): Register only splitter, combine with the splitter.
1105 (bswaphi2 splitter): Likewise.
1106 (bswapsi2_reg): Likewise.
1107 (bswapsi2 splitter): Likewise.
1108 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
1109 the insns into load, store, and register/register insns.
1110 (bswapdi2_ldbrx): Likewise.
1111 (bswapdi2_load): Likewise.
1112 (bswapdi2_store): Likewise.
1113 (bswapdi2_reg): Likewise.
1115 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
1117 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
1118 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
1120 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1123 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
1125 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1126 special handling for target option conflicts between dform
1127 options (-mpower9-dform, -mpower9-dform-vector,
1128 -mpower9-dform-scalar) and -mno-direct-move.
1130 2017-03-27 Richard Biener <rguenther@suse.de>
1132 PR tree-optimization/80181
1133 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
1135 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1137 * config/arc/predicates.md (move_double_src_operand): Replace the
1138 call to move_double_src_operand with a call to address_operand.
1140 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1142 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
1143 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
1144 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
1146 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1148 * config/arc/predicates.md (long_immediate_loadstore_operand):
1149 Consider scaled addresses cases.
1151 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1153 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
1154 restored when in interrupt.
1155 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
1156 doesn't have delay slot.
1158 2017-03-27 Richard Biener <rguenther@suse.de>
1161 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
1162 inlined thunk clones.
1164 2017-03-27 Jakub Jelinek <jakub@redhat.com>
1167 * asan.c (instrument_derefs): Copy over last operand from
1168 original COMPONENT_REF to the new COMPONENT_REF with
1169 DECL_BIT_FIELD_REPRESENTATIVE.
1170 * ubsan.c (instrument_object_size): Likewise.
1172 2017-03-27 Richard Biener <rguenther@suse.de>
1174 PR tree-optimization/80170
1175 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
1176 sure DR/SCEV didnt fold in constants we do not see when looking
1177 at the reference base alignment.
1179 2017-03-27 Richard Biener <rguenther@suse.de>
1182 * gimple-fold.c (fold_ctor_reference): Properly guard against
1183 NULL return value from canonicalize_constructor_val.
1185 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
1188 * config/i386/i386.c (ix86_expand_builtin)
1189 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
1190 flags reg setting and flags reg using instructions.
1191 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
1192 clobbering instructions to zero extend op2.
1194 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
1196 * doc/install.texi (Configuration) <--with-aix-soname>:
1197 Update link to AIX ld.
1199 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
1201 PR rtl-optimization/80160
1202 PR rtl-optimization/80159
1203 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
1204 reg_alternate_class into account.
1206 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
1209 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
1210 to consider in curr_insn_transform.
1212 2017-03-24 Jakub Jelinek <jakub@redhat.com>
1214 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
1215 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
1216 and emit_mode_inner.
1218 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1220 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
1221 argument to the overloaded builtin variants. Use the new flag to
1222 deprecate certain builtin variants.
1223 * config/s390/s390-builtin-types.def: Add new builtin types.
1224 * config/s390/s390-builtins.h: Support new flags field for
1225 overloaded builtins.
1226 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
1227 (s390_macro_to_expand): Enable vector float data type.
1228 (s390_cpu_cpp_builtins_internal): Indicate support of the new
1229 builtins by incrementing the __VEC__ version number.
1230 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
1232 (s390_resolve_overloaded_builtin): Emit error messages depending
1233 on the builtin flags.
1234 * config/s390/s390.c (s390_expand_builtin): Support additional
1235 flags argument. Change error message to match the messages
1236 emitted in s390-c.c.
1237 * config/s390/s390.md: New UNSPEC_* constants.
1238 (op_type): Add new instruction types.
1239 * config/s390/vecintrin.h: Add new builtins and test data class
1241 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
1242 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
1243 (VEC_INEXACT, VEC_NOINEXACT): New constants.
1244 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
1245 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
1246 ("vec_mergel<mode>"): V_HW -> VEC_HW.
1248 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
1249 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
1250 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
1251 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
1253 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
1254 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
1255 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
1256 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
1258 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
1259 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
1260 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
1261 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
1262 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
1263 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
1264 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
1266 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
1267 ("vec_scatter_element<V_HW_4:mode>_DI")
1268 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
1269 ("vec_fpint<mode>", "vflls")
1270 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
1271 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
1272 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
1273 ("*vec_cmphe<mode>_cc"): ... these.
1275 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
1276 mode constant instead of magic value.
1278 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1280 * config/s390/s390.c (s390_expand_vec_compare): Support other
1281 vector floating point modes than just V2DF.
1282 (s390_expand_vcond): Likewise.
1283 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
1284 (s390_cannot_change_mode_class): Prevent mode changes between TF
1285 and V1TF in vector registers.
1286 * config/s390/s390.md (DF, SF): New mode attributes.
1287 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
1288 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
1289 SFmode support for VRs.
1290 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
1292 (VFT, VF_HW): New mode iterators.
1293 (vw, sdx): New mode attributes.
1294 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
1295 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
1296 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
1297 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
1298 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
1299 also the new vector floating point modes. Renaming to ...
1301 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
1302 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
1303 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
1304 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
1305 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
1306 ("vec_unordered<mode>"): ... these.
1308 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
1309 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
1310 ("*vec_extendv2df"): New insn definitions.
1312 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1314 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
1315 ("mulditi3_2", "*muldi3_sign"): New patterns.
1316 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
1317 rename the pattern definition.
1319 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1321 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
1323 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
1325 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1327 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
1328 instruction if possible.
1329 * config/s390/vector.md (vec_halfnumelts): New mode
1331 ("*vec_vllezlf<mode>"): New pattern.
1333 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1335 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
1336 ("popcountv4si2", "popcountv2di2"): Rename to ...
1337 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
1338 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
1340 ("popcount<mode>2_vxe"): New pattern.
1342 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1344 * common/config/s390/s390-common.c (processor_flags_table): Add
1346 * config.gcc: Add arch12.
1347 * config/s390/driver-native.c (s390_host_detect_local_cpu):
1348 Default to arch12 for unknown CPU model numbers.
1349 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
1350 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
1351 PROCESSOR_max sanity check.
1352 * config/s390/s390-opts.h (enum processor_type): Add
1354 * config/s390/s390.c (processor_table): Add arch12.
1355 (s390_expand_builtin): Add check for B_VXE flag.
1356 (s390_issue_rate): Add PROCESSOR_ARCH12.
1357 (s390_get_sched_attrmask): Likewise.
1358 (s390_get_unit_mask): Likewise.
1359 (s390_sched_score): Enable z13 scheduling for arch12.
1360 (s390_sched_reorder): Likewise.
1361 (s390_sched_variable_issue): Likewise.
1362 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
1364 (s390_tune_attr): Use z13 scheduling also for arch12.
1365 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
1366 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
1367 (TARGET_VXE_P): New macros.
1368 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
1369 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
1370 * config/s390/s390.opt: Add arch12 as processor_type.
1372 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1374 * config/s390/s390.md
1375 ("fixuns_truncdddi2", "fixuns_trunctddi2")
1376 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
1377 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
1379 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
1380 Rename expanders to ...
1382 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
1383 ("fixuns_truncdddi2_emu"): ... these.
1385 ("fixuns_trunc<mode>si2_emu"): New expander.
1387 ("*fixuns_truncdfdi2_z13"): Rename to ...
1388 ("*fixuns_truncdfdi2_vx"): ... this.
1390 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1392 * config/s390/2964.md: Remove the single element vector compare
1393 instructions which are no longer used.
1394 * config/s390/s390.c (s390_select_ccmode): Remove handling of
1396 (s390_canonicalize_comparison): Remove handling of DFmode
1398 (s390_expand_vec_compare_scalar): Remove function.
1399 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
1400 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
1402 ("*cmp<mode>_ccs"): Add wfcdb instruction.
1404 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1406 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
1408 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
1409 will anyway by matched by mov<mode>_64dfp.
1411 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1413 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
1414 vlef/vstef. Add missing operand to vleif.
1416 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1418 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
1419 pair for all vector types with 64 bit elements.
1420 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
1421 * config/s390/vector.md (V_HW_64): ... here.
1422 (V_128_NOSINGLE): New mode iterator.
1423 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
1424 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
1425 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
1426 ("*vec_load_pairv2di"): Change to ...
1427 ("*vec_load_pair<mode>"): ... this one.
1429 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1431 * config/s390/constraints.md: Add comments.
1432 (jKK): Reject element sizes > 8 bytes.
1433 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
1435 * config/s390/s390.md: Add the s_operand checks formerly in
1436 s390_split_ok_p to various splitters where they are still
1438 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
1439 for 128 bit vectors. Plus two splitters.
1441 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1443 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
1446 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1449 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
1450 error if the boundary argument is not constant.
1452 2017-03-24 Jakub Jelinek <jakub@redhat.com>
1454 PR rtl-optimization/80112
1455 * loop-doloop.c (doloop_condition_get): Don't check condition
1456 if cmp isn't SET with IF_THEN_ELSE src.
1458 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1460 PR tree-optimization/80158
1461 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
1462 replacing a candidate statement, also replace it for the
1463 candidate's alternate interpretation.
1464 (replace_rhs_if_not_dup): Likewise.
1465 (replace_one_candidate): Likewise.
1467 2017-03-24 Richard Biener <rguenther@suse.de>
1469 PR tree-optimization/80167
1470 * graphite-isl-ast-to-gimple.c
1471 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
1473 (translate_isl_ast_to_gimple::get_rename): Likewise.
1475 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
1477 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
1478 handling of certain combinations of target options, including the
1479 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
1480 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
1482 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1485 * config/arm/arm.md (*load_multiple): Add reload_completed to
1488 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1489 Richard Biener <rguenth@suse.de>
1491 PR tree-optimization/79908
1492 PR tree-optimization/80136
1493 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
1494 been cast away, gimplify_and_add suffices.
1496 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
1498 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
1500 2017-03-23 Richard Biener <rguenther@suse.de>
1502 PR tree-optimization/80032
1503 * gimplify.c (gimple_push_cleanup): Forced unconditional
1504 cleanups still have to go to the conditional_cleanups
1507 2017-03-22 Jakub Jelinek <jakub@redhat.com>
1509 PR tree-optimization/80072
1510 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
1512 (next_operand_entry_id): Change type to unsigned int.
1513 (sort_by_operand_rank): Make sure to return the right return value
1514 even if unsigned fields are bigger than INT_MAX.
1515 (struct oecount): Change cnt and id type to unsigned int.
1516 (oecount_hasher::equal): Formatting fix.
1517 (oecount_cmp): Make sure to return the right return value
1518 even if unsigned fields are bigger than INT_MAX.
1519 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
1522 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
1523 TREE_READONLY on result if writing it more than once.
1526 * doc/invoke.texi (-fsanitize=thread): Document that with
1527 -fnon-call-exceptions atomics are not able to throw
1531 * tsan.c: Include tree-eh.h.
1532 (instrument_builtin_call): Call maybe_clean_eh_stmt or
1533 maybe_clean_or_replace_eh_stmt where needed.
1534 (instrument_memory_accesses): Add cfg_changed argument.
1535 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
1536 if it returned true.
1537 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
1539 PR rtl-optimization/63191
1540 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
1541 wrapper function, moved the whole old content into ...
1542 (ix86_delegitimize_address_1): ... this. New inline function.
1543 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
1544 true as last argument instead of ix86_delegitimize_address.
1546 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
1548 * config/aarch64/aarch64.c (generic_branch_cost): Copy
1549 cortexa57_branch_cost.
1551 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
1553 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
1555 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1558 * doc/md.texi (Constraints): Document wA constraint.
1559 * config/rs6000/constraints.md (wA): New.
1560 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
1561 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
1562 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
1563 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
1565 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
1568 * gimplify.c (is_oacc_declared): New function.
1569 (oacc_default_clause): Use it to set default flags for acc declared
1570 variables inside parallel regions.
1571 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
1573 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
1574 declare attribute to any decl as necessary.
1576 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1579 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
1580 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
1581 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
1582 (arm_arch_lpae): This.
1583 * config/arm/arm.c (arm_arch7ve): Rename into ...
1584 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
1585 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
1588 2017-03-22 Martin Liska <mliska@suse.cz>
1591 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
1592 error message instead of an ICE.
1594 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1596 * doc/extend.texi (6.11 Additional Floating Types): Revise.
1598 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
1600 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
1602 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1605 2017-03-21 Martin Sebor <msebor@redhat.com>
1607 * doc/extend.texi: Use "cannot" instead of "can't."
1608 * doc/hostconfig.texi: Same.
1609 * doc/install.texi: Same.
1610 * doc/invoke.texi: Same.
1611 * doc/loop.texi: Same.
1612 * doc/md.texi: Same.
1613 * doc/objc.texi: Same.
1614 * doc/rtl.texi: Same.
1615 * doc/tm.texi: Same.
1616 * doc/tm.texi.in: Same.
1617 * doc/trouble.texi: Same.
1619 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
1622 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
1623 (collect_checksum_attributes): Set it.
1624 (die_checksum_ordered): Use it.
1626 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1628 PR tree-optimization/79908
1629 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
1630 change: For a VA_ARG whose LHS has been cast away, use
1631 force_gimple_operand to construct the side effects.
1633 2017-03-21 David Malcolm <dmalcolm@redhat.com>
1635 PR translation/80001
1636 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
1637 more amenable to translation.
1638 (oacc_loop_auto_partitions): Likewise.
1640 2017-03-21 Marek Polacek <polacek@redhat.com>
1641 Martin Sebor <msebor@redhat.com>
1643 PR tree-optimization/80109
1644 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
1647 2017-03-21 Jakub Jelinek <jakub@redhat.com>
1648 Segher Boessenkool <segher@kernel.crashing.org>
1651 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
1652 check reg_used_between_p between insn and one of succ or succ2
1653 depending on if succ is artificial insn not inserted into insn
1656 2017-03-21 Martin Liska <mliska@suse.cz>
1658 PR gcov-profile/80081
1659 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
1660 * doc/gcc.texi: Include gcov-dump stuff.
1661 * doc/gcov-dump.texi: New file.
1663 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
1665 PR rtl-optimization/79150
1666 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
1667 conditional jump, if the jump is the last insn of the loop.
1669 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1670 Richard Biener <rguenth@suse.de>
1672 PR tree-optimization/79908
1673 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
1674 been cast away, use force_gimple_operand to construct the side
1677 2017-03-21 Martin Liska <mliska@suse.cz>
1680 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
1683 2017-03-21 Brad Spengler <spender@grsecurity.net>
1686 * plugin.c (htab_hash_plugin): New function.
1687 (add_new_plugin): Use it and adjust.
1688 (parse_plugin_arg_opt): Adjust.
1689 (init_one_plugin): Likewise.
1691 2017-03-21 Richard Biener <rguenther@suse.de>
1693 PR tree-optimization/80032
1694 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
1695 if set force the cleanup to happen unconditionally.
1696 (gimplify_target_expr): Push inserted clobbers with force_uncond
1697 to avoid them being removed by control-dependent DCE.
1699 2017-03-21 Richard Biener <rguenther@suse.de>
1701 PR tree-optimization/80122
1702 * tree-inline.c (copy_bb): Do not expans va-arg packs or
1703 va_arg_pack_len when the inlined call stmt requires pack
1705 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
1707 2017-03-21 Jakub Jelinek <jakub@redhat.com>
1710 * tsan.c (instrument_builtin_call): If the memory model argument
1711 is not a constant, assume it is valid.
1714 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
1717 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
1719 PR rtl-optimization/79910
1720 * combine.c (can_combine_p): Do not allow combining an I0 or I1
1721 if its dest is used by an insn before I2 (other than the combined
1722 insns themselves, which are properly handled already).
1724 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
1727 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
1729 * combine.c (record_used_regs): New static function.
1730 (try_combine): Handle situations where there is an additional
1731 instruction between I2 and I3 which needs to have a LOG_LINK
1735 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
1737 * combine.c (try_combine): Delete redundant i1 test. Call
1738 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
1740 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1743 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
1746 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1748 PR tree-optimization/80054
1749 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
1750 the optimization if a PHI or any of its arguments is not dominated
1751 by the candidate's basis. Use gphi* rather than gimple* as
1753 (replace_profitable_candidates): Clean up a gimple* variable that
1754 should be a gphi* variable.
1756 2017-03-20 Martin Sebor <msebor@redhat.com>
1759 * doc/extend.texi (attribute constructor): Document present limitation.
1761 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
1764 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
1765 __POWER9_VECTOR__ #ifdef control, change template definition to
1766 use Power9-specific built-in function.
1767 (vec_any_eq): Likewise.
1768 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
1769 to control outcomes from this test.
1770 (vector_ae_<mode>p): For VEC_F modes, likewise.
1772 2017-03-20 Ian Lance Taylor <iant@google.com>
1774 * config/i386/i386.c (ix86_function_regparm): Save an extra
1775 register for -fsplit-stack with DECL_STATIC_CHAIN.
1777 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
1780 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
1781 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
1783 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
1785 * config/riscv/riscv.c (riscv_print_operand): Use "fence
1787 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
1790 2017-03-20 Marek Polacek <polacek@redhat.com>
1793 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
1795 2017-03-20 Richard Biener <rguenther@suse.de>
1797 PR tree-optimization/80113
1798 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
1799 allocate extra SSA name for PHI def.
1800 (add_close_phis_to_outer_loops): Likewise.
1801 (add_close_phis_to_merge_points): Likewise.
1802 (copy_loop_close_phi_args): Likewise.
1803 (copy_cond_phi_nodes): Likewise.
1805 2017-03-20 Martin Liska <mliska@suse.cz>
1808 * tree-chkp.c (chkp_build_returned_bound): Do not build
1809 returned bounds for a LHS that's not a BOUNDED_P type.
1811 2017-03-20 Martin Liska <mliska@suse.cz>
1815 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
1816 COMPLEX_CST and VECTOR_CST.
1818 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1821 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
1822 target operand. A new splitter adds the clobber statement in case
1823 the target operand is dead anyway.
1825 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
1827 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
1828 to age-old versions of binutils and glibc.
1830 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
1832 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
1834 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
1836 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
1838 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
1840 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
1841 requirement for binutils 2.13.
1843 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
1845 * combine.c (try_combine): Delete redundant i1 test. Call
1846 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
1848 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
1850 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
1851 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
1853 <riscv64-*-elf>: Re-arrange section
1854 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
1855 <riscv32-*-linux>: Likewise.
1856 <riscv64-*-elf>: Likewise
1857 <riscv64-*-linux>: Likewise.
1859 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
1862 * aarch64.opt(verbose-cost-dump): Fix typo.
1864 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
1867 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
1868 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
1870 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
1872 * reload.c (find_reloads): When reloading a nonoffsettable address,
1873 use RELOAD_OTHER for it and its address reloads.
1875 PR rtl-optimization/79910
1876 * combine.c (record_used_regs): New static function.
1877 (try_combine): Handle situations where there is an additional
1878 instruction between I2 and I3 which needs to have a LOG_LINK
1881 2017-03-17 Jeff Law <law@redhat.com>
1883 PR tree-optimization/71437
1884 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
1885 conditional in the hash table first.
1886 (vrp_dom_walker::before_dom_children): Extract condition from
1887 ASSERT_EXPR. Record condition, its inverion and any implied
1890 2017-03-17 Marek Polacek <polacek@redhat.com>
1891 Markus Trippelsdorf <markus@trippelsdorf.de>
1893 PR tree-optimization/80079
1894 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
1897 2017-03-17 Richard Biener <rguenther@suse.de>
1900 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
1901 Properly verify the LHS before the RHS possibly claims to be
1903 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
1906 2017-03-17 Martin Jambor <mjambor@suse.cz>
1908 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
1909 (List of -O2 options): Likewise.
1910 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
1913 2017-03-17 Tom de Vries <tom@codesourcery.com>
1915 * gcov-dump.c (print_usage): Print bug_report_url.
1917 2017-03-17 Richard Biener <rguenther@suse.de>
1920 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
1921 (parser::peek): Likewise.
1923 2017-03-17 Richard Biener <rguenther@suse.de>
1925 PR tree-optimization/80048
1926 * sese.c (free_sese_info): Properly release rename_map and
1927 copied_bb_map elements.
1929 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
1931 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
1932 Add linked-list forward and backlinks. Insert on
1933 construction, remove on destruction.
1934 (class pass_store_merging): Add m_stores_head field.
1935 (pass_store_merging::terminate_and_process_all_chains):
1936 Iterate over m_stores_head list.
1937 (pass_store_merging::terminate_all_aliasing_chains):
1939 (pass_store_merging::execute): Check for debug stmts first.
1940 Push new chains onto the m_stores_head stack.
1942 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
1945 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
1946 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
1947 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
1949 2017-03-16 Jeff Law <law@redhat.com>
1951 PR tree-optimization/71437
1952 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
1953 member function. Implementation moved into after_dom_children
1954 member function and into the threader's thread_outgoing_edges
1956 (dom_opt_dom_walker::after_dom_children): Simplify by moving
1957 some code into new thread_outgoing_edges.
1958 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
1959 definition. Simplify marker handling (do it here). Assume we always
1960 have the available expression and the const/copies tables.
1961 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
1963 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
1964 * tree-vrp.c (equiv_stack): No longer file scoped.
1965 (vrp_dom_walker): New class.
1966 (vrp_dom_walker::before_dom_children): New member function.
1967 (vrp_dom_walker::after_dom_children): Likewise.
1968 (identify_jump_threads): Setup domwalker. Use it rather than
1969 walking edges in a random order by hand. Simplify setup/finalization.
1970 (finalize_jump_threads): Remove.
1971 (vrp_finalize): Do not call identify_jump_threads here.
1972 (execute_vrp): Do it here instead and call thread_through_all_blocks
1975 PR tree-optimization/71437
1976 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
1978 (simplify_stmt_for_jump_threading): Add basic_block argument. All
1980 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
1981 (dom_opt_dom_walker::thread_across_edge): Remove
1982 handle_dominating_asserts argument. All callers changed.
1983 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
1984 changes. Remove calls to lhs_of_dominating_assert. Other
1985 uses of handle_dominating_asserts turn into unconditional code
1986 (simplify_control_stmt_condition_1): Likewise.
1987 (simplify_control_stmt_condition): Likewise.
1988 (thread_through_normal_block, thread_across_edge): Likewise.
1989 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
1990 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
1991 object if it is not an SSA_NAME.
1992 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
1993 before calling into the VRP specific simplifiers.
1994 (identify_jump_threads): Remove handle_dominating_asserts
1997 2017-03-16 Jakub Jelinek <jakub@redhat.com>
2000 * tree-diagnostic.c (default_tree_printer): No longer static.
2001 * tree-diagnostic.h (default_tree_printer): New prototype.
2003 2017-03-16 Tamar Christina <tamar.christina@arm.com>
2005 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
2006 Change ins into fmov.
2008 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2010 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
2011 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
2012 Use h_con constraint for operand 1.
2013 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
2014 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
2016 2017-03-15 Jeff Law <law@redhat.com>
2018 PR tree-optimization/71437
2019 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
2020 (record_temporary_equivalences): Use it.
2022 PR tree-optimization/71437
2023 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
2024 tree-ssa-scopedtables.
2025 (lookup_avail_expr, build_and_record_new_cond): Likewise.
2026 (record_conditions, record_cond, vuse_eq): Likewise.
2027 (record_edge_info): Adjust to API tweak of record_conditions.
2028 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
2029 (record_temporary_equivalences, optimize_stmt): Likewise.
2030 (eliminate_redundant_computations): Likewise.
2031 (record_equivalences_from_stmt): Likewise.
2032 * tree-ssa-scopedtables.c: Include options.h and params.h.
2033 (vuse_eq): New function, moved from tree-ssa-dom.c
2034 (build_and_record_new_cond): Likewise.
2035 (record_conditions): Likewise. Accept vector of conditions rather
2036 than edge_equivalence structure for first argument.
2037 for the first argument.
2038 (avail_exprs_stack::lookup_avail_expr): New member function, moved
2039 from tree-ssa-dom.c.
2040 (avail_exprs_stack::record_cond): Likewise.
2041 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
2042 from tree-ssa-dom.c.
2043 (avail_exprs_stack): Add new member functions lookup_avail_expr
2045 (record_conditions): Declare.
2047 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
2050 * lra-constraints.c (process_alt_operands): Increase reject for
2051 reloading an input/output operand.
2053 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2056 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
2057 insns to convert from signed/unsigned char/short to IEEE 128-bit
2059 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
2061 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
2064 * config/i386/i386.c (ix86_vector_duplicate_value): Create
2065 subreg of inner mode for values already in registers.
2067 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
2069 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
2070 iteration reg is used after the loop.
2072 2017-03-14 Martin Sebor <msebor@redhat.com>
2074 PR tree-optimization/79800
2075 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
2076 precision in negative-positive range.
2077 (format_floating): Call non-const overload with adjusted precision.
2079 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2082 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
2085 2017-03-14 Martin Sebor <msebor@redhat.com>
2088 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
2089 * builtins.def (aligned_alloc): Use it.
2092 * Makefile.in (GTFILES): Add calls.c.
2093 * calls.c: Include "gt-calls.h".
2095 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
2097 PR rtl-optimization/79728
2098 * regs.h (struct target_regs): New field
2099 x_contains_allocatable_regs_of_mode.
2100 (contains_allocatable_regs_of_mode): New macro.
2101 * reginfo.c (init_reg_sets_1): Initialize it, and change
2102 contains_reg_of_mode so it includes global regs as well.
2103 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
2104 rather than contains_regs_of_mode.
2106 2017-03-14 Martin Liska <mliska@suse.cz>
2108 * doc/invoke.texi: Document options that can't be combined with
2109 -fcheck-pointer-bounds.
2111 2017-03-14 Martin Liska <mliska@suse.cz>
2114 * doc/invoke.texi (-Wchkp): Document the option.
2116 2017-03-14 Martin Liska <mliska@suse.cz>
2118 * Makefile.in: Install gcov-dump.
2120 2017-03-14 Martin Liska <mliska@suse.cz>
2122 * multiple_target.c (expand_target_clones): Bail out for
2123 an invalid attribute.
2125 2017-03-14 Richard Biener <rguenther@suse.de>
2127 * alias.c (struct alias_set_entry): Pack properly.
2128 * cfgloop.h (struct loop): Likewise.
2129 * cse.c (struct set): Likewise.
2130 * ipa-utils.c (struct searchc_env): Likewise.
2131 * loop-invariant.c (struct invariant): Likewise.
2132 * lra-remat.c (struct cand): Likewise.
2133 * recog.c (struct change_t): Likewise.
2134 * rtl.h (struct address_info): Likewise.
2135 * symbol-summary.h (function_summary): Likewise.
2136 * tree-loop-distribution.c (struct partition): Likewise.
2137 * tree-object-size.c (struct object_size_info): Likewise.
2138 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
2139 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
2140 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
2141 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
2142 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
2143 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
2144 (struct _stmt_vec_info): Likewise.
2146 2017-03-14 Martin Liska <mliska@suse.cz>
2149 * multiple_target.c (create_dispatcher_calls): Check that
2150 a target can create a function dispatcher.
2152 2017-03-14 Martin Liska <mliska@suse.cz>
2155 * multiple_target.c (expand_target_clones): Drop local.local
2156 flag for default implementation.
2158 2017-03-14 Richard Biener <rguenther@suse.de>
2160 PR tree-optimization/80030
2161 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
2163 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
2165 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
2166 gcc_fallthrough() instead of __attribute__((fallthrough));
2168 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
2170 * doc/gcc.texi: Remove "up" link to (DIR).
2171 * doc/gccint.texi: Ditto.
2173 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
2175 * doc/install.texi (Specific) <avr>: Remove reference to
2178 2017-03-13 Jeff Law <law@redhat.com>
2180 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
2181 attribute rather than comments.
2183 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
2184 match_scratch operand is highest.
2186 2017-03-13 Martin Liska <mliska@suse.cz>
2189 * ipa-pure-const.c (warn_function_noreturn): If the declarations
2190 is a CHKP clone, use original declaration.
2192 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2194 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
2195 (arc_conditional_register_usage): Use a different allocation order
2196 when optimizing for size.
2197 * common/config/arc/arc-common.c (arc_option_optimization_table):
2198 Section anchors default on when optimizing for size.
2200 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2202 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
2204 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2206 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
2207 * config/arc/arc.md (cpu_facility): Add cd variant.
2208 (*movqi_insn): Add code density variant.
2209 (*movhi_insn): Likewise.
2210 (*movqi_insn): Likewise.
2211 (*addsi3_mixed): Likewise.
2212 (subsi3_insn): Likewise.
2214 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2216 * config/arc/arc.md (movsi_cond_exec): Update constraint.
2218 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2220 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
2221 expressions with MINUS and UNARY ops.
2223 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2226 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
2228 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
2229 between vec_select and vector argument.
2230 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
2231 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
2232 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
2233 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
2234 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
2235 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
2237 2017-03-13 Richard Biener <rguenther@suse.de>
2240 * params.def (vect-max-peeling-for-alignment): Fix typo.
2242 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2244 * doc/install.texi (Specific) <mips-*-*>: Remove description of
2245 issue that only occurred with binutils below 2.18.
2247 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2249 * doc/install.texi (Specific) <cris-axis-elf>: No longer
2250 refer to binutils 2.11/2.12 minimum.
2252 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2254 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
2255 ftp.kernel.org and simplify binutils requirement.
2257 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
2259 * doc/invoke.texi (Warning Options): Fix spelling of link-time
2261 (Optimize Options): Ditto. Also remove redundancy.
2263 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2265 PR translation/79848
2266 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
2268 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
2269 to G_ to avoid double translation.
2271 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2273 PR translation/79923
2274 * auto-profile.c (get_combined_location): Convert leading
2275 character of diagnostics to lower case and remove trailing period.
2276 (read_profile): Likewise for various diagnostics.
2277 * config/arm/arm.c (arm_option_override): Remove trailing period
2278 from various diagnostics.
2279 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
2280 (msp430_expand_delay_cycles): Likewise.
2282 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2285 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
2286 full command-line argument, rather than just "str".
2287 (aarch64_validate_march): Likewise.
2288 (aarch64_validate_mtune): Likewise.
2290 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
2292 PR rtl-optimization/78911
2293 * lra-assigns.c (must_not_spill_p): New function.
2294 (spill_for): Use it.
2296 2017-03-10 Jakub Jelinek <jakub@redhat.com>
2298 PR tree-optimization/79981
2299 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
2300 ATOMIC_COMPARE_EXCHANGE ifn result.
2301 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
2302 IFN_ATOMIC_COMPARE_EXCHANGE.
2304 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2307 * opts.c (parse_sanitizer_options): Add missing question mark to
2308 "did you mean" message.
2310 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2312 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
2314 (VMULEUH_UNS): Likewise.
2315 (VMULOUB_UNS): Likewise.
2316 (VMULOUH_UNS): Likewise.
2317 * config/rs6000/rs6000.c (builtin_function_type): Remove
2318 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
2320 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2323 * read-rtl-function.c (function_reader::read_rtx_operand): Update
2324 x with result of extra_parsing_for_operand_code_0.
2325 (function_reader::extra_parsing_for_operand_code_0): Convert
2326 return type from void to rtx, returning x. When reading
2327 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
2328 larger size containing struct block_symbol.
2330 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
2332 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
2333 -mfloat128-hardware without -m64.
2335 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
2338 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
2339 entries to the case statement that marks unsigned arguments to
2340 overloaded functions.
2342 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
2344 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
2345 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
2347 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
2350 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
2351 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
2353 2017-03-10 Martin Liska <mliska@suse.cz>
2357 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
2358 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
2361 2017-03-10 Olivier Hainque <hainque@adacore.com>
2363 * tree-switch-conversion (array_value_type): Start by resetting
2364 candidate type to it's main variant.
2366 2017-03-10 Jakub Jelinek <jakub@redhat.com>
2368 PR rtl-optimization/79909
2369 * combine.c (try_combine): Use simplify_replace_rtx on individual
2370 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
2371 of the whole CALL_INSN_FUNCTION_USAGE.
2373 PR tree-optimization/79972
2374 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
2375 get_range_info on SSA_NAMEs. Formatting fixes.
2377 2017-03-10 Richard Biener <rguenther@suse.de>
2378 Jakub Jelinek <jakub@redhat.com>
2380 PR tree-optimization/77975
2381 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
2382 edge to be constant.
2383 (get_val_for): For constant x return it. Formatting fix.
2384 (loop_niter_by_eval): Avoid pointless looping if the next iteration
2385 would use the same bases as the current one.
2387 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2389 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
2390 instead of vec_select for V1TImode.
2391 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
2393 (VSX_LE_128): Add V1TI to this mode iterator.
2394 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
2395 (*vsx_le_perm_store_<mode>): Likewise.
2396 (pre-reload splitter for VSX stores): Likewise.
2397 (post-reload splitter for VSX stores): Likewise.
2398 (*vsx_xxpermdi2_le_<mode>): Likewise.
2399 (*vsx_lxvd2x2_le_<mode>): Likewise.
2400 (*vsx_stxvd2x2_le_<mode>): Likewise.
2402 2017-03-09 Michael Eager <eager@eagercon.com>
2404 Correct failures with --enable-checking=yes,rtl.
2406 * config/microblaze/microblaze.c (microblaze_expand_shift):
2407 Replace GET_CODE test with CONST_INT_P and INTVAL test with
2408 test for const0_rtx.
2409 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
2410 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
2412 2017-03-09 Richard Biener <rguenther@suse.de>
2414 PR tree-optimization/79977
2415 * graphite-scop-detection.c (scop_detection::merge_sese):
2416 Handle the case of extra exits to blocks dominating the entry.
2418 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
2420 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2423 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
2425 PR rtl-optimization/79949
2426 * lra-constraints.c (process_alt_operands): Check memory when
2427 trying to predict a cycle. Print about the overall increase.
2429 2017-03-09 Richard Biener <rguenther@suse.de>
2432 * gimple-expr.c (useless_type_conversion_p): Preserve
2433 TYPE_SATURATING for fixed-point types.
2435 2017-03-09 Richard Biener <rguenther@suse.de>
2438 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
2439 alignment of BLKmode params.
2441 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2444 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
2445 (VALL_NO_V2Q): Likewise.
2447 * config/aarch64/aarch64-simd.md
2448 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
2450 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
2451 VALL_NO_V2Q mode iterator.
2452 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
2454 2017-03-09 Martin Liska <mliska@suse.cz>
2456 PR tree-optimization/79631
2457 * tree-chkp-opt.c (chkp_is_constant_addr): Call
2458 tree_int_cst_sign_bit just for INTEGER constants.
2460 2017-03-09 Martin Liska <mliska@suse.cz>
2464 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
2467 2017-03-09 Marek Polacek <polacek@redhat.com>
2470 * tree.c (inchash::add_expr): Handle TREE_VEC.
2472 2017-03-09 Martin Liska <mliska@suse.cz>
2475 (chkp_narrow_size_and_offset): New function.
2476 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
2477 (void chkp_parse_bit_field_ref): New function.
2478 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
2479 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
2481 2017-03-09 Martin Liska <mliska@suse.cz>
2484 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
2485 (chkp_find_bounds_1): Remove gcc_unreachable.
2487 2017-03-09 Jakub Jelinek <jakub@redhat.com>
2490 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
2491 BUILT_IN_SYNC*, determine the access type from the size suffix and
2492 always build a MEM_REF with that type. Handle forgotten
2493 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
2496 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
2497 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
2498 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
2499 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
2500 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
2501 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
2502 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
2503 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
2504 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
2505 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
2506 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
2507 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
2508 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
2509 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
2510 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
2511 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
2512 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
2513 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
2514 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
2515 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
2516 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
2517 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
2518 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
2519 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
2520 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
2521 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
2522 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
2523 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
2524 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
2525 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
2526 definitions outside of __OPTIMIZE__ guarded section.
2529 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
2530 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
2531 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
2532 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
2535 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2537 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
2538 ("vfenez<mode>"): Add missing constraints.
2540 2017-03-08 Martin Sebor <msebor@redhat.com>
2543 * config/nds32/nds32.c (nds32_option_override):
2544 Fix misspelled diagnostic.
2546 2017-03-08 Jakub Jelinek <jakub@redhat.com>
2549 * gimplify.c (gimplify_omp_for): Replace index var in outer
2550 taskloop statement with an artificial variable and add
2551 OMP_CLAUSE_PRIVATE clause for it.
2553 2017-03-08 Richard Biener <rguenther@suse.de>
2555 PR tree-optimization/79955
2556 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
2557 for accesses that are completely outside of the variable.
2559 2017-03-08 Andrew Haley <aph@redhat.com>
2561 PR tree-optimization/79943
2562 * tree-ssa-loop-split.c (compute_new_first_bound): When
2563 calculating the new upper bound, (END-BEG) should be added, not
2566 2017-03-08 Jakub Jelinek <jakub@redhat.com>
2568 * config/avr/avr.md (setmemhi): Make sure match_dup
2569 operand number comes before match_scratch.
2571 2017-03-08 Richard Biener <rguenther@suse.de>
2573 PR tree-optimization/79920
2574 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
2575 with ncopies == 1 to ...
2576 (vect_transform_slp_perm_load): ... here. Properly compute
2577 all element loads by iterating VF times over the group. Do
2578 not handle ncopies (computed in a broken way) in
2579 vect_create_mask_and_perm.
2581 2017-03-08 Jakub Jelinek <jakub@redhat.com>
2584 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
2585 is a uniform vector, use uniform_vector_p return value instead of
2586 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
2588 2017-03-07 Marek Polacek <polacek@redhat.com>
2591 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
2592 (alloca_call_type): Likewise.
2594 2017-03-07 Martin Liska <mliska@suse.cz>
2596 * gcov.c (process_args): Put comment to correct location.
2598 2017-03-07 Martin Liska <mliska@suse.cz>
2601 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
2602 Use array_at_struct_end_p instead of DECL_CHAIN (field).
2603 (chkp_narrow_bounds_for_field): Likewise.
2604 (chkp_parse_array_and_component_ref): Pass one more argument to
2607 2017-03-07 Richard Biener <rguenther@suse.de>
2609 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
2612 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
2614 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
2615 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
2617 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2620 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
2621 to end of description.
2622 (PARAM_MAX_STORES_TO_MERGE): Likewise.
2624 2017-03-07 Jakub Jelinek <jakub@redhat.com>
2626 PR rtl-optimization/79901
2627 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
2629 (*avx512f_<code><mode>3<mask_name>): ... this.
2630 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
2631 iterator instead of VI8_AVX2_AVX512BW.
2633 PR rtl-optimization/79901
2634 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
2635 min/max expander, expand it using expand_vec_cond_expr.
2638 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
2641 2017-03-06 Jakub Jelinek <jakub@redhat.com>
2644 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
2645 to void * for PCH reasons.
2646 * dwarf2out.c (output_loc_operands, output_die): Cast
2647 v.val_vec.array to unsigned char *.
2649 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
2652 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
2655 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
2657 PR rtl-optimization/79571
2658 * lra-constraints.c (process_alt_operands): Calculate static
2659 reject and subtract it from overall when only addresses will be
2662 2017-03-06 Julia Koval <julia.koval@intel.com>
2665 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
2666 incoming stack boundary to 128 for 64-bit targets.
2668 2017-03-06 Richard Biener <rguenther@suse.de>
2670 PR tree-optimization/79894
2671 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
2672 to NULL after folding it.
2674 2017-03-06 Richard Biener <rguenther@suse.de>
2676 PR tree-optimization/79824
2677 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
2678 check disabling peeling for gaps.
2680 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
2682 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
2683 attributes): Document gettimeofday.
2685 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
2687 * config/s390/s390.c (s390_option_override_internal): Set
2688 PARAM_MIN_VECT_LOOP_BOUND
2690 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
2692 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
2693 * config/s390/s390.md: Likewise.
2695 2017-03-06 Jakub Jelinek <jakub@redhat.com>
2698 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
2699 (<avx2_avx512>_perm<mode>): Rename to ...
2700 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
2702 (<avx512>_perm<mode>_mask): Rename to ...
2703 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
2705 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
2706 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
2707 instead of VI8F_256_512.
2708 (avx512f_perm<mode>): New define_expand.
2709 (avx512f_perm<mode>_mask): Likewise.
2710 (avx512f_perm<mode>_1<mask_name>): New define_insn.
2711 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
2713 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
2715 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
2716 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
2718 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
2720 2017-03-06 Martin Liska <mliska@suse.cz>
2723 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
2724 when having a SSA NAME w/o VAR_DECL assigned to it.
2726 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
2728 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
2729 msa_dpsub_<su>_d): Fix MODE for vec_select.
2731 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
2733 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
2735 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
2737 2017-03-06 Richard Biener <rguenther@suse.de>
2739 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
2740 * plugin.c (register_plugin_info): Likewise.
2741 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
2743 2017-03-05 Jakub Jelinek <jakub@redhat.com>
2745 * config/i386/sse.md (sse_storehps, sse_storelps,
2746 avx_<castmode><avxsizesuffix>_<castmode>,
2747 avx512f_<castmode><avxsizesuffix>_<castmode>,
2748 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
2749 in condition that at least one operand is not a MEM.
2751 2017-03-03 Jakub Jelinek <jakub@redhat.com>
2754 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
2755 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
2757 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
2758 gimple_call_nothrow_p flag based on whether original builtin can throw.
2759 If it can, emit following stmts on the fallthrough edge.
2760 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
2761 don't create new bb if inserting just debug stmts on the edge, try to
2762 insert them on the fallthru bb or just reset debug stmts.
2764 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
2767 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
2768 restore recog_data (including the operand rtxes inside it) around
2769 the call to get_insn_template.
2771 2017-03-03 Martin Sebor <msebor@redhat.com>
2773 PR tree-optimization/79699
2774 * context.c (context::~context): Free MPFR caches to avoid
2775 a memory leak on program exit.
2777 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2779 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
2780 Use wide_int::ulow () instead of .elt (0).
2782 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
2784 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
2785 (*pushxf): Limit oF constraint to 32bit targets and add oC
2786 constraint for 64bit targets.
2787 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
2788 (*pushdf): Change rmF constraint to rmC.
2790 2017-03-03 Martin Liska <mliska@suse.cz>
2792 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
2793 Remove unused variable.
2795 2017-03-03 Jakub Jelinek <jakub@redhat.com>
2798 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
2799 is a memory operand, increase num_memory.
2800 (ix86_expand_args_builtin): Likewise.
2802 2017-03-03 Jan Hubicka <jh@suse.cz>
2805 * ipa-devirt.c (maybe_record_node): Properly handle
2806 __cxa_pure_virtual visibility.
2808 2017-03-03 Martin Liska <mliska@suse.cz>
2810 PR tree-optimization/79803
2811 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
2813 (pass_loop_prefetch::execute): Disabled optimization if an
2814 assumption about L1 cache size is not met.
2816 2017-03-03 Martin Liska <mliska@suse.cz>
2818 PR rtl-optimization/79574
2819 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
2820 (hash_scan_set): Likewise.
2821 (dump_hash_table): Likewise.
2822 (hoist_code): Likewise.
2824 2017-03-03 Richard Biener <rguenther@suse.de>
2826 * fixed-value.c (fixed_from_string): Restore use of elt (1)
2827 in place of uhigh ().
2828 (fixed_convert_from_real): Likewise.
2830 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
2833 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
2835 2017-03-03 Richard Biener <rguenther@suse.de>
2838 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
2839 TYPE_OVERFLOW_UNDEFINED check.
2841 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2843 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
2845 (vector_ae_<mode>_p): Likewise.
2846 (vector_nez_<mode>_p): Likewise.
2847 (vector_ne_v2di_p): Likewise.
2848 (vector_ae_v2di_p): Likewise.
2849 (vector_ne_<mode>_p): Likewise.
2850 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
2852 (vsx_tsqrt<mode>2_fe): Likewise.
2854 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
2857 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
2859 2017-03-02 Jakub Jelinek <jakub@redhat.com>
2861 PR rtl-optimization/79780
2862 * cprop.c (one_cprop_pass): When second and further conditional trap
2863 in a single basic block is turned into an unconditional trap, turn it
2864 into a deleted note to avoid RTL verification failures.
2866 2017-03-02 Richard Biener <rguenther@suse.de>
2868 * fold-const.c (const_binop): Use ulow () instead of elt (0).
2870 2017-03-02 Richard Biener <rguenther@suse.de>
2872 PR tree-optimization/79345
2874 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
2875 param and abort the walk, returning -1 if it is hit.
2876 (walk_aliased_vdefs): Take a limit param and pass it on.
2877 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
2878 defaulting to 0 and return a signed int.
2879 * tree-ssa-uninit.c (struct check_defs_data): New struct.
2880 (check_defs): New helper.
2881 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
2882 about uninitialized memory.
2883 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
2884 bogus uninitialized warning.
2885 (fixed_convert_from_real): Likewise.
2887 2017-03-02 Bin Cheng <bin.cheng@arm.com>
2889 PR tree-optimization/66768
2890 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
2891 iv_use if base object can't be determined.
2893 2017-03-02 Jakub Jelinek <jakub@redhat.com>
2895 PR tree-optimization/79345
2896 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
2897 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
2898 (get_pattern_stats): Initialize it.
2899 * genemit.c (gen_expand): Verify match_scratch numbers come after
2900 match_operand/match_dup numbers.
2901 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
2902 match_scratch numbers.
2903 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
2905 * config/s390/s390.md (trunctdsd2): Likewise.
2907 2017-03-02 Richard Biener <rguenther@suse.de>
2909 * wide-int.h (wide_int_storage::operator=): Implement in terms
2912 2017-03-02 Richard Biener <rguenther@suse.de>
2914 PR tree-optimization/79777
2915 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
2916 the to insert expression to sth existing.
2918 2017-03-01 Martin Sebor <msebor@redhat.com>
2921 * gimple-ssa-sprintf.c
2922 (directive::known_width_and_precision): New function.
2923 (format_integer): Use it.
2924 (get_mpfr_format_length): Consider the full range of precision
2925 when computing %g output with the # flag. Set the likely byte
2926 count to 3 rather than 1 when precision is indeterminate.
2927 (format_floating): Correct the lower bound of precision.
2929 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2931 * doc/invoke.texi: Document default code model for 64-bit Linux.
2933 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2936 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
2937 udiv rather than div since input pattern is unsigned.
2939 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
2941 * config/i386/i386.c (print_reg): Warn for values of
2942 unsupported size in integer register.
2944 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
2947 * config/rs6000/predicates.md (current_file_function_operand): Do
2948 not allow self calls to be local if the function is replaceable.
2950 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
2953 * config/rs6000/altivec.h (vec_ctz and others): Change the
2954 preprocessor macro that controls conditional compilation from
2955 _ARCH_PWR9 to __POWER9_VECTOR__.
2956 (vec_all_ne): Change parameterization of __altivec_scalar_pred
2957 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
2958 control (instead of _ARCH_PWR9 control) so that template
2959 definition uses power9-specific function.
2960 (vec_any_eq): Likewise.
2961 (vec_all_ne): Change macro definition to use a power9-specific
2962 expansion under #ifdef __POWER9_VECTOR__ control (instead of
2963 _ARCH_PWR9 control).
2964 (vec_any_eq) Likewise.
2965 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
2966 expansion for CMPNEF to remove support for xvcmpnesp instruction.
2967 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
2968 support for xvcmpnedp instruction.
2969 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
2970 macro expansion so that Power9 implementation of vec_all_ne does
2971 not use the AltiVec predicate framework.
2972 (VCMPNEH_P): Likewise.
2973 (VCMPNEW_P): Likewise.
2974 (VCMPNED_P): Likewise.
2975 (VCMPNEFP_P): Likewise.
2976 (VCMPNEDP_P): Likewise.
2977 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
2978 implementation of vec_any_eq to not use AltiVec predicate
2980 (VCMPAEH_P): Likewise.
2981 (VCMPAEW_P): Likewise.
2982 (VCMPAED_P): Likewise.
2983 (VCMPAEFP_P): Likewise.
2984 (VCMPAEDP_P): Likewise.
2985 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
2986 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
2987 not use the AltiVec predicate framework.
2988 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
2989 of vec_any_eq to not use AltiVec predicate framework.
2990 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
2991 support for predefined __POWER9_VECTOR__ macro to indicate that
2992 Power9 instruction selection is enabled.
2993 (altivec_overloaded_builtins): Remove extraneous
2994 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
2995 function argument types RS6000_BTI_bool_V16QI and
2996 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
2997 entry for overloaded function argument types RS6000_BTI_bool_V4SI
2998 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
2999 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
3000 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
3001 Power9 for implementations of vec_cmpne. Change the signature for
3002 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
3003 (representing vec_all_ne) to remove the previously described first
3004 argument of type RS6000_BTI_INTSI, as this was an artifact of
3005 reliance on the AltiVec predicate framework, which is no longer
3006 used in the implementation of these functions. Add
3007 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
3008 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
3009 since, unlike the AltiVec predicate framework implementation, we
3010 do not share function descriptors between vec_alle and vec_anyeq.
3011 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
3012 set of modes that receive special treatment even when
3013 TARGET_P9_VECTOR is true. The special treatment emits code that
3014 does not depend on Power9 instructions.
3015 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
3016 define_expand to not rely on AltiVec predicate framework.
3017 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
3019 (vector_ne_v2di_p): Change this define_expand to not rely on
3020 AltiVec predicate framework.
3021 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
3023 (vector_ne_<mode>_p): Change this define_expand to not rely on
3024 AltiVec predicate framework.
3025 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
3027 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
3028 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
3029 define_insn pattern.
3030 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
3031 define_insn pattern because the xvcmpne<VSs>. instruction is not
3033 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
3034 instruction is not supported.
3036 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3038 * config/nvptx/nvptx.c: Include intl.h.
3040 2017-03-01 Martin Jambor <mjambor@suse.cz>
3043 * ipa-prop.h (ipa_bits): Removed field known.
3044 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
3045 to pointers. Adjusted their comments to warn about their sharing.
3046 (ipcp_transformation_summary): Change bits to a vector of pointers.
3047 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
3048 (ipa_get_ipa_bits_for_value): Declare.
3049 * tree-vrp.h (value_range): Mark as GTY((for_user)).
3050 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
3051 (ipa_bits_hash_table): Likewise.
3052 (ipa_vr_ggc_hash_traits): Likewise.
3053 (ipa_vr_hash_table): Likewise.
3054 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
3055 being pointers and vr_known being removed.
3056 (ipa_set_jf_unknown): Likewise.
3057 (ipa_get_ipa_bits_for_value): New function.
3058 (ipa_set_jfunc_bits): Likewise.
3059 (ipa_get_value_range): New overloaded functions.
3060 (ipa_set_jfunc_vr): Likewise.
3061 (ipa_compute_jump_functions_for_edge): Use the above functions to
3062 construct bits and vr parts of jump functions.
3063 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
3064 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
3066 (ipcp_grow_transformations_if_necessary): Also allocate
3067 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
3069 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
3070 them. Fix too long lines.
3071 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
3072 vr_known being removed.
3073 (ipa_read_jump_function): Use new setter functions to construct bits
3074 and vr parts of jump functions or set them to NULL.
3075 (write_ipcp_transformation_info): Adjust for bits being pointers.
3076 (read_ipcp_transformation_info): Likewise.
3077 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
3079 Include gt-ipa-prop.h.
3080 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
3082 (ipcp_store_bits_results): Likewise.
3083 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
3084 Do not write to existing jump functions but use a temporary instead.
3086 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3089 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
3090 attempt to use its first operand as BIT_FIELD_REF base.
3092 2017-03-01 Richard Biener <rguenther@suse.de>
3095 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
3096 interpolating formula in wrapping arithmetic.
3097 (chrec_apply): Convert chrec_evaluate return value to wanted type.
3099 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3101 PR tree-optimization/79734
3102 * tree-vect-generic.c (expand_vector_condition): Optimize
3103 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
3104 Handle VEC_COND_EXPR where comparison has different inner width from
3107 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
3109 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
3110 markup, and similar issues. Remove @opindex entries for things
3111 that aren't options. Add missing -mmpy-option entries.
3113 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3115 PR tree-optimization/79737
3116 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
3117 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
3118 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
3119 instead of byte_size. Formatting fix.
3120 (shift_bytes_in_array_right): Formatting fix.
3122 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
3125 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
3126 condition on optimize for the leaf function test.
3128 2017-02-28 Martin Liska <mliska@suse.cz>
3131 * read-rtl-function.c (function_reader::handle_unknown_directive):
3132 Bail out when one uses -flto.
3134 2017-02-28 Martin Liska <mliska@suse.cz>
3136 * common.opt: Replace space with tabular for options of <number>
3138 * config/i386/i386.opt: Show <number> value for
3139 -mlarge-data-threshold.
3140 * opts.c (print_filtered_help): Do not display number in hexadecimal
3143 2017-02-28 Martin Liska <mliska@suse.cz>
3145 * common.opt: Fix --help=option -Q for options which are of
3148 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
3150 * config/i386/i386.c (print_reg): Error out for values
3151 of 8-bit size in invalid integer register.
3153 2017-02-28 Martin Sebor <msebor@redhat.com>
3155 PR tree-optimization/79691
3156 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
3158 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3161 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
3162 gcc_unreachable with output_operand_lossage.
3164 2017-02-28 Richard Biener <rguenther@suse.de>
3166 PR tree-optimization/79740
3167 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
3169 (visit_nary_op): Insert the nary into the hashtable if we
3170 pattern-matched sth.
3171 * tree-ssa-pre.c (eliminate_insert): Robustify.
3173 2017-02-28 Richard Biener <rguenther@suse.de>
3176 * fold-const.c (decode_field_reference): Reject out-of-bound
3179 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3181 * config/i386/i386.c: Include intl.h.
3182 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
3183 instead of just cond ? "..." : "...".
3184 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
3185 * coverage.c (read_counts_file): Likewise.
3186 * omp-offload.c: Include intl.h.
3187 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
3188 of just cond ? "..." : "...".
3189 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
3190 of just cond ? "..." : "...".
3192 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
3195 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
3197 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
3198 'tune for' CPU name.
3199 * config/arm/arm-cpu-data.h: Regenerated.
3201 2017-02-28 Richard Biener <rguenther@suse.de>
3203 PR tree-optimization/79732
3204 * tree-inline.c (expand_call_inline): Do not shadow var.
3206 2017-02-28 Richard Biener <rguenther@suse.de>
3208 PR tree-optimization/79723
3209 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
3210 address-space properly.
3212 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
3214 * doc/optinfo.texi (Optimization groups): Fix option used for
3216 * doc/invoke.texi (-fopt-info): Document "omp".
3217 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
3218 (OPTGROUP_ALL): Add OPTGROUP_OMP.
3219 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
3220 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
3221 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
3223 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
3225 * dumpfile.c (optgroup_options): Instead of "openmp", associate
3226 OPTGROUP_OMP with "omp".
3228 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
3231 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
3232 for arithmetic shift of unsigned V2DI.
3234 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
3236 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
3237 arc/linux.h headers.
3238 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
3239 (LINK_SPEC): Likewise.
3240 (ARC_TLS_EXTRA_START_SPEC): Likewise.
3241 (EXTRA_SPECS): Likewise.
3242 (STARTFILE_SPEC): Likewise.
3243 (ENDFILE_SPEC): Likewise.
3244 (LIB_SPEC): Likewise.
3245 (TARGET_SDATA_DEFAULT): Likewise.
3246 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
3247 (MULTILIB_DEFAULTS): Likewise.
3248 (DWARF2_UNWIND_INFO): Likewise.
3249 * config/arc/big.h: New file.
3250 * config/arc/elf.h: Likewise.
3251 * config/arc/linux.h: Likewise.
3252 * config/arc/t-uClibc: Remove.
3254 2017-02-27 Bin Cheng <bin.cheng@arm.com>
3256 PR tree-optimization/77536
3257 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
3258 (tree_transform_and_unroll_loop): Use above function to compute the
3259 estimated niter of unrolled loop and use it when scaling profile.
3260 Also use count info rather than frequency if it's non-zero.
3261 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
3262 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
3263 (vect_transform_loop): Call above function.
3265 2017-02-27 Richard Biener <rguenther@suse.de>
3267 PR tree-optimization/45397
3268 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
3269 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
3270 (visit_nary_op): Add pattern matching for CSEing sign-changed
3271 or truncated operations with wider ones.
3273 2017-02-27 Richard Biener <rguenther@suse.de>
3275 PR tree-optimization/79690
3276 * tree-vect-stmts.c (vectorizable_store): Use vector type
3277 built from the DR with address-space.
3279 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
3281 * doc/invoke.texi (Optimize Options): Refine the description
3282 of asan-use-after-return.
3284 2017-02-25 Alan Modra <amodra@gmail.com>
3286 PR rtl-optimization/79584
3287 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
3288 base, not ad->base_term, the reg within base. Remove assertion
3289 that ad->base == ad->base_term. Replace gen_int_mode using
3290 bogus mode with const0_rtx.
3292 2017-02-25 Jakub Jelinek <jakub@redhat.com>
3295 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
3296 FMA_EXPR like tcc_binary or tcc_unary.
3298 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
3301 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
3303 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
3304 (output_loc_operands): Handle DW_OP_call_ref and
3305 DW_OP_GNU_variable_value.
3306 (struct variable_value_struct): New type.
3307 (struct variable_value_hasher): Likewise.
3308 (variable_value_hash): New variable.
3309 (string_types): Remove.
3310 (copy_loc_descr): New function.
3311 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
3312 (prepend_loc_descr_to_each): New function.
3313 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
3314 instead of add_loc_descr_to_each if the first argument is single
3315 location list and the second has multiple.
3316 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
3317 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
3318 when looking for variable value which doesn't have other location info.
3319 (loc_list_from_tree): Formatting fix.
3320 (gen_array_type_die): Simplify DW_AT_string_length handling.
3321 (adjust_string_types): Remove.
3322 (gen_subprogram_die): Don't call adjust_string_types nor test/set
3323 string_types. Call resolve_variable_values.
3324 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
3325 (resolve_addr_in_expr): Likewise. Add A argument.
3326 (copy_deref_exprloc): Remove deref argument. Adjust for the
3327 original expression being DW_OP_GNU_variable_value with optionally
3328 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
3329 optionally after it.
3330 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
3331 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
3332 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
3333 (variable_value_hasher::hash, variable_value_hasher::equal): New
3335 (resolve_variable_value_in_expr, resolve_variable_value,
3336 resolve_variable_values, note_variable_value_in_expr,
3337 note_variable_value): New functions.
3338 (dwarf2out_early_finish): Call note_variable_value on all toplevel
3341 2017-02-24 Jakub Jelinek <jakub@redhat.com>
3344 * opts.h (handle_generated_option): Add GENERATED_P argument.
3345 * opts-common.c (handle_option): Adjust function comment.
3346 (handle_generated_option): Add GENERATED_P argument, pass it to
3348 (control_warning_option): Pass false to handle_generated_option
3350 * opts.c (maybe_default_option): Pass true to handle_generated_option
3352 * optc-gen.awk: Likewise.
3354 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3356 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
3357 a REG, look at the REG it is a SUBREG of.
3358 (splitter for cmpeqsi_t): Ditto.
3360 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3362 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
3363 the special USEs with the pattern of the insn, not the insn itself.
3365 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
3368 * doc/invoke.texi: Document -mload-store-pairs.
3370 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3371 Sandra Loosemore <sandra@codesourcery.com>
3373 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
3374 argument isn't a CONST_INT.
3375 (nios2_alternate_compare_const): Assert op is a CONST_INT.
3376 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
3377 (nios2_validate_compare): Bypass alternate compare logic if *op2
3379 (ldstwm_operation_p): Return false if first_base is not a REG or
3380 if first_offset is not a CONST_INT.
3382 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3384 * config/cris/cris.md: Use correct operand in a define_peephole2.
3386 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3388 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
3390 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3392 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
3393 this_insn if it is an INSN or JUMP_INSN.
3394 (force_offsettable): Look at base, not at addr.
3395 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
3396 on things that aren't necessarily CONST_INTs.
3398 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
3400 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
3401 -mfpmath=sse is the default also for x86-32 targets with SSE2
3402 instruction set when @option{-ffast-math} is enabled
3404 2017-02-24 Jeff Law <law@redhat.com>
3406 PR rtl-optimizatoin/79286
3407 * ira.c (update_equiv_regs): Drop may_trap_p exception to
3410 2017-02-24 Richard Biener <rguenther@suse.de>
3412 PR tree-optimization/79389
3413 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
3416 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
3418 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
3419 function comment to reflect reality.
3420 (loop_exits_before_overflow): Fix typo in function description.
3422 2017-02-24 Richard Biener <rguenther@suse.de>
3424 PR tree-optimization/79389
3425 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
3426 properly that a threading opportunity exists. Detect conditional
3427 copy/constant propagation opportunities.
3429 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
3431 * config/visium/visium.md (type): Add trap.
3432 (b): New mode attribute.
3433 (*btst): Rename into...
3434 (*btst<mode>): ...this and adjust.
3435 (*cbranchsi4_btst_insn): Rename into...
3436 (*cbranch<mode>4_btst_insn): ...this and adjust.
3437 (trap): New define_insn.
3439 2017-02-23 Jakub Jelinek <jakub@redhat.com>
3441 PR tree-optimization/79389
3442 * ifcvt.c (struct noce_if_info): Add rev_cond field.
3443 (noce_reversed_cond_code): New function.
3444 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
3445 reversed_comparison_code. Formatting fix.
3446 (noce_try_store_flag): Test rev_cond != NULL in addition to
3447 reversed_comparison_code.
3448 (noce_try_store_flag_constants): Likewise.
3449 (noce_try_store_flag_mask): Likewise.
3450 (noce_try_addcc): Use rev_cond if non-NULL instead of
3451 reversed_comparison_code.
3452 (noce_try_cmove_arith): Likewise. Formatting fixes.
3453 (noce_try_minmax, noce_try_abs): Clear rev_cond.
3454 (noce_find_if_block): Initialize rev_cond.
3455 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
3456 instead of false as last argument never attempt to reverse it
3459 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3461 PR tree-optimization/79663
3462 * tree-predcom.c (combine_chains): Process refs in reverse order
3463 only for ZERO length chains, and add explaining comment.
3465 2017-02-23 Jeff Law <law@redhat.com>
3467 PR tree-optimization/79578
3468 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
3469 in call to operand_equal_p.
3471 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
3474 * config/i386/cpuid.h: Fix another undefined behavior.
3476 2017-02-23 Richard Biener <rguenther@suse.de>
3478 PR tree-optimization/79683
3479 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
3480 vector types for data-refs.
3482 2017-02-23 Martin Liska <mliska@suse.cz>
3484 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
3486 2017-02-23 Jakub Jelinek <jakub@redhat.com>
3489 * internal-fn.c (get_range_pos_neg): Moved to ...
3490 * tree.c (get_range_pos_neg): ... here. No longer static.
3491 * tree.h (get_range_pos_neg): New prototype.
3492 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
3493 are known to be in between 0 and signed maximum inclusive, try to
3494 expand both unsigned and signed divmod and use the cheaper one from
3497 2017-02-22 Jeff Law <law@redhat.com>
3499 PR tree-optimization/79578
3500 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
3501 to compare base operands.
3503 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
3506 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
3507 gpc_reg_operand instead of fpr_reg_operand.
3509 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
3511 * config/mips/mips.c (mips_return_in_memory): Force FP
3512 vector types to be returned in memory for o32 ABI.
3514 2017-02-22 Jakub Jelinek <jakub@redhat.com>
3516 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
3517 instead of DW_TAG_member for static data member declarations and don't
3518 set no_linkage_name for static inline data members.
3519 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
3522 2017-02-22 Martin Liska <mliska@suse.cz>
3524 * doc/invoke.texi: Replace inequality signs with square brackets
3527 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3529 PR tree-optimization/68644
3530 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
3532 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
3535 * lra-constraints.c (simplify_operand_subreg): Handle
3536 WORD_REGISTER_OPERATIONS targets.
3538 2017-02-22 Jakub Jelinek <jakub@redhat.com>
3541 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
3542 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
3543 elimination by swapping fld*.
3545 2017-02-22 Richard Biener <rguenther@suse.de>
3547 PR tree-optimization/79673
3548 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
3549 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
3550 irrelevant address-space qualifiers and avoiding a
3551 ADDR_SPACE_CONVERT_EXPR from fold_convert.
3553 2017-02-22 Richard Biener <rguenther@suse.de>
3555 PR tree-optimization/79666
3556 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
3557 to not symbolically negate if that may introduce undefined
3560 2017-02-22 Martin Liska <mliska@suse.cz>
3563 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
3564 * data-streamer-out.c (streamer_write_gcov_count_stream):
3566 * value-prof.c (stream_out_histogram_value): Make assert more
3567 precise based on type of counter.
3569 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
3572 * config/i386/i386.md (standard_x87sse_constant_load splitter):
3573 Use nonimmediate_operand instead of memory_operand for operand 1.
3574 (float-extend standard_x87sse_constant_load splitter): Ditto.
3576 2017-02-21 Jeff Law <law@redhat.com>
3578 PR tree-optimization/79621
3579 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
3580 blocks with edges to themselves.
3582 2017-02-21 Jakub Jelinek <jakub@redhat.com>
3585 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
3586 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
3587 Use gimple_call_builtin_p.
3590 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
3591 on temporarily removed DEBUG_INSNs.
3593 PR tree-optimization/79649
3594 * tree-loop-distribution.c (classify_partition): Give up on
3595 non-generic address space loads/stores.
3597 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
3599 * doc/loop.texi (Loop manipulation): Remove nonexistent
3600 tree_ssa_loop_version from the documentation.
3601 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
3603 2017-02-21 Jakub Jelinek <jakub@redhat.com>
3606 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
3607 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
3608 * config/rs6000/rs6000.c: Include except.h.
3609 (rs6000_expand_split_stack_prologue): Call
3610 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
3612 2017-02-21 Martin Jambor <mjambor@suse.cz>
3615 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
3618 2017-02-21 Martin Jambor <mjambor@suse.cz>
3620 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
3621 for backward compatibility only.
3622 * doc/invoke.texi (Option Summary): Remove all references to
3625 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
3629 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3631 * lra-constraints.c (curr_insn_transform): Handle
3632 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
3634 2017-02-21 Martin Liska <mliska@suse.cz>
3636 * config/i386/i386.opt: Replace -masm-dialect with -masm.
3638 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
3640 PR translation/79638
3641 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
3643 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
3646 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
3647 (arm_function_ok_for_sibcall): Return false for an indirect call by
3648 descriptor if all the argument registers are used.
3649 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
3650 alignment of the function.
3652 2017-02-21 Jakub Jelinek <jakub@redhat.com>
3654 PR tree-optimization/61441
3655 * simplify-rtx.c (simplify_const_unary_operation): For
3656 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
3657 the sNaN unmodified.
3659 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
3661 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
3662 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
3663 instead of SYSTEM_HEADER_DIR.
3665 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
3666 Martin Liška <mliska@suse.cz>
3668 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
3669 Fix typos and grammar, use active voice, and clarify.
3671 2017-02-20 Marek Polacek <polacek@redhat.com>
3674 * gimplify.c (gimplify_expr): Handle unused *&&L;.
3677 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
3679 2017-02-20 Jakub Jelinek <jakub@redhat.com>
3682 * config/i386/i386.c (ix86_expand_builtin): Handle
3683 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
3684 ix86_builtins_isa[fcode].isa as a requirement of those
3685 flags and any other flag in the bitmask.
3686 (ix86_init_mmx_sse_builtins): Use 0 instead of
3687 ~OPTION_MASK_ISA_64BIT as mask.
3688 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
3689 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
3690 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
3691 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
3693 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3696 * lra-constraints.c (split_reg): Check requested split mode
3697 is supported by the register.
3699 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3701 * lra-constraints.c (simplify_operand_subreg): Remove early
3704 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3707 * lra-constraints.c (curr_insn_transform): Tighten condition
3708 for converting SUBREG reloads from OP_OUT to OP_INOUT.
3710 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3713 * lra-constraints.c (curr_insn_transform): Handle
3714 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
3716 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
3719 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
3721 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
3723 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
3726 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
3729 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3731 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
3734 2017-02-18 Jakub Jelinek <jakub@redhat.com>
3736 * final.c (last_columnnum, override_columnnum): New variables.
3737 (final_start_function): Set last_columnnum, pass it to begin_prologue
3738 hook and pass 0 to dwarf2out_begin_prologue.
3739 (final_scan_insn): Update override_columnnum. Pass last_columnnum
3740 to source_line debug hook.
3741 (notice_source_line): Compute last_columnnum and for debug_column_info
3742 return true on column changes.
3743 * debug.h (struct gcc_debug_hooks): Add column argument to
3744 source_line and begin_prologue hooks.
3745 (debug_nothing_int_charstar_int_bool): Remove prototype.
3746 (debug_nothing_int_int_charstar,
3747 debug_nothing_int_int_charstar_int_bool): New prototypes.
3748 (dwarf2out_begin_prologue): Add column argument.
3749 * debug.c (do_nothing_debug_hooks): Adjust source_line and
3750 begin_prologue hooks.
3751 (debug_nothing_int_charstar_int_bool): Remove.
3752 (debug_nothing_int_int_charstar,
3753 debug_nothing_int_int_charstar_int_bool): New functions.
3754 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
3755 through to dwarf2out_source_line.
3756 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
3757 (dwarf2out_source_line): Add column argument, emit it if requested.
3758 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
3760 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
3761 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
3762 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
3763 through to dwarf2out_begin_prologue.
3764 (vmsdbgout_source_line): Add column argument, pass it through to
3765 dwarf2out_source_line.
3766 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
3767 dbxout_source_line caller.
3768 (dbxout_source_line): Add column argument.
3770 * common.opt (gno-column-info, gcolumn-info): New options.
3771 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
3772 (check_die): Also test for multiple DW_AT_decl_column attributes.
3773 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
3774 DW_AT_decl_column if requested.
3775 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
3777 (gen_variable_die): Likewise.
3778 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
3779 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
3782 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
3783 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
3784 (ix86_handle_option): Handle OPT_m3dnowa.
3785 * doc/invoke.texi (-m3dnowa): Document.
3786 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
3787 -m3dnowa instead of -m3dnow -march=athlon.
3790 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
3791 instead of gcc_assert for K, r and R code checks. Formatting fixes.
3793 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3796 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
3797 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
3798 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
3799 generator for vsx_xxpermdi_<mode>_be.
3800 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
3801 force big-endian semantics.
3802 (vsx_xxpermdi_<mode>_be): New define_expand with same
3803 implementation as previous version of vsx_xxpermdi_<mode>.
3805 2017-02-17 Jakub Jelinek <jakub@redhat.com>
3807 PR tree-optimization/79327
3808 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
3809 variable, its initialization and use.
3811 2017-02-17 Julia Koval <julia.koval@intel.com>
3813 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
3814 (OPTION_MASK_ISA_PKU_UNSET): New.
3815 (ix86_handle_option): Handle -mrdpid.
3816 * config/i386/cpuid.h (bit_RDPID): New.
3817 * config/i386/driver-i386.c (host_detect_local_cpu):
3818 Detect RDPID feature.
3819 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
3820 * config/i386/i386-c.c (ix86_target_macros_internal):
3822 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
3823 (ix86_valid_target_attribute_inner_p): Add "rdpid".
3824 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
3825 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
3826 * config/i386/i386.md (define_insn "rdpid"): New.
3827 * config/i386/i386.opt Add -mrdpid.
3828 * config/i386/immintrin.h (_rdpid_u32): New.
3830 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
3832 PR rtl-optimization/79541
3833 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
3834 instead of transforming it into USE.
3836 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
3838 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
3839 If HONOR_SNANS (SFmode) force the input to a register.
3840 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
3841 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
3842 an frsp or similar insn.
3844 2017-02-17 Martin Liska <mliska@suse.cz>
3846 PR rtl-optimization/79577
3847 * params.def (selsched-max-sched-times): Increase minimum to 1.
3849 2017-02-17 Martin Liska <mliska@suse.cz>
3851 PR rtl-optimization/79574
3852 * gcse.c (want_to_gcse_p): Prevent integer overflow.
3854 2017-02-17 Martin Liska <mliska@suse.cz>
3856 PR tree-optimization/79529
3857 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
3858 ssa_defined_default_def_p to handle cases which are implicitly
3860 * tree-ssa.c (ssa_defined_default_def_p): New function.
3861 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
3862 which are implicitly defined.
3863 * tree-ssa.h (ssa_defined_default_def_p): Declare.
3865 2017-02-17 Richard Biener <rguenther@suse.de>
3868 * params.def (max-ssa-name-query-depth): Limit to 10.
3870 2017-02-17 Richard Biener <rguenther@suse.de>
3872 PR tree-optimization/79552
3873 * tree-ssa-structalias.c (visit_loadstore): Properly verify
3876 2017-02-17 Richard Biener <rguenther@suse.de>
3879 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
3881 2017-02-17 Marek Polacek <polacek@redhat.com>
3884 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
3885 (fold_negate_expr): New wrapper.
3887 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
3889 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
3890 Correct terminology and de-emphasize pre-standard behavior.
3892 2017-02-16 Alan Modra <amodra@gmail.com>
3894 PR rtl-optimization/79286
3895 * ira.c (def_dominates_uses): New function.
3896 (update_equiv_regs): Don't create an equivalence for insns that
3897 may trap where the register def does not dominate the use.
3899 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
3901 PR rtl-optimization/78127
3902 * lra.c (lra): Call lra_eliminate before finish the loop after
3905 2017-02-16 Richard Biener <rguenther@suse.de>
3907 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
3909 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
3910 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
3911 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
3912 (isl_val_int_from_wi): New function.
3913 (extract_affine_gmp): Rename to ...
3914 (extract_affine_wi): ... this, take a widest_int.
3915 (extract_affine_int): Just wrap extract_affine_wi.
3916 (add_param_constraints): Use isl_val_int_from_wi.
3917 (add_loop_constraints): Likewise, and extract_affine_wi.
3919 2017-02-15 Jeff Law <law@redhat.com>
3922 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
3923 ira_init_register_move_cost_if_necessary.
3925 2017-02-15 Martin Sebor <msebor@redhat.com>
3928 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
3929 removed in a prior commit.
3931 2017-02-15 Bin Cheng <bin.cheng@arm.com>
3933 PR tree-optimization/79347
3934 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
3935 counters during peeling.
3937 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
3939 * Makefile.in (site.exp): Remove "set ISLVER".
3941 2017-02-15 Jakub Jelinek <jakub@redhat.com>
3944 * real.c (real_from_integer): Call real_convert even for decimal.
3946 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
3949 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
3951 2017-02-14 Andrew Pinski <apinski@cavium.com>
3953 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
3954 cores and change the partno/implementer to be correct.
3955 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
3956 the 'B" as the implementer.
3957 * config/aarch64/aarch64-tune.md: Regenerate.
3959 2017-02-14 Carl Love <cel@us.ibm.com>
3961 * config/rs6000/rs6000.c: Add case statement entry to make the
3962 xvcvuxdsp built-in argument unsigned.
3963 * config/rs6000/vsx.md: Fix the source and return operand types so they
3964 match the instruction definitions from the ISA document. Fix typo
3965 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
3968 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
3971 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
3972 member early_clobber_alts.
3973 * lra-lives.c (reg_early_clobber_p): New.
3974 (process_bb_lives): Use it.
3975 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
3976 (debug_operand_data): Initialize early_clobber_alts.
3977 (setup_operand_alternative): Set up early_clobber_alts.
3978 (collect_non_operand_hard_regs): Ditto. Pass early clobber
3979 alternatives to new_insn_reg.
3980 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
3982 (lra_update_insn_regno_info): Pass the new arg.
3984 2017-02-14 Jakub Jelinek <jakub@redhat.com>
3987 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
3988 (new_oacc_loop_raw): Don't clear already cleared fields.
3991 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
3992 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
3993 _mm512_prefetch_i64gather_ps): New inline functions and macros.
3995 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
3998 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
4000 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
4003 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
4004 the extra instruction to the right place to store 128-bit constant
4007 2017-02-14 Martin Sebor <msebor@redhat.com>
4010 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
4011 warning for strings of unknown length.
4013 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
4015 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
4017 2017-02-14 Jeff Law <law@redhat.com>
4020 * ira-costs.c (scan_one_insn): Initialize register move costs
4021 for pseudos seen in USE/CLOBBER insns.
4023 PR tree-optimization/79095
4024 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
4025 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
4026 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
4027 if the operands are known to be not equal, then the resulting range
4029 (intersect_ranges): If the new range is ~[0,0] and the old range is
4030 wide, then prefer ~[0,0].
4031 * tree-vrp.c (overflow_comparison_p_1): New function.
4032 (overflow_comparison_p): New function.
4033 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
4034 if NAME is used in an overflow test.
4035 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
4036 overflow check that can be expressed as an equality test, then adjust
4037 ops to be that equality test.
4039 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4041 * config/s390/s390-builtin-types.def: Remove flags argument.
4042 * config/s390/s390.c (s390_init_builtins): Likewise.
4044 2017-02-14 Martin Liska <mliska@suse.cz>
4046 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
4047 vector. Fix trailing white spaces.
4049 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
4051 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
4054 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4056 PR rtl-optimization/68664
4057 * config/arm/arm.c (arm_sched_can_speculate_insn):
4058 New function. Declare prototype.
4059 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
4061 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4063 PR rtl-optimization/68664
4064 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
4066 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
4068 2017-02-14 Amit Pawar <amit.pawar@amd.com>
4070 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
4071 max skip bytes for function, loop and jump.
4073 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4075 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
4076 ABS_EXPR for gimple dump.
4078 2017-02-14 Jakub Jelinek <jakub@redhat.com>
4081 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
4083 PR tree-optimization/79408
4084 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
4085 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
4086 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
4087 also if rhs1 is INTEGER_CST.
4089 2017-02-14 Richard Biener <rguenther@suse.de>
4092 * tree-into-ssa.c (insert_phi_nodes): When the function can
4093 have abnormal edges rewrite SSA names with broken use-def
4094 dominance out of SSA and register them for PHI insertion.
4096 2017-02-13 Martin Sebor <msebor@redhat.com>
4099 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
4100 clearing info.nowrite flag when snprintf size argument is a range.
4102 2017-02-13 Jakub Jelinek <jakub@redhat.com>
4104 * cprop.c (cprop_jump): Add missing space in string literal.
4105 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
4106 (get_constraint_for_component_ref): Likewise.
4107 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
4108 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
4109 * lra-constraints.c (process_alt_operands): Likewise.
4110 * ipa-inline.c (inline_small_functions): Likewise.
4111 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
4112 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
4113 * trans-mem.c (diagnose_tm_1_op): Likewise.
4114 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
4115 (grid_parallel_clauses_gridifiable): Likewise.
4117 * config/nvptx/mkoffload.c (process): Add space in between
4120 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
4121 "MOD4_SSE_REGS" and "ALL_REGS".
4123 * spellcheck.c (test_data): Add , in between "foo" and "food".
4125 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4128 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
4129 boundary crossing check and subsequent code generation agree.
4131 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4133 * config/aarch64/aarch64.c (has_memory_op): Delete.
4134 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
4137 2017-02-13 Jakub Jelinek <jakub@redhat.com>
4139 PR rtl-optimization/79388
4140 PR rtl-optimization/79450
4141 * combine.c (distribute_notes): When removing TEM_INSN for which
4142 corresponding dest has last value recorded, invalidate that last
4145 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4147 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
4148 of explicit '@'. Add missing assembly comment marker on branch costs
4151 2017-02-13 Nathan Sidwell <nathan@acm.org>
4153 * gengtype-lex.l (<in_struct>): Add '/'.
4155 2017-02-13 Martin Liska <mliska@suse.cz>
4158 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
4160 2017-02-13 Richard Biener <rguenther@suse.de>
4162 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
4164 * configure: Re-generate.
4165 * config.in: Likewise.
4166 * graphite-dependences.c: Simplify as if
4167 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
4168 * graphite-isl-ast-to-gimple.c: Likewise.
4169 * graphite-optimize-isl.c: Likewise.
4170 * graphite-poly.c: Likewise.
4171 * graphite-sese-to-poly.c: Likewise.
4172 * graphite.h: Likewise.
4173 * toplev.c: Include isl/version.h and use isl_version () for
4174 printing the ISL version.
4175 * doc/install.texi: Update ISL requirement.
4177 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
4179 * doc/standards.texi (Standards): Update reference to
4182 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
4184 * doc/extend.texi (Named Address Spaces): sourceware.org now
4186 * doc/install.texi (Binaries): Ditto.
4189 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
4191 * doc/cpp.texi: Replace "stringify"/"stringification" with C
4192 standard terminology "stringize"/"stringizing" throughout.
4193 * doc/cppinternals.texi: Likewise.
4195 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
4197 * doc/extend.texi: Fix some spelling mistakes and typos.
4198 * doc/invoke.texi: Likewise.
4200 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
4203 * params.def (inline-min-speedup) Change from 10 to 8.
4205 2017-02-11 Jakub Jelinek <jakub@redhat.com>
4207 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
4210 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
4213 * ipa-inline-analysis.c (get_minimal_bb): New function.
4214 (record_modified): Use it.
4215 (remap_edge_change_prob): Handle also ancestor functions.
4217 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
4219 * doc/contrib.texi (Contributors): Remove broken link into
4220 the Mauve CVS repository.
4222 2017-02-11 Jakub Jelinek <jakub@redhat.com>
4225 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
4226 result computation whenever lhs doesn't have vector mode, not
4227 just when it has BLKmode.
4229 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
4231 * doc/makefile.texi (profiledbootstrap): Refer to the
4232 installation instructions only in textual form.
4234 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4237 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
4239 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
4241 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
4242 (Specific): Update mingw-w64 reference.
4244 (Specific): Remove broken link to Renesas RX processor.
4246 2017-02-10 Richard Biener <rguenther@suse.de>
4248 * toplev.c (process_options): Do not mention obsolete graphite
4249 options when printing sorry message about missing graphite support.
4250 Mention -floop-nest-optimize.
4252 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
4254 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
4255 (vtst_p16): Likewise.
4256 (vtstq_p8): Likewise.
4257 (vtstq_p16): Likewise.
4259 (vtstq_p64): Likewise.
4260 * config/arm/arm_neon.h (vgetq_lane_p64): New.
4261 (vset_lane_p64): New.
4262 (vsetq_lane_p64): New.
4264 2017-02-10 Jakub Jelinek <jakub@redhat.com>
4266 PR tree-optimization/79411
4267 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
4268 stmt operands are SSA_NAMEs used in abnormal phis.
4269 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
4272 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
4275 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
4278 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
4280 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
4282 2017-02-09 Jakub Jelinek <jakub@redhat.com>
4284 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
4288 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
4289 not arbitrary TREE_CONSTANT.
4292 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
4293 "omp declare target link" attribute unless is_global_var.
4294 * omp-offload.c (find_link_var_op): Likewise.
4296 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
4297 Chung-Lin Tang <cltang@codesourcery.com>
4299 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
4301 (gimplify_adjust_omp_clauses): Don't delete TILE.
4302 (gimplify_omp_for): Deal with TILE.
4303 * internal-fn.c (expand_GOACC_TILE): New function.
4304 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
4306 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
4307 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
4309 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
4310 avoid DIV for outermost collapse var.
4311 (expand_oacc_for): Insert tile element loop as needed. Adjust.
4312 Remove out of date comments, fix whitespace.
4313 * omp-general.c (omp_extract_for_data): Deal with tiling.
4314 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
4315 adjust OLF_DIM_BASE value.
4316 (struct omp_for_data): Add tiling field.
4317 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
4318 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
4319 for auto loops. Remove default auto determining, moved to
4320 oacc_loop_fixed_partitions.
4321 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
4322 stmts, add e_mask field.
4323 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
4324 (oacc_thread_numbers): Use oacc_dim_call.
4325 (oacc_xform_tile): New.
4326 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
4327 (finish_oacc_loop): Adjust for ifns vector.
4328 (oacc_loop_discover_walk): Append loop abstraction sites to list,
4329 add case for GOACC_TILE fns.
4330 (oacc_loop_xform_loop): Delete.
4331 (oacc_loop_process): Iterate over call list directly, and add
4332 handling for GOACC_TILE fns.
4333 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
4335 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
4336 vector partitioning to outer loops. Assign 2 partitions to loops
4337 when available. Add TILE handling.
4338 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
4339 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
4340 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
4341 * tree.c (omp_clause_num_ops): Adjust TILE ops.
4342 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
4344 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
4346 * configure.ac (ACX_BUGURL): Update.
4347 * configure: Regenerate.
4349 2017-02-09 Richard Biener <rguenther@suse.de>
4351 PR tree-optimization/69823
4352 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
4353 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
4355 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
4357 * config/arc/arc-c.def: Add __NPS400__ definition.
4358 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
4359 (TARGET_NPS400): Define.
4361 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
4363 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
4365 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
4367 (arc_cpu_types): Fill the arch_info field with a pointer into the
4368 arc_arch_types table.
4369 (arc_selected_cpu): Declare.
4370 * config/arc/arc.c (arc_selected_cpu): Make global.
4371 (arc_selected_arch): Delete.
4372 (arc_base_cpu): Delete.
4373 (arc_override_options): Remove references to deleted variables,
4374 update access to arch information.
4375 (ARC_OPT): Update access to arch information.
4376 (ARC_OPTX): Likewise.
4377 * config/arc/arc.h (arc_base_cpu): Remove declaration.
4378 (TARGET_ARC600): Update access to arch information.
4379 (TARGET_ARC601): Likewise.
4380 (TARGET_ARC700): Likewise.
4381 (TARGET_EM): Likewise.
4382 (TARGET_HS): Likewise.
4383 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
4386 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
4389 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
4390 condition/operands for integer GE/LE/GEU/LEU operations.
4392 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
4394 PR translation/79397
4395 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
4398 2017-02-08 Martin Jambor <mjambor@suse.cz>
4401 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
4402 whether allocation happened.
4403 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
4404 nothing was allocated.
4406 2017-02-08 Jakub Jelinek <jakub@redhat.com>
4408 PR tree-optimization/79408
4409 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
4410 constant, but SSA_NAME with a known integer range, use the minimum
4411 of that range instead of op1 to determine if modulo can be replaced
4412 with its first operand.
4414 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4416 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
4418 2017-02-08 Richard Biener <rguenther@suse.de>
4420 PR tree-optimization/71824
4421 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
4422 Check all loops contained in the merged region.
4424 2017-02-07 Andrew Pinski <apinski@cavium.com>
4426 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
4428 2017-02-07 Andrew Pinski <apinski@cavium.com>
4430 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
4431 (thunderxt88): Likewise.
4432 (thunderxt81): Disable LSE and change v8.1 to v8.
4433 (thunderxt83): Likewise.
4435 2017-02-07 Jakub Jelinek <jakub@redhat.com>
4436 Richard Biener <rguenther@suse.de>
4439 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
4440 type from int to size_t.
4441 * ira-costs.c (struct_costs_size): Change type from int to size_t.
4443 2017-02-07 Jakub Jelinek <jakub@redhat.com>
4445 PR rtl-optimization/79386
4446 * cprop.c (bypass_conditional_jumps): Initialize
4447 bypass_last_basic_block already before splitting bbs after
4448 unconditional traps...
4449 (bypass_conditional_jumps): ... rather than here.
4452 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
4453 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
4454 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
4455 fix -masm=intel patterns.
4457 2017-02-07 Richard Biener <rguenther@suse.de>
4459 PR tree-optimization/79256
4461 * builtins.c (get_object_alignment_2): Use min_align_of_type
4462 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
4463 and ADJUST_FIELD_ALIGN.
4465 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
4467 * doc/tm.texi: Regenerate.
4468 * stor-layout.c (layout_decl): Adjust.
4469 (update_alignment_for_field): Likewise.
4470 (place_field): Likewise.
4471 (min_align_of_type): Likewise.
4472 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
4473 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
4474 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
4475 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
4476 * config/frv/frv.c (frv_adjust_field_align): Likewise.
4477 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
4478 * config/i386/i386.c (x86_field_alignment): Likewise.
4479 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
4480 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
4481 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
4482 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
4483 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
4484 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
4488 2017-01-30 Richard Biener <rguenther@suse.de>
4490 PR tree-optimization/79256
4491 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
4492 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
4495 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
4497 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
4498 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
4499 builtins to SImode and emit a zero-extend, if necessary.
4501 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
4503 * docs/invoke.texi (RISC-V Options): Alphabetize.
4505 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
4507 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
4510 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
4512 * config/riscv/riscv.c: New file.
4513 * gcc/common/config/riscv/riscv-common.c: Likewise.
4514 * config.gcc: Likewise.
4515 * config/riscv/constraints.md: Likewise.
4516 * config/riscv/elf.h: Likewise.
4517 * config/riscv/generic.md: Likewise.
4518 * config/riscv/linux.h: Likewise.
4519 * config/riscv/multilib-generator: Likewise.
4520 * config/riscv/peephole.md: Likewise.
4521 * config/riscv/pic.md: Likewise.
4522 * config/riscv/predicates.md: Likewise.
4523 * config/riscv/riscv-builtins.c: Likewise.
4524 * config/riscv/riscv-c.c: Likewise.
4525 * config/riscv/riscv-ftypes.def: Likewise.
4526 * config/riscv/riscv-modes.def: Likewise.
4527 * config/riscv/riscv-opts.h: Likewise.
4528 * config/riscv/riscv-protos.h: Likewise.
4529 * config/riscv/riscv.h: Likewise.
4530 * config/riscv/riscv.md: Likewise.
4531 * config/riscv/riscv.opt: Likewise.
4532 * config/riscv/sync.md: Likewise.
4533 * config/riscv/t-elf-multilib: Likewise.
4534 * config/riscv/t-linux: Likewise.
4535 * config/riscv/t-linux-multilib: Likewise.
4536 * config/riscv/t-riscv: Likewise.
4537 * configure.ac: Likewise.
4538 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
4539 Waterman as RISC-V maintainers.
4540 * doc/install.texi: Add RISC-V entries.
4541 * doc/invoke.texi: Add RISC-V options section.
4542 * doc/md.texi: Add RISC-V constraints section.
4543 * configure: Regenerated.
4545 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4548 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
4549 false values to be constant vectors with all 0 or all 1 bits set.
4550 (vcondu<mode><mode>): Likewise.
4551 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
4553 (fpmask_comparison_operator): Update comment.
4554 (vecint_comparison_operator): New predicate.
4555 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
4556 vector conditionals when the true and false values are constant
4557 vectors with all 0 bits or all 1 bits set.
4559 2017-02-06 Martin Sebor <msebor@redhat.com>
4561 PR tree-optimization/79376
4562 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
4564 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
4566 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
4567 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
4568 to simplify split condition.
4570 2017-02-06 Jakub Jelinek <jakub@redhat.com>
4572 * omp-expand.c (oxpand_omp_atomic_fetch_op,
4573 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
4576 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
4578 PR rtl-optimization/68664
4579 * target.def (can_speculate_insn): New hook.
4580 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
4581 * doc/tm.texi: Regenerate.
4582 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
4583 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
4584 (rs6000_sched_can_speculate_insn): New function.
4586 2017-02-06 Jakub Jelinek <jakub@redhat.com>
4588 PR tree-optimization/79284
4589 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
4590 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
4591 vectorizable_mask_load_store, vectorizable_operation,
4592 vect_is_simple_cond, get_same_sized_vectype): Use it instead
4593 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
4594 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
4595 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
4596 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4597 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
4598 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
4599 is_gimple_assign (stmt). Replace another such test with
4600 is_gimple_assign (stmt).
4602 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
4605 * config/avr/avr.c (rtl-iter.h): Include it.
4606 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
4607 (avr_legitimate_combined_insn): ...and implementation.
4609 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
4611 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
4612 * config/s390/s390.c (s390_const_operand_ok)
4613 (s390_canonicalize_comparison, s390_extract_part)
4614 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
4615 (s390_contiguous_bitmask_p, s390_rtx_costs)
4616 (legitimize_pic_address): Likewise.
4617 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
4618 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
4619 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
4620 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
4621 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
4623 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
4625 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
4626 REGNO($0) == REGNO($1).
4628 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4630 * config/s390/linux.h(SIZE_TYPE): Add comment.
4632 2017-02-06 Julian Brown <julian@codesourcery.com>
4633 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4634 Virendra Pathak <virendra.pathak@broadcom.com>
4636 * config/aarch64/aarch64-cores.def: Change the scheduler
4638 * config/aarch64/aarch64.md: Include thunderx2t99.md.
4639 * config/aarch64/thunderx2t99.md: New file.
4641 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
4643 * doc/standards.texi (Go Language): Update link to language
4646 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
4648 * tree-eh.c (lower_resx): Sanitize profile.
4649 (cleanup_empty_eh_move_lp): Likewise.
4651 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
4654 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
4656 * cfgloopmanip.h (loop_version): Update prototype.
4657 * modulo-sched.c (sms_schedule): Update call of loop_version.
4658 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
4659 * tree-parloops.c (gen_parallel_loop): Likewise.
4660 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
4661 * tree-ssa-loop-split.c (split_loop): Likewise.
4662 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
4663 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
4665 2017-02-05 Martin Liska <mliska@suse.cz>
4668 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
4670 (print_operand_address): Initialize a struct to zero.
4672 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
4674 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
4675 garbage collector only in textual form.
4677 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
4679 * doc/extend.texi (x86 specific memory model extensions for
4680 transactional memory): Simplify a phrase.
4682 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
4685 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
4686 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
4687 (atomic_storedi_1): Likewise.
4689 2017-02-04 Jakub Jelinek <jakub@redhat.com>
4691 PR tree-optimization/79338
4692 * tree-parloops.c (gather_scalar_reductions): Don't call
4693 vect_analyze_loop_form for loop->inner before destroying loop's
4696 2017-02-03 Martin Sebor <msebor@redhat.com>
4698 PR tree-optimization/79327
4699 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
4700 when precision has resulted in leading zeros.
4701 (format_integer): Adjust the likely counter to assume an unknown
4702 argument that may be zero is non-zero.
4704 2017-02-03 Jason Merrill <jason@redhat.com>
4707 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
4708 avoid copying non-taken branch.
4710 2017-02-03 Jakub Jelinek <jakub@redhat.com>
4712 PR tree-optimization/79340
4713 * tree-vect-loop.c (vectorizable_reduction): Release
4714 vec_defs elements after safe_splicing them into other vectors.
4717 PR tree-optimization/79327
4718 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
4719 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
4721 (format_integer): Use wide_int_to_tree instead of build_int_cst
4722 + to_?hwi. If argmin is NULL, just set argmin and argmax to
4723 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
4724 of shortest and longest sequence.
4726 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
4728 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
4729 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
4731 2017-02-03 Walter Lee <walt@tilera.com>
4734 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
4735 after initial stackframe link reg save.
4736 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
4738 2017-02-03 Jakub Jelinek <jakub@redhat.com>
4741 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
4742 wu for stxssp alternative.
4744 2017-02-03 Martin Sebor <msebor@redhat.com>
4746 PR tree-optimization/79352
4747 * gimple-fold.c (get_range_strlen): Add argument.
4748 (get_range_strlen): Change return type to bool.
4749 (get_maxval_strlen): Pass in a dummy argument.
4750 * gimple-fold.h (get_range_strlen): Change return type to bool.
4751 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
4752 * tree.h (array_at_struct_end_p): Add argument.
4753 * tree.c (array_at_struct_end_p): Handle it.
4755 2017-02-03 Martin Liska <mliska@suse.cz>
4758 * multiple_target.c (create_dispatcher_calls): Redirect edge
4759 from a caller of a dispatcher.
4760 (expand_target_clones): Make the clones local.
4761 (ipa_target_clone): Do both target clones and resolvers.
4762 (ipa_dispatcher_calls): Remove the pass.
4763 (pass_dispatcher_calls::gate): Likewise.
4764 (make_pass_dispatcher_calls): Likewise.
4765 * passes.def (pass_target_clone): Put as very first IPA early
4768 2017-02-03 Martin Liska <mliska@suse.cz>
4770 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
4771 in case of a function with ifunc attribute.
4773 2017-02-03 Martin Liska <mliska@suse.cz>
4775 * cgraph.c (cgraph_node::dump): Dump function version info.
4776 * symtab.c (symtab_node::dump_base): Add missing new line.
4778 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
4780 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
4781 (ifcombine_ifandif): Use it.
4783 2017-02-03 Martin Liska <mliska@suse.cz>
4785 * doc/invoke.texi: Document default value for
4786 use-after-scope-direct-emission-threshold.
4788 2017-02-03 Martin Liska <mliska@suse.cz>
4790 PR tree-optimization/79339
4791 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
4792 (format_floating): Likewise.
4794 2017-02-03 Martin Liska <mliska@suse.cz>
4797 * ipa-prop.c (ipa_node_params_t::insert): Remove current
4799 (ipa_node_params_t::remove): Likewise.
4800 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
4801 initialization from removed ipa_node_params_t::insert.
4802 (ipa_node_params::~ipa_node_params): Move from removed
4803 ipa_node_params_t::release.
4804 * symbol-summary.h (symbol_summary::m_released): New member.
4805 Do not release a summary twice. Do not allow to call finalizer
4806 for types of a summary that live in GGC memory.
4808 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4810 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
4813 2017-02-02 Martin Sebor <msebor@redhat.com>
4816 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
4817 (format_string): Tighten up the range of output for non-constant
4818 strings and correct the expected range for wide non-constant strings.
4820 2017-02-02 Martin Sebor <msebor@redhat.com>
4822 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
4825 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
4827 (-fdump-tree-@var): Add to index and document how to come up
4828 with pass-specific option and dump file names.
4829 (-fdump-passes): Clarify where to look for output.
4831 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
4834 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
4835 statistics of the analyzed path; allow threading for speed when
4836 any of BBs along the path are optimized for speed.
4838 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
4841 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
4842 settings of the virtual registers.
4845 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
4847 * explow.c (get_dynamic_stack_size): Take known alignment of stack
4848 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
4851 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4853 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
4854 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
4856 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4858 * config/s390/s390.md: Add missing comments with the expanded
4860 * config/s390/vector.md: Likewise.
4861 * config/s390/vx-builtins.md: Likewise.
4863 2017-02-02 Jakub Jelinek <jakub@redhat.com>
4866 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
4867 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
4868 conditions on a single line.
4870 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4872 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
4873 __S390_VX__ to __VX__.
4875 2017-02-01 Andrew Pinski <apinski@cavium.com>
4877 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
4878 stmt_info to record_stmt_cost.
4879 (vect_get_known_peeling_cost): Pass stmt_info if known to
4881 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
4882 cpu_vector_cost field into
4883 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
4884 field into vec_int_stmt_cost and vec_fp_stmt_cost.
4885 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
4886 splitting of scalar_stmt_cost and vec_stmt_cost.
4887 (thunderx_vector_cost): Likewise.
4888 (cortexa57_vector_cost): LIkewise.
4889 (exynosm1_vector_cost): Likewise.
4890 (xgene1_vector_cost): Likewise.
4891 (thunderx2t99_vector_cost): Improve after the splitting of the two
4893 (aarch64_builtin_vectorization_cost): Update for the splitting of
4894 scalar_stmt_cost and vec_stmt_cost.
4896 2017-02-01 Torvald Riegel <triegel@redhat.com>
4897 Richard Henderson <rth@redhat.com>
4899 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
4900 conditional on existance of a fast atomic load.
4901 * optabs-query.c (can_atomic_load_p): New function.
4902 * optabs-query.h (can_atomic_load_p): Declare it.
4903 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
4904 no fast atomic load is available for the particular size of access.
4905 (expand_atomic_compare_and_swap): Likewise.
4906 (expand_atomic_load): Likewise.
4907 (expand_atomic_store): Likewise.
4908 (expand_atomic_fetch_op): Likewise.
4909 * testsuite/lib/target-supports.exp
4910 (check_effective_target_sync_int_128): Remove x86 because it provides
4911 no fast atomic load.
4912 (check_effective_target_sync_int_128_runtime): Likewise.
4914 2017-02-01 Richard Biener <rguenther@suse.de>
4916 * graphite.c: Include tree-vectorizer.h for find_loop_location.
4917 (graphite_transform_loops): Provide opt-info for optimized nests.
4918 * tree-parloop.c (parallelize_loops): Provide opt-info for
4921 2017-02-01 Richard Biener <rguenther@suse.de>
4924 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
4927 2017-02-01 Richard Biener <rguenther@suse.de>
4929 PR tree-optimization/71824
4930 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
4931 Verify the loops are valid in the merged SESE region.
4932 (scop_detection::can_represent_loop_1): Check analyzing the
4933 evolution of the number of iterations in the region succeeds.
4935 2017-01-31 Ian Lance Taylor <iant@golang.org>
4937 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
4938 REG_ARGS_SIZE note to 32-bit push insns and call insn.
4940 2017-01-31 David Malcolm <dmalcolm@redhat.com>
4942 PR preprocessor/79210
4943 * input.c (get_substring_ranges_for_loc): Replace line_width
4944 assertion with error-handling.
4946 2017-01-31 Richard Biener <rguenther@suse.de>
4948 PR tree-optimization/77318
4949 * graphite-sese-to-poly.c (extract_affine): Fix assert.
4950 (create_pw_aff_from_tree): Take loop parameter.
4951 (add_condition_to_pbb): Pass loop of the condition to
4952 create_pw_aff_from_tree.
4954 2017-01-31 Jakub Jelinek <jakub@redhat.com>
4956 * config/s390/s390.c (s390_asan_shadow_offset): New function.
4957 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
4959 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
4963 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
4965 (convert_int_to_float128): Likewise.
4966 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
4967 (convert_int_to_float128): Likewise.
4968 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
4969 (UNSPEC_IEEE128_CONVERT): Likewise.
4970 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
4971 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
4972 Use local variables for IBM extended format.
4973 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
4974 (fix_trunc<mode>si2_fprs): Likewise.
4975 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
4976 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
4977 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
4978 to know that we can now have integers of all sizes in vector
4980 (fix<uns>_<mode>di2_hw): Likewise.
4981 (float<uns>_<mode>si2_hw): Likewise.
4982 (fix_<mode>si2_hw): Likewise.
4983 (fixuns_<mode>si2_hw): Likewise.
4984 (float<uns>_<mode>di2_hw): Likewise.
4985 (float_<mode>di2_hw): Likewise.
4986 (float_<mode>si2_hw): Likewise.
4987 (floatuns_<mode>di2_hw): Likewise.
4988 (floatuns_<mode>si2_hw): Likewise.
4989 (xscvqp<su>wz_<mode>): Delete, no longer used.
4990 (xscvqp<su>dz_<mode>): Likewise.
4991 (xscv<su>dqp_<mode>): Likewise.
4992 (ieee128_mfvsrd_64bit): Likewise.
4993 (ieee128_mfvsrd_32bit): Likewise.
4994 (ieee128_mfvsrwz): Likewise.
4995 (ieee128_mtvsrw): Likewise.
4996 (ieee128_mtvsrd_64bit): Likewise.
4997 (ieee128_mtvsrd_32bit): Likewise.
4999 2017-01-31 Martin Liska <mliska@suse.cz>
5002 * ipa-prop.c (ipa_free_all_node_params): Call release method
5003 instead of ~sumbol_summary to not to trigger double times
5006 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
5008 PR tree-optimization/71691
5009 * bitmap.h (class auto_bitmap): New.
5010 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
5011 is_maybe_undefined instead of ssa_undefined_value_p.
5013 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5015 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
5016 __S390_ARCH_LEVEL__ to __ARCH__.
5018 2017-01-31 Jakub Jelinek <jakub@redhat.com>
5020 PR tree-optimization/79267
5021 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
5022 if should_remove_lhs_p is true.
5024 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
5027 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
5028 (add_alignment_attribute): New.
5029 (base_type_die): Add alignment attribute.
5030 (subrange_type_die): Likewise.
5031 (modified_type_die): Likewise.
5032 (gen_array_type_die): Likewise.
5033 (gen_descr_array_type_die: Likewise.
5034 (gen_enumeration_type_die): Likewise.
5035 (gen_subprogram_die): Likewise.
5036 (gen_variable_die): Likewise.
5037 (gen_field_die): Likewise.
5038 (gen_ptr_to_mbr_type_die): Likewise.
5039 (gen_struct_or_union_type_die): Likewise.
5040 (gen_subroutine_type_die): Likewise.
5041 (gen_typedef_die): Likewise.
5042 (base_type_cmp): Compare alignment attribute.
5044 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
5047 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
5048 (setb_unsigned) New pattern for setb with CCUNS.
5049 * config/rs6000/rs6000.c (expand_block_compare): Use a different
5050 subfc./subfe sequence to avoid overflow problems. Generate a
5051 shorter sequence with cmpld/setb for power9.
5052 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
5053 for generating subfc. instruction.
5054 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
5055 now uses this instruction.
5057 2017-01-30 Ian Lance Taylor <iant@google.com>
5060 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
5061 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
5063 2017-01-30 Martin Sebor <msebor@redhat.com>
5065 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
5066 Move constant to the right of a relational operator.
5067 (get_mpfr_format_length, format_character, format_string): Ditto.
5068 (should_warn_p, maybe_warn): Same.
5070 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
5072 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5075 * asan.c (get_translation_unit_decl): Remove function.
5076 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
5078 2017-01-30 Martin Liska <mliska@suse.cz>
5080 PR gcov-profile/79259
5081 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
5084 2017-01-30 Martin Liska <mliska@suse.cz>
5087 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
5088 Initialize variables with NULL value.
5090 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
5093 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
5095 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
5097 2017-01-30 Richard Biener <rguenther@suse.de>
5099 PR tree-optimization/79276
5100 * tree-vrp.c (process_assert_insertions): Properly adjust common
5101 when removing a duplicate.
5103 * gcc.dg/torture/pr79276.c: New testcase.
5105 2017-01-30 Richard Biener <rguenther@suse.de>
5107 PR tree-optimization/79256
5108 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
5109 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
5111 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
5113 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
5116 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
5117 ("*r<noxa>sbg_<mode>_sll_bitmask")
5118 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
5119 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
5120 Use contiguous_bitmask_nowrap_operand.
5122 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5125 * config/rs6000/altivec.h (vec_xl): Revise #define.
5126 (vec_xst): Likewise.
5128 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
5130 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
5132 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
5134 PR rtl-optimization/79194
5135 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
5136 traps before call to bypass_conditional_jumps.
5138 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
5140 PR tree-optimization/71374
5141 * lra-constraints.c (check_conflict_input_operands): New.
5142 (match_reload): Use it.
5144 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
5147 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
5148 account to calculate conflict_set.
5150 2017-01-27 Bin Cheng <bin.cheng@arm.com>
5152 PR rtl-optimization/78559
5153 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
5154 other_insn in combine.
5156 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
5158 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
5159 uint16_type_node for BT_UINT16.
5161 2017-01-27 David Malcolm <dmalcolm@redhat.com>
5163 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
5164 "RTL Tests" to menu.
5165 (GIMPLE Tests): New node.
5166 (RTL Tests): New node.
5168 2017-01-27 Richard Biener <rguenther@suse.de>
5170 PR tree-optimization/79245
5171 * tree-loop-distribution.c (distribute_loop): Apply cost
5172 modeling also to detected patterns.
5174 2017-01-27 Richard Biener <rguenther@suse.de>
5176 PR tree-optimization/71433
5177 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
5178 (compare_assert_loc): New function.
5179 (process_assert_insertions): Sort and optimize assert locations
5180 to remove duplicates and push down identical assertions on
5181 edges to their destination block.
5183 2017-01-27 Richard Biener <rguenther@suse.de>
5185 PR tree-optimization/79244
5186 * tree-vrp.c (remove_range_assertions): Forcefully propagate
5187 out SSA names even if abnormal.
5189 2017-01-27 Jakub Jelinek <jakub@redhat.com>
5191 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
5192 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
5193 instead of MPFR_RNDN.
5195 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
5198 * arm.c (arm_option_override): Don't call build_target_option_node
5199 until after doing all option overrides.
5200 (arm_valid_target_attribute_tree): Likewise.
5202 2017-01-27 Martin Liska <mliska@suse.cz>
5204 * doc/invoke.texi (-fprofile-arcs): Document profiling support
5205 for {cd}tors and C++ {cd}tors.
5207 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
5209 * config/s390/s390.md ("*setmem_long_and")
5210 ("*setmem_long_and_31z"): Use zero_extend instead of and.
5212 2017-01-26 Martin Sebor <msebor@redhat.com>
5214 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
5217 2017-01-26 Martin Sebor <msebor@redhat.com>
5219 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
5220 HAVE_DFmode before using XFmode or DFmode.
5221 (parse_directive): Avoid using the z length modifier to avoid
5222 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
5225 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
5226 to accept adjustment as an array.
5227 (get_int_range): New function.
5228 (struct directive): Make width and prec arrays.
5229 (directive::set_width, directive::set_precision): Call get_int_range.
5230 (format_integer, format_floating): Handle width and precision ranges.
5231 (format_string, parse_directive): Same.
5233 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5236 * dwarf2out.c (generate_skeleton_bottom_up): For children with
5237 comdat_type_p set, just clone them, but keep the children in the
5241 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
5242 which have direct callers with -fvar-tracking-assignments enabled
5244 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
5245 inside of type units.
5247 2017-01-26 Martin Sebor <msebor@redhat.com>
5250 * gimple-ssa-sprintf.c (struct result_range): Add likely and
5252 (struct format_result): Replace number_chars, number_chars_min,
5253 and number_chars_max with a single member of struct result_range.
5255 (format_result::operator+=): Adjust.
5256 (struct fmtresult): Remove bounded. Handle likely and unlikely
5258 (fmtresult::adjust_for_width_or_precision): New function.
5259 (fmtresult:type_max_digits): New function.
5260 (bytes_remaining): Handle likely and unlikely counters.
5261 (min_bytes_remaining): Remove.
5262 (format_percent): Simplify.
5263 (format_integer, format_floating): Set likely and unlikely counters.
5264 (get_string_length, format_character, format_string): Same.
5265 (format_plain, should_warn_p): New function.
5266 (maybe_warn): Call should_warn_p. Update diagnostic messages
5267 and handle those for all directives, including plain strings.
5268 (format_directive): Handle likely and unlikely counters.
5269 Remove unnecessary quoting from diagnostics. Add an informational
5271 (add_bytes): Remove.
5272 (pass_sprintf_length::compute_format_length): Simplify.
5273 (try_substitute_return_value): Handle likely and unlikely counters.
5275 2017-01-26 Carl Love <cel@us.ibm.com>
5277 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
5278 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
5280 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
5283 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
5284 endianess for subregs into account.
5285 * lra-constraints.c (lra_constraints): Do risky transformations
5286 always on the first iteration.
5287 * lra-lives.c (check_pseudos_live_through_calls): Add arg
5288 last_call_used_reg_set.
5289 (process_bb_lives): Define and use last_call_used_reg_set.
5290 * lra.c (lra): Always continue after lra_constraints on the first
5293 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
5295 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
5297 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
5299 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5301 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
5302 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
5303 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
5304 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
5305 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
5306 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
5307 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
5308 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
5309 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
5311 2017-01-26 Marek Polacek <polacek@redhat.com>
5314 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
5315 for the third operand.
5317 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5320 * omp-low.c (struct omp_context): Add simt_stmt field.
5321 (scan_omp_for): Return omp_context *.
5322 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
5323 context to the _simt_ SIMD stmt.
5324 (lower_omp_for): For combined SIMD with sibling _simt_
5325 SIMD, make sure to use the same decls in _looptemp_
5326 clauses as in the sibling.
5328 2017-01-26 David Sherwood <david.sherwood@arm.com>
5331 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
5334 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5337 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
5338 emit fld b; fld a; if possible.
5340 * brig-builtins.def: Update copyright years.
5341 * config/arm/arm_acle_builtins.def: Update copyright years.
5343 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5346 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
5347 constraint instead of o for the stxsd instruction.
5349 2017-01-25 Carl Love <cel@us.ibm.com>
5351 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
5352 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
5354 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
5356 * doc/invoke.texi (C++ Dialect Options): Fix typo.
5358 2017-01-25 Richard Biener <rguenther@suse.de>
5360 PR tree-optimization/69264
5361 * target.def (vector_alignment_reachable): Improve documentation.
5362 * doc/tm.texi: Regenerate.
5363 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
5365 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
5366 earlier changes with respect to TYPE_USER_ALIGN.
5367 (vector_alignment_reachable_p): Likewise. Improve dumping.
5369 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5372 * config/arm/arm.md (xordi3): Force constant operand into a register
5375 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5377 * doc/invoke.texi (-fstore-merging): Correct default optimization
5378 levels at which it is enabled.
5379 (-O): Move -fstore-merging from list to...
5382 2017-01-25 Richard Biener <rguenther@suse.de>
5385 * omp-expand.c: Include debug.h.
5386 (expand_omp_taskreg): Make sure to generate early debug before
5387 outlining anything from a function.
5388 (expand_omp_target): Likewise.
5389 (grid_expand_target_grid_body): Likewise.
5391 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
5394 * asan.c (get_translation_unit_decl): New function.
5395 (asan_add_global): Extract modules file name from globals
5396 TRANSLATION_UNIT_DECL name.
5398 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
5401 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
5402 for long calls with APCS frame and VFP.
5404 2017-01-24 David Malcolm <dmalcolm@redhat.com>
5406 * cfg.c (original_copy_tables_initialized_p): New function.
5407 * cfg.h (original_copy_tables_initialized_p): New decl.
5408 * cfgrtl.c (relink_block_chain): Guard the call to
5409 free_original_copy_tables with a call to
5410 original_copy_tables_initialized_p.
5411 * cgraph.h (symtab_node::native_rtl_p): New decl.
5412 * cgraphunit.c (symtab_node::native_rtl_p): New function.
5413 (symtab_node::needed_p): Don't assert for early assembly output
5414 for __RTL functions.
5415 (cgraph_node::finalize_function): Set "force_output" for __RTL
5417 (cgraph_node::analyze): Bail out early for __RTL functions.
5418 (analyze_functions): Update assertion to support __RTL functions.
5419 (cgraph_node::expand): Bail out early for __RTL functions.
5420 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
5422 * function.h (struct function): Update comment for field
5424 * gimple-expr.c: Include "tree-pass.h".
5425 (gimple_has_body_p): Return false for __RTL functions.
5426 * Makefile.in (OBJS): Add run-rtl-passes.o.
5427 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
5429 (gcc::pass_manager::get_clean_slate): New accessor.
5430 * passes.c: Include "insn-addr.h".
5431 (should_skip_pass_p): Add logging. Update logic for running
5432 "expand" to be compatible with both __GIMPLE and __RTL. Guard
5433 property-provider override so it is only done for gimple passes.
5435 (skip_pass): New function.
5436 (execute_one_pass): Call skip_pass when skipping passes.
5437 * read-md.c (md_reader::read_char): Support filtering
5438 the input to a subset of line numbers.
5439 (md_reader::md_reader): Initialize fields
5440 m_first_line and m_last_line.
5441 (md_reader::read_file_fragment): New function.
5442 * read-md.h (md_reader::read_file_fragment): New decl.
5443 (md_reader::m_first_line): New field.
5444 (md_reader::m_last_line): New field.
5445 * read-rtl-function.c (function_reader::create_function): Only
5446 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
5447 curr_properties. Set DECL_INITIAL to a dummy block.
5448 (read_rtl_function_body_from_file_range): New function.
5449 * read-rtl-function.h (read_rtl_function_body_from_file_range):
5451 * run-rtl-passes.c: New file.
5452 * run-rtl-passes.h: New file.
5454 2017-01-24 Jeff Law <law@redhat.com>
5456 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
5459 2017-01-24 Bin Cheng <bin.cheng@arm.com>
5461 PR tree-optimization/79159
5462 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
5463 (record_nonwrapping_iv): Improve boundary using above function if no
5464 value range information.
5466 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
5467 Martin Jambor <mjambor@suse.cz>
5469 * brig-builtins.def: New file.
5470 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
5471 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
5472 (DEF_HSAIL_SAT_BUILTIN): Likewise.
5473 (DEF_HSAIL_INTR_BUILTIN): Likewise.
5474 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
5475 * builtin-types.def (BT_INT8): New.
5476 (BT_INT16): Likewise.
5477 (BT_UINT8): Likewise.
5478 (BT_UINT16): Likewise.
5479 (BT_FN_ULONG): Likewise.
5480 (BT_FN_UINT_INT): Likewise.
5481 (BT_FN_UINT_ULONG): Likewise.
5482 (BT_FN_UINT_LONG): Likewise.
5483 (BT_FN_UINT_PTR): Likewise.
5484 (BT_FN_ULONG_PTR): Likewise.
5485 (BT_FN_INT8_FLOAT): Likewise.
5486 (BT_FN_INT16_FLOAT): Likewise.
5487 (BT_FN_UINT32_FLOAT): Likewise.
5488 (BT_FN_UINT16_FLOAT): Likewise.
5489 (BT_FN_UINT8_FLOAT): Likewise.
5490 (BT_FN_UINT64_FLOAT): Likewise.
5491 (BT_FN_UINT16_UINT32): Likewise.
5492 (BT_FN_UINT32_UINT16): Likewise.
5493 (BT_FN_UINT16_UINT16_UINT16): Likewise.
5494 (BT_FN_INT_PTR_INT): Likewise.
5495 (BT_FN_UINT_PTR_UINT): Likewise.
5496 (BT_FN_LONG_PTR_LONG): Likewise.
5497 (BT_FN_ULONG_PTR_ULONG): Likewise.
5498 (BT_FN_VOID_UINT64_UINT64): Likewise.
5499 (BT_FN_UINT8_UINT8_UINT8): Likewise.
5500 (BT_FN_INT8_INT8_INT8): Likewise.
5501 (BT_FN_INT16_INT16_INT16): Likewise.
5502 (BT_FN_INT_INT_INT): Likewise.
5503 (BT_FN_UINT_FLOAT_UINT): Likewise.
5504 (BT_FN_FLOAT_UINT_UINT): Likewise.
5505 (BT_FN_ULONG_UINT_UINT): Likewise.
5506 (BT_FN_ULONG_UINT_PTR): Likewise.
5507 (BT_FN_ULONG_ULONG_ULONG): Likewise.
5508 (BT_FN_UINT_UINT_UINT): Likewise.
5509 (BT_FN_VOID_UINT_PTR): Likewise.
5510 (BT_FN_UINT_UINT_PTR: Likewise.
5511 (BT_FN_UINT32_UINT64_PTR): Likewise.
5512 (BT_FN_INT_INT_UINT_UINT): Likewise.
5513 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
5514 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
5515 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
5516 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
5517 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
5518 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
5519 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
5520 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
5521 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
5522 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
5523 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
5524 * doc/frontends.texi: List BRIG FE.
5525 * doc/install.texi (Testing): Add BRIG tesring requirements.
5526 * doc/invoke.texi (Overall Options): Mention BRIG.
5527 * doc/standards.texi (Standards): Doucment BRIG HSA version.
5529 2017-01-24 Richard Biener <rguenther@suse.de>
5531 PR translation/79208
5532 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
5534 2017-01-24 Martin Jambor <mjambor@suse.cz>
5537 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
5538 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
5541 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
5544 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
5545 casts from signed to unsigned really don't have a range.
5547 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
5549 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
5550 GMP_RNDx for compatiblity.
5552 2017-01-24 Martin Liska <mliska@suse.cz>
5555 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
5556 that would prevent us to call alloca with -1 as argument.
5558 2017-01-24 Jakub Jelinek <jakub@redhat.com>
5560 * dwarf2out.c (output_compilation_unit_header, output_file_names):
5561 Avoid -Wformat-security warning.
5563 2017-01-23 Andrew Pinski <apinski@cavium.com>
5565 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
5568 2017-01-23 Martin Sebor <msebor@redhat.com>
5571 * gimple-ssa-sprintf.c (warn_level): New global.
5572 (format_integer): Use it here and throughout the rest of the file.
5573 Use the same switch to compute sign as base.
5574 (maybe_warn): New function.
5575 (format_directive): Factor out warnings into maybe_warn.
5576 Add debugging output. Use warn_level.
5577 (add_bytes): Use warn_level.
5578 (pass_sprintf_length::compute_format_length): Add debugging output.
5579 (try_substitute_return_value): Same.
5580 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
5583 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
5584 (struct fmtresult, format_integer, format_floating): Adjust.
5585 (fmtresult::fmtresult): Set max correctly in two argument ctor.
5586 (get_string_length, format_string,format_directive): Same.
5587 (pass_sprintf_length::compute_format_length): Same.
5588 (try_substitute_return_value): Simplify slightly.
5591 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
5592 (fmtresult::operator+=): Outlined.
5593 (struct fmtresult): Add ctors.
5594 (struct conversion_spec): Rename...
5595 (struct directive): ...to this. Add and remove data members.
5596 (directive::set_width, directive::set_precision): New functions.
5597 (format_percent): Use fmtresult ctor.
5598 (get_width_and_precision): Remove.
5599 (format_integer): Make naming changes. Avoid computing width and
5601 (format_floating): Same. Adjust indentation.
5602 (format_character, format_none): New functions.
5603 (format_string): Moved character handling to format_character.
5604 (format_directive): Remove arguments, change return type.
5605 (parse_directive): New function.
5606 (pass_sprintf_length::compute_format_length): Move directive
5607 parsing to parse_directive.
5609 2017-01-23 Jakub Jelinek <jakub@redhat.com>
5611 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
5612 (assign_assembler_name_if_needed): ... this.
5613 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
5614 (assign_assembler_name_if_needed): ... this.
5615 (free_lang_data_in_cgraph): Adjust callers.
5616 * cgraphunit.c (cgraph_node::analyze): Likewise.
5617 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
5620 2017-01-23 Richard Biener <rguenther@suse.de>
5622 PR tree-optimization/79088
5623 PR tree-optimization/79188
5624 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
5625 resetting loop bounds after last path deletion. Reset loop
5626 bounds of the target loop, make code match the comments.
5627 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
5628 Make sure loops need no fixups.
5630 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
5632 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
5633 exponent support with double type for first argument.
5634 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
5635 type returned by __builtin_vec_extract_sig,
5636 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
5637 functions from "vector int" to "vector unsigned int" or from
5638 "vector long long int" to "vector unsigned long long int".
5639 Changed type returned by __builtin_vec_extract_exp,
5640 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
5641 functions from "vector int" to "vector unsigned int" or from
5642 "vector long long int" to "vector unsigned long long int".
5643 Changed return type of __builtin_vec_test_data_class,
5644 __builtin_vec_test_data_class_sp, and
5645 __builtin_vec_test_data_class_dp from "vector int" to
5646 "vector bool int" or from "vector long long int" to "vector bool
5647 long long int" and changed second argument type from "unsigned
5648 int" to "int". Added new overloaded function forms "vector float
5649 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
5650 "vector float __builtin_vec_insert_exp_sp (vector float, vector
5651 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
5652 double, vector unsigned long long int)" and "vector double
5653 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
5654 long int)". Changed return type of
5655 __builtin_scalar_test_data_class and
5656 __builtin_scalar_test_data_class_sp and
5657 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
5658 int" and changed second argument from "unsigned int" to "int".
5659 Changed type returned by __builtin_scalar_test_neg,
5660 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
5661 from "int" to "bool int". Added new overloaded function form
5662 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
5663 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
5664 exponent double-precision with floating point first argument.
5665 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
5666 documentation of scalar_test_data_class, scalar_test_neg,
5667 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
5668 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
5669 vec_test_data_class built-in functions to reflect refinements in
5670 their type signatures.
5672 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
5674 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
5676 (aarch64_elf_asm_destructor): Likewise.
5678 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
5680 PR rtl-optimization/78634
5681 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
5682 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
5683 * ifcvt.c (noce_try_cmove): Add missing cost check.
5685 PR rtl-optimization/71724
5686 * combine.c (if_then_else_cond): Look for situations where it is
5687 beneficial to undo the work of one of the recursive calls.
5689 2017-01-23 Bin Cheng <bin.cheng@arm.com>
5691 PR tree-optimization/70754
5692 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
5693 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
5694 combined stmt before it if not NULL.
5695 (combine_chains): Process refs reversely and compute dominance point
5698 2017-01-23 Martin Liska <mliska@suse.cz>
5700 PR tree-optimization/79196
5701 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
5702 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
5704 (strlen_optimize_stmt): Call the renamed function.
5706 2017-01-23 Michael Matz <matz@suse.de>
5708 PR tree-optimization/78384
5709 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
5711 2017-01-23 Richard Biener <rguenther@suse.de>
5713 PR tree-optimization/79186
5714 * tree-vrp.c (register_new_assert_for): Make sure we've seen
5715 both incoming edges before moving an assert.
5717 2017-01-23 Martin Jambor <mjambor@suse.cz>
5719 * ipa-prop.c (load_from_param_1): Removed.
5720 (load_from_unmodified_param): Bits from load_from_param_1 put back
5722 (load_from_param): Removed.
5723 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
5724 with stmt. Reverted back to use of load_from_unmodified_param.
5726 2017-01-23 Martin Jambor <mjambor@suse.cz>
5729 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
5730 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
5731 field a pointer to garbage collected vector, mark lattices and
5732 ipcp_orig_node with GTY((skip)).
5733 (ipa_get_param_count): Adjust to descriptors being a pointer.
5734 (ipa_get_param): Likewise.
5735 (ipa_get_type): Likewise.
5736 (ipa_get_param_move_cost): Likewise.
5737 (ipa_set_param_used): Likewise.
5738 (ipa_get_controlled_uses): Likewise.
5739 (ipa_set_controlled_uses): Likewise.
5740 (ipa_is_param_used): Likewise.
5741 (ipa_node_params_t): Move into garbage collector. New methods insert
5743 (ipa_node_params_sum): Annotate wth GTY(()).
5744 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
5746 (ipa_load_from_parm_agg): Adjust declaration.
5747 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
5748 * ipa-profile.c (ipa_profile): Likewise.
5749 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
5750 (ipa_populate_param_decls): Make descriptors parameter garbage
5752 (ipa_dump_param): Adjust to descriptors being a pointer.
5753 (ipa_alloc_node_params): Likewise.
5754 (ipa_initialize_node_params): Likewise.
5755 (load_from_param_1): Make descriptors parameter garbage collected.
5756 (load_from_unmodified_param): Likewise.
5757 (load_from_param): Likewise.
5758 (ipa_load_from_parm_agg): Likewise.
5759 (ipa_node_params::~ipa_node_params): Removed.
5760 (ipa_free_all_node_params): Remove call to delete operator.
5761 (ipa_node_params_t::insert): New.
5762 (ipa_node_params_t::remove): Likewise.
5763 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
5764 copy known_csts and known_contexts vectors.
5765 (ipa_read_node_info): Adjust to descriptors being a pointer.
5766 (ipcp_modif_dom_walker): Make m_descriptors field garbage
5768 (ipcp_transform_function): Make descriptors variable garbage
5771 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
5773 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
5774 * config/i386/avx512dqintrin.h: Ditto.
5775 * config/i386/avx512fintrin.h: Ditto.
5776 * gcc/config/i386/i386.c: Handle new builtins.
5777 * config/i386/i386-builtin.def: Add new builtins.
5778 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
5779 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
5781 2017-01-23 Jakub Jelinek <jakub@redhat.com>
5782 Martin Liska <mliska@suse.cz>
5784 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
5785 * asan.c (asan_expand_poison_ifn): Support stores and use
5786 appropriate ASAN report function.
5787 * internal-fn.c (expand_ASAN_POISON_USE): New function.
5788 * internal-fn.def (ASAN_POISON_USE): Declare.
5789 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
5790 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
5791 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
5792 ASAN_POISON calls w/o LHS.
5793 * tree-ssa.c (execute_update_addresses_taken): Create clobber
5794 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
5795 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
5796 * gimplify.c (asan_poison_variables): Add attribute
5797 use_after_scope_memory to variables that really needs to live
5799 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
5800 having the attribute.
5802 2017-01-23 Martin Liska <mliska@suse.cz>
5804 * asan.c (create_asan_shadow_var): New function.
5805 (asan_expand_poison_ifn): Likewise.
5806 * asan.h (asan_expand_poison_ifn): New declaration.
5807 * internal-fn.c (expand_ASAN_POISON): Likewise.
5808 * internal-fn.def (ASAN_POISON): New builtin.
5809 * sanopt.c (pass_sanopt::execute): Expand
5810 asan_expand_poison_ifn.
5811 * tree-inline.c (copy_decl_for_dup_finish): Make function
5813 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
5814 * tree-ssa.c (is_asan_mark_p): New function.
5815 (execute_update_addresses_taken): Rewrite local variables
5816 (identified just by use-after-scope as addressable) into SSA.
5818 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
5820 * doc/install.texi (Specific): opensource.apple.com uses https
5821 now. Remove trailing slash.
5823 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
5825 * README.Portability: Remove note on an Irix compatibility issue.
5827 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
5829 * gcov.c (INCLUDE_ALGORITHM): Define.
5830 (INCLUDE_VECTOR): Define.
5831 No longer include <vector> and <algorithm> directly.
5833 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
5835 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
5837 * doc/invoke.texi (Code Gen Options): Ditto.
5839 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
5842 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
5844 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
5846 rtl-optimization/79125
5847 * cprop.c (local_cprop_pass): Handle cases where we make an
5850 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
5854 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
5855 read from, for big endian.
5857 2017-01-20 Jiong Wang <jiong.wang@arm.com>
5859 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
5860 register pauth builtins for LP64 only.
5862 2017-01-20 Marek Polacek <polacek@redhat.com>
5865 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
5868 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
5870 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
5872 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
5873 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
5874 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
5876 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5879 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
5880 in big-endian mode when they are not a single duplicated value.
5882 2017-01-20 Richard Biener <rguenther@suse.de>
5884 * BASE-VER: Bump to 7.0.1.
5886 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
5888 * omp-low.c (omplow_simd_context): New struct. Use it...
5889 (lower_rec_simd_input_clauses): ...here and...
5890 (lower_rec_input_clauses): ...here to hold common data. Adjust all
5891 references to idx, lane, max_vf, is_simt.
5893 2017-01-20 Graham Markall <graham.markall@embecosm.com>
5895 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
5898 2017-01-20 Martin Jambor <mjambor@suse.cz>
5900 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
5901 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
5903 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
5904 (GTFILES): Rename hsa.c to hsa-common.c.
5905 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
5906 * hsa-dump.c: Likewise.
5907 * hsa-gen.c: Likewise.
5908 * hsa-regalloc.c: Likewise.
5909 * ipa-hsa.c: Likewise.
5910 * omp-expand.c: Likewise.
5911 * omp-low.c: Likewise.
5912 * toplev.c: Likewise.
5914 2017-01-20 Marek Polacek <polacek@redhat.com>
5917 * doc/invoke.texi: Document -Wduplicated-branches.
5918 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
5919 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
5920 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
5921 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
5922 return 0 only when not OEP_LEXICOGRAPHIC.
5923 (fold_build_cleanup_point_expr): Use the expression
5924 location when building CLEANUP_POINT_EXPR.
5925 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
5926 * tree.c (add_expr): Handle error_mark_node.
5928 2017-01-20 Martin Liska <mliska@suse.cz>
5931 * tree-profile.c (init_ic_make_global_vars): Do not call
5933 (gimple_init_gcov_profiler): Likewise.
5935 2017-01-20 Martin Liska <mliska@suse.cz>
5938 * cgraph.h (maybe_create_reference): Remove argument and
5940 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
5942 * ipa-cp.c (create_specialized_node): Likewise.
5943 * symtab.c (symtab_node::maybe_create_reference): Handle
5944 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
5946 2017-01-20 Martin Liska <mliska@suse.cz>
5948 * read-rtl-function.c (function_reader::create_function): Use
5949 build_decl instread of build_decl_stat.
5951 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
5953 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
5954 * config/i386/avx512dqintrin.h: Ditto.
5955 * config/i386/avx512fintrin.h: Ditto.
5956 * config/i386/i386-builtin-types.def: Add new types.
5957 * gcc/config/i386/i386.c: Handle new types.
5958 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
5959 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
5960 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
5961 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
5962 (__builtin_ia32_kshiftridi): New.
5963 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
5965 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
5969 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
5970 define to rs6000_init_stack_protect_guard.
5971 (rs6000_init_stack_protect_guard): New function.
5973 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
5974 Yunqiang Su <yunqiang.su@imgtec.com>
5976 * config.gcc (supported_defaults): Add madd4.
5977 (with_madd4): Add validation.
5978 (all_defaults): Add madd4.
5979 * config/mips/mips.opt (mmadd4): New option.
5980 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
5982 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
5984 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
5985 (ISA_HAS_FUSED_MADD4): Likewise.
5986 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
5987 * gcc/doc/install.texi (--with-madd4): Document the new option.
5989 2017-01-19 Jiong Wang <jiong.wang@arm.com>
5991 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
5992 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
5993 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
5994 (aarch64_init_pauth_hint_builtins): New.
5995 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
5996 (aarch64_expand_builtin): Expand new builtins.
5998 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6000 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
6001 * combine-stack-adj.c (no_unhandled_cfa): Handle
6002 REG_CFA_TOGGLE_RA_MANGLE.
6003 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
6004 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
6005 info for return address signing.
6006 (aarch64_expand_epilogue): Likewise.
6008 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6010 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
6011 * config/aarch64/aarch64-protos.h
6012 (aarch64_return_address_signing_enabled): New declaration.
6013 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
6015 (aarch64_expand_prologue): Sign return address before it's pushed onto
6017 (aarch64_expand_epilogue): Authenticate return address fetched from
6019 (aarch64_override_options): Sanity check for ILP32 and ISA level.
6020 (aarch64_attributes): New function attributes for "sign-return-address".
6021 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
6022 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
6023 ("*do_return"): Generate combined instructions according to key index.
6024 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
6025 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
6027 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
6028 * config/aarch64/aarch64.opt (msign-return-address=): New.
6029 * doc/extend.texi (AArch64 Function Attributes): Documents
6030 "sign-return-address=".
6031 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
6033 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
6035 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
6036 overall option summary.
6038 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6040 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
6041 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
6042 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
6043 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
6045 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
6047 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
6048 -mpower9-minmax by default for -mcpu=power9.
6049 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
6050 128-bit floating point.
6052 2017-01-20 Alan Modra <amodra@gmail.com>
6054 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
6055 optimizing for size.
6057 2017-01-20 Alan Modra <amodra@gmail.com>
6060 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
6061 for strcmp and strncmp from corresponding builtin decl.
6063 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
6065 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
6066 instead of i386/rtems-64.h.
6067 * config/i386/rtems-64.h: Remove.
6069 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
6073 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
6075 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
6077 2017-01-19 Tamar Christina <tamar.christina@arm.com>
6079 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6080 Change int to HOST_WIDE_INT.
6081 * config/aarch64/aarch64-protos.h
6082 (aarch64_simd_gen_const_vector_dup): Likewise.
6083 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
6085 2017-01-19 David Malcolm <dmalcolm@redhat.com>
6087 * langhooks-def.h (lhd_type_for_size): New decl.
6088 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
6089 * langhooks.c (lhd_type_for_size): New function, taken from
6092 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
6094 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
6095 define_bypass for CR latency.
6096 (power9-cracked-alu): Update bypass latency and remove power9-branch.
6097 (power9-alu2): Add define_bypass for CR latency.
6099 (power9-mul): Update insn latency.
6100 (power9-mul-compare): Update insn latency, bypass latency and remove
6103 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6105 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
6107 * config/aarch64/aarch64.md
6108 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
6109 aarch64_nopcrelative_literal_loads.
6110 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
6112 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
6114 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
6116 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
6118 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
6121 * config.gcc (supported_defaults): Add lxc1-sxc1.
6122 (with_lxc1_sxc1): Add validation.
6123 (all_defaults): Add lxc1-sxc1.
6124 * config/mips/mips.opt (mlxc1-sxc1): New option.
6125 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
6127 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
6128 __mips_no_lxc1_sxc1.
6129 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
6130 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
6131 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
6133 2017-01-19 Richard Biener <rguenther@suse.de>
6135 PR tree-optimization/72488
6136 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
6137 sure to restore SSA info.
6138 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
6140 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
6142 PR rtl-optimization/79121
6143 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
6144 of the inner type when shifting an extended value.
6146 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
6149 * symtab.c (symtab_node::equal_address_to): Fix comparing of
6150 interposable aliases.
6152 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
6155 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
6156 Use the evmergelohi instruction.
6157 (mov_si<mode>_e500_subreg4_2_le): Likewise.
6158 (mov_sitf_e500_subreg8_2_be): Likewise.
6159 (mov_sitf_e500_subreg12_2_le): Likewise.
6160 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
6161 (mov_si<mode>_e500_subreg4_2_be): Likewise.
6162 (mov_sitf_e500_subreg8_2_le): Likewise.
6163 (mov_sitf_e500_subreg12_2_be): Likewise.
6165 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6167 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
6168 attribute from vecsimple to vecperm.
6169 (altivec_vbpermq2): Likewise.
6171 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6174 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
6176 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6177 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
6178 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
6179 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
6180 case where N arg is SIZE_MAX.
6181 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
6182 (cmpstrsi): Add pattern.
6184 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
6186 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6187 __builtin_vec_revb builtins.
6188 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
6189 built-in functions to support generation of the ISA 3.0 XXBR<x>
6190 vector byte reverse instructions.
6191 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
6192 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
6193 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
6194 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
6195 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
6196 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
6197 (P9V_BUILTIN_VEC_REVB): Likewise.
6198 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
6199 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
6200 (p9_xxbrq_v16qi): Likewise.
6201 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
6202 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
6203 (p9_xxbrh_v8hi): Likewise.
6204 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
6205 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
6206 vec_revb built-in functions.
6208 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
6210 PR rtl-optimization/78952
6211 * config/i386/i386.md (any_extract): New code iterator.
6212 (*insvqi_2): Use any_extract for source operand.
6213 (*insvqi_3): Use any_shiftrt for source operand.
6215 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
6217 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
6219 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
6221 2017-01-18 Matthias Klose <doko@ubuntu.com>
6223 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
6225 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6227 * config/rs6000/altivec.h (vec_bperm): Change #define.
6228 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
6229 (altivec_vbpermq2): New define_insn.
6230 (altivec_vbpermd): Likewise.
6231 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
6233 (VBPERMD): Likewise.
6234 (VBPERM): New polymorphic function interface.
6235 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
6236 Add entries for P9V_BUILTIN_VEC_VBPERM.
6237 * doc/extend.texi: Add interfaces for vec_bperm.
6239 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6241 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
6242 first letter of error messages.
6243 (s390_resolve_overloaded_builtin): Likewise.
6244 * config/s390/s390.c (s390_expand_builtin): Likewise.
6245 (s390_invalid_arg_for_unprototyped_fn): Likewise.
6246 (s390_valid_target_attribute_inner_p): Likewise.
6247 * config/s390/s390.md ("tabort"): Likewise.
6249 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
6251 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
6252 (ISA_AVOID_DIV_HILO): New macro.
6253 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
6254 (ISA_HAS_DDIV): Likewise.
6256 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6258 * doc/invoke.texi (fabi-version): Correct number of occurrences.
6260 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6262 * doc/invoke.texi (fabi-version): Spelling fix.
6264 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6267 * doc/invoke.texi (fabi-version): Mention mangling fix for
6270 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6273 * doc/invoke.texi (fabi-version): Document discriminator mangling.
6275 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
6278 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
6279 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
6281 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
6283 (stack_protect_test): Ditto.
6284 * config/rs6000/rs6000.opt (mstack-protector-guard=,
6285 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
6287 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
6288 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
6289 -mstack-protector-guard-offset=.
6290 (RS/6000 and PowerPC Options): Ditto.
6292 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
6294 * config/i386/i386.h (MASK_CLASS_P): New define.
6295 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
6296 there are no registers from different register sets also when
6297 mask registers are used. Update function comment.
6298 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
6299 to (*k/*r) and (*k/*km) alternatives.
6301 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
6303 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
6304 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
6305 (EH_RETURN_HANDLER_RTX): New define.
6306 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
6307 Force frame pointer in EH return functions.
6308 (aarch64_expand_epilogue): Add barrier for eh_return.
6309 (aarch64_final_eh_return_addr): Remove.
6310 (aarch64_eh_return_handler_rtx): New function.
6311 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
6313 (aarch64_eh_return_handler_rtx): New prototype.
6315 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6317 * config/rs6000/altivec.h (vec_rlmi): New #define.
6318 (vec_vrlnm): Likewise.
6319 (vec_rlnm): Likewise.
6320 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
6321 (UNSPEC_VRLNM): Likewise.
6322 (VIlong): New mode iterator.
6323 (altivec_vrl<VI_char>mi): New define_insn.
6324 (altivec_vrl<VI_char>nm): Likewise.
6325 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
6328 (RLNM): New polymorphic function entry.
6329 (VRLWMI): New monomorphic function entry.
6331 (RLMI): New polymorphic function entry.
6332 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
6333 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
6334 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
6337 2017-01-17 Jakub Jelinek <jakub@redhat.com>
6340 * dwarf2out.c (field_byte_offset): Restore the
6341 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
6342 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
6343 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
6346 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
6349 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
6351 2017-01-17 Jakub Jelinek <jakub@redhat.com>
6354 * dwarf2out.c (add_data_member_location_attribute): For constant
6355 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
6356 instead of DW_AT_data_member_location, DW_AT_bit_offset and
6357 DW_AT_byte_size attributes.
6359 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
6361 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
6362 after forcing to constant memory when the code model is medium.
6364 2017-01-17 Julia Koval <julia.koval@intel.com>
6367 * config/i386/avx512fintrin.h
6368 (_mm512_i32gather_ps): Change __addr type to void const*.
6369 (_mm512_mask_i32gather_ps): Ditto.
6370 (_mm512_i32gather_pd): Ditto.
6371 (_mm512_mask_i32gather_pd): Ditto.
6372 (_mm512_i64gather_ps): Ditto.
6373 (_mm512_mask_i64gather_ps): Ditto.
6374 (_mm512_i64gather_pd): Ditto.
6375 (_mm512_mask_i64gather_pd): Ditto.
6376 (_mm512_i32gather_epi32): Ditto.
6377 (_mm512_mask_i32gather_epi32): Ditto.
6378 (_mm512_i32gather_epi64): Ditto.
6379 (_mm512_mask_i32gather_epi64): Ditto.
6380 (_mm512_i64gather_epi32): Ditto.
6381 (_mm512_mask_i64gather_epi32): Ditto.
6382 (_mm512_i64gather_epi64): Ditto.
6383 (_mm512_mask_i64gather_epi64): Ditto.
6384 (_mm512_i32scatter_ps): Change __addr type to void*.
6385 (_mm512_mask_i32scatter_ps): Ditto.
6386 (_mm512_i32scatter_pd): Ditto.
6387 (_mm512_mask_i32scatter_pd): Ditto.
6388 (_mm512_i64scatter_ps): Ditto.
6389 (_mm512_mask_i64scatter_ps): Ditto.
6390 (_mm512_i64scatter_pd): Ditto.
6391 (_mm512_mask_i64scatter_pd): Ditto.
6392 (_mm512_i32scatter_epi32): Ditto.
6393 (_mm512_mask_i32scatter_epi32): Ditto.
6394 (_mm512_i32scatter_epi64): Ditto.
6395 (_mm512_mask_i32scatter_epi64): Ditto.
6396 (_mm512_i64scatter_epi32): Ditto.
6397 (_mm512_mask_i64scatter_epi32): Ditto.
6398 (_mm512_i64scatter_epi64): Ditto.
6399 (_mm512_mask_i64scatter_epi64): Ditto.
6400 * config/i386/avx512pfintrin.h
6401 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
6402 (_mm512_mask_prefetch_i32gather_ps): Ditto.
6403 (_mm512_mask_prefetch_i64gather_pd): Ditto.
6404 (_mm512_mask_prefetch_i64gather_ps): Ditto.
6405 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
6406 (_mm512_prefetch_i32scatter_ps): Ditto.
6407 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
6408 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
6409 (_mm512_prefetch_i64scatter_pd): Ditto.
6410 (_mm512_prefetch_i64scatter_ps): Ditto.
6411 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
6412 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
6413 * config/i386/avx512vlintrin.h
6414 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
6415 (_mm_mmask_i32gather_ps): Ditto.
6416 (_mm256_mmask_i32gather_pd): Ditto.
6417 (_mm_mmask_i32gather_pd): Ditto.
6418 (_mm256_mmask_i64gather_ps): Ditto.
6419 (_mm_mmask_i64gather_ps): Ditto.
6420 (_mm256_mmask_i64gather_pd): Ditto.
6421 (_mm_mmask_i64gather_pd): Ditto.
6422 (_mm256_mmask_i32gather_epi32): Ditto.
6423 (_mm_mmask_i32gather_epi32): Ditto.
6424 (_mm256_mmask_i32gather_epi64): Ditto.
6425 (_mm_mmask_i32gather_epi64): Ditto.
6426 (_mm256_mmask_i64gather_epi32): Ditto.
6427 (_mm_mmask_i64gather_epi32): Ditto.
6428 (_mm256_mmask_i64gather_epi64): Ditto.
6429 (_mm_mmask_i64gather_epi64): Ditto.
6430 (_mm256_i32scatter_ps): Change __addr type to void*.
6431 (_mm256_mask_i32scatter_ps): Ditto.
6432 (_mm_i32scatter_ps): Ditto.
6433 (_mm_mask_i32scatter_ps): Ditto.
6434 (_mm256_i32scatter_pd): Ditto.
6435 (_mm256_mask_i32scatter_pd): Ditto.
6436 (_mm_i32scatter_pd): Ditto.
6437 (_mm_mask_i32scatter_pd): Ditto.
6438 (_mm256_i64scatter_ps): Ditto.
6439 (_mm256_mask_i64scatter_ps): Ditto.
6440 (_mm_i64scatter_ps): Ditto.
6441 (_mm_mask_i64scatter_ps): Ditto.
6442 (_mm256_i64scatter_pd): Ditto.
6443 (_mm256_mask_i64scatter_pd): Ditto.
6444 (_mm_i64scatter_pd): Ditto.
6445 (_mm_mask_i64scatter_pd): Ditto.
6446 (_mm256_i32scatter_epi32): Ditto.
6447 (_mm256_mask_i32scatter_epi32): Ditto.
6448 (_mm_i32scatter_epi32): Ditto.
6449 (_mm_mask_i32scatter_epi32): Ditto.
6450 (_mm256_i32scatter_epi64): Ditto.
6451 (_mm256_mask_i32scatter_epi64): Ditto.
6452 (_mm_i32scatter_epi64): Ditto.
6453 (_mm_mask_i32scatter_epi64): Ditto.
6454 (_mm256_i64scatter_epi32): Ditto.
6455 (_mm256_mask_i64scatter_epi32): Ditto.
6456 (_mm_i64scatter_epi32): Ditto.
6457 (_mm_mask_i64scatter_epi32): Ditto.
6458 (_mm256_i64scatter_epi64): Ditto.
6459 (_mm256_mask_i64scatter_epi64): Ditto.
6460 (_mm_i64scatter_epi64): Ditto.
6461 (_mm_mask_i64scatter_epi64): Ditto.
6462 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
6463 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
6464 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
6465 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
6466 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
6467 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
6468 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
6469 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
6470 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
6471 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
6472 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
6473 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
6474 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
6475 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
6476 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
6477 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
6478 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
6479 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
6480 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
6481 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
6482 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
6483 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
6484 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
6485 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
6486 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
6487 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
6488 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
6489 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
6490 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
6491 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
6492 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
6493 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
6494 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
6495 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
6496 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
6497 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
6498 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
6499 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
6500 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
6501 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
6502 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
6503 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
6504 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
6505 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
6506 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
6507 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
6508 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
6509 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
6510 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
6511 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
6512 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
6513 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
6514 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
6515 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
6516 definitions accordingly.
6518 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
6519 Kuan-Lin Chen <kuanlinchentw@gmail.com>
6522 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
6525 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
6528 * ira-conflicts.c (ira_build_conflicts): Update total conflict
6529 hard regs for inner regno.
6531 2017-01-17 Martin Liska <mliska@suse.cz>
6534 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
6535 assumption and add comment.
6537 2017-01-17 Nathan Sidwell <nathan@acm.org>
6539 * ipa-visibility.c (localize_node): New function, broken out of ...
6540 (function_and_variable_visibility): ... here. Call it.
6542 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
6545 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
6546 correctly set frequency of oudgoing edge.
6547 (duplicate_thread_path): Fix profile updating.
6549 2017-01-17 Jakub Jelinek <jakub@redhat.com>
6552 * configure.ac: Add GCC_BASE_VER.
6553 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
6554 version from BASE-VER file.
6555 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
6556 (gcc.o): Depend on $(BASEVER).
6557 * common.opt (dumpfullversion): New option.
6558 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
6559 * doc/invoke.texi: Document -dumpfullversion.
6560 * doc/install.texi: Document --with-gcc-major-version-only.
6561 * configure: Regenerated.
6563 2017-01-17 Richard Biener <rguenther@suse.de>
6565 PR tree-optimization/71433
6566 * tree-vrp.c (register_new_assert_for): Merge same asserts
6567 on all incoming edges.
6568 (process_assert_insertions_for): Handle insertions at the
6571 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
6573 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
6574 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
6576 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
6579 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
6582 2017-01-17 Alan Modra <amodra@gmail.com>
6585 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
6586 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
6587 symbolic stack limit when pic.
6589 2017-01-16 Martin Sebor <msebor@redhat.com>
6591 PR tree-optimization/78608
6592 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
6594 2017-01-16 Jeff Law <law@redhat.com>
6597 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
6598 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
6599 for several include directories that may be relative to sysroot.
6600 * config/i386/x-mingw32 (gplus_includedir): Define.
6601 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
6602 (native_system_includedir): Likewise.
6603 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
6604 override if TARGET_SYSTEM_ROOT is defined.
6605 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
6607 PR tree-optimization/79090
6608 PR tree-optimization/33562
6609 PR tree-optimization/61912
6610 PR tree-optimization/77485
6611 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
6612 and computed trims into the dump file.
6614 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
6616 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
6618 2017-01-16 Jakub Jelinek <jakub@redhat.com>
6621 * gimplify.c (gimplify_init_constructor): If want_value and
6622 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
6626 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
6627 sequence. Formatting fixes.
6628 (doloop_optimize): Formatting fixes.
6631 * gcc.c (debug_level_greater_than_spec_func): New function.
6632 (static_spec_functions): Add debug-level-gt spec function.
6633 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
6635 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
6636 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
6637 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
6638 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
6639 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
6640 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
6642 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
6644 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
6645 QImode fixups to general and mask registers only.
6647 2017-01-16 Carl Love <cel@us.ibm.com>
6649 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
6650 for built-in functions
6651 vector signed char vec_nabs (vector signed char)
6652 vector signed short vec_nabs (vector signed short)
6653 vector signed int vec_nabs (vector signed int)
6654 vector signed long long vec_nabs (vector signed long long)
6655 vector float vec_nabs (vector float)
6656 vector double vec_nabs (vector double)
6657 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
6659 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
6660 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
6661 * doc/extend.texi: Update the documentation file for the new built-in
6664 2017-01-16 Martin Sebor <msebor@redhat.com>
6666 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
6669 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6671 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
6672 UNSPEC_VSX__XXSPLTD to require special splat handling.
6674 2017-01-16 David Malcolm <dmalcolm@redhat.com>
6677 * system.h: Poison strndup.
6679 2017-01-16 Alan Modra <amodra@gmail.com>
6682 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
6685 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
6687 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
6689 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
6691 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
6692 call recog here. Assert that INSN_CODE (insn) is non-negative.
6694 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
6697 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
6699 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
6700 in the currently scheduled RTL fragment.
6702 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
6704 PR rtl-optimization/78751
6705 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
6708 2017-01-14 Jeff Law <law@redhat.com>
6710 PR tree-optimization/79090
6711 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
6712 variable length stores.
6713 (compute_trims): Delete dead assignment to *trim_tail.
6714 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
6717 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
6719 PR rtl-optimization/78626
6720 PR rtl-optimization/78727
6721 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
6722 of a block, and split such blocks after everything else is finished.
6724 2017-01-14 Alan Modra <amodra@gmail.com>
6727 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
6728 target legitimate_combined_insn.
6729 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
6730 (rs6000_legitimate_combined_insn): New function.
6731 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
6733 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
6734 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
6735 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
6737 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
6739 * doc/frontends.texi (G++ and GCC): Remove references to Java.
6741 2017-01-13 Jeff Law <law@redhat.com>
6743 PR tree-optimization/33562
6744 PR tree-optimization/61912
6745 PR tree-optimization/77485
6746 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
6748 (delete_dead_assignment): Likewise.
6749 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
6750 statement to delete_dead_call and delete_dead_assignment.
6752 2017-01-13 David Malcolm <dmalcolm@redhat.com>
6755 * substring-locations.c (format_warning_va): Strengthen case 1 so
6756 that both endpoints of the substring must be within the format
6757 range for just the substring to be printed.
6759 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
6761 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
6762 * config/i386/i386.c (ix86_target_string): Add missing options
6763 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
6764 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
6765 flags_other and ix86_target_other to flags2_other. Display unknown
6767 (ix86_valid_target_attribute_inner_p): Add missing options and
6768 reorder options by implied ISAs, as in ix86_target_string.
6770 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
6772 * hash-table.h (hash_table::too_empty_p): New function.
6773 (hash_table::expand): Use it.
6774 (hash_table::traverse): Likewise.
6775 (hash_table::empty_slot): Use sizeof (value_type) instead of
6776 sizeof (PTR) to convert bytes to elements. Shrink the table
6777 if the current size is excessive for the current number of
6780 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
6782 * ira-costs.c (record_reg_classes): Break from the inner loop
6783 early once alt_fail is known to be true. Update outer loop
6784 handling accordingly.
6786 2017-01-13 Jeff Law <law@redhat.com>
6788 * tree-ssa-dse.c (decrement_count): New function.
6789 (increment_start_addr, maybe_trim_memstar_call): Likewise.
6790 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
6791 when we know the partially dead statement is a mem* function.
6793 PR tree-optimization/61912
6794 PR tree-optimization/77485
6795 * tree-ssa-dse.c: Include expr.h.
6796 (maybe_trim_constructor_store): New function.
6797 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
6799 PR tree-optimization/33562
6800 PR tree-optimization/61912
6801 PR tree-optimization/77485
6802 * doc/invoke.texi: Document new dse-max-object-size param.
6803 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
6804 * tree-ssa-dse.c: Include params.h.
6805 (dse_store_status): New enum.
6806 (initialize_ao_ref_for_dse): New, partially extracted from
6808 (valid_ao_ref_for_dse, normalize_ref): New.
6809 (setup_live_bytes_from_ref, compute_trims): Likewise.
6810 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
6811 (maybe_trim_partially_dead_store): Likewise.
6812 (maybe_trim_complex_store): Likewise.
6813 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
6814 Track what bytes live from the original store. Return tri-state
6815 for dead, partially dead or live.
6816 (dse_dom_walker): Add constructor, destructor and new private members.
6817 (delete_dead_call, delete_dead_assignment): New extracted from
6819 (dse_optimize_stmt): Make a member of dse_dom_walker.
6820 Use initialize_ao_ref_for_dse.
6822 PR tree-optimization/33562
6823 PR tree-optimization/61912
6824 PR tree-optimization/77485
6825 * sbitmap.h (bitmap_count_bits): Prototype.
6826 (bitmap_clear_range, bitmap_set_range): Likewise.
6827 * sbitmap.c (bitmap_clear_range): New function.
6828 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
6830 2017-01-13 Martin Liska <mliska@suse.cz>
6833 * function.c (set_cfun): Add new argument force.
6834 * function.h (set_cfun): Likewise.
6835 * ipa-inline-transform.c (inline_call): Use the function when
6836 strict alising from is dropped for function we inline to.
6838 2017-01-13 Richard Biener <rguenther@suse.de>
6840 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
6841 for dumping GIMPLE INTEGER_CSTs.
6843 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6845 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
6846 to 201112L since C++17.
6848 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
6851 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
6852 if -fsanitize=kernel-address is present.
6854 2017-01-13 Richard Biener <rguenther@suse.de>
6856 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
6857 as _Literal ( type ) number in case usual suffixes do not
6858 preserve all information.
6860 2017-01-13 Richard Biener <rguenther@suse.de>
6862 PR tree-optimization/77283
6863 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
6864 and ssa-iterators.h.
6865 (is_feasible_trace): Implement a cost model based on joiner
6868 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6871 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
6872 char or short to __float128/_Float128 directly.
6874 2017-01-12 Martin Sebor <msebor@redhat.com>
6876 to -Wformat-overflow.
6877 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
6878 (min_bytes_remaining): Same.
6879 (get_string_length): Same.
6880 (format_string): Same.
6881 (format_directive): Same.
6883 (pass_sprintf_length::handle_gimple_call): Same.
6885 2017-01-12 Jakub Jelinek <jakub@redhat.com>
6887 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
6888 info.nowrite calls with no lhs that can't throw. Return bool
6889 whether gsi_remove has been called or not.
6890 (pass_sprintf_length::handle_gimple_call): Return bool whether
6891 try_substitute_return_value called gsi_remove. Formatting fix.
6892 (pass_sprintf_length::execute): Don't use gsi_remove if
6893 handle_gimple_call returned true.
6896 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
6897 be removed due to side-effects, don't remove following barrier nor
6898 turn the successor edge into fallthru edge.
6900 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6903 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
6904 element-reversing loads and stores as not swappable.
6906 2017-01-12 Nathan Sidwell <nathan@acm.org>
6907 Nicolai Stange <nicstange@gmail.com>
6909 * combine.c (try_combine): Don't ignore result of overlap checking
6910 loop. Combine overlap & asm check into single loop.
6912 2017-01-12 Richard Biener <rguenther@suse.de>
6914 * tree-pretty-print.c (dump_generic_node): Provide -gimple
6915 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
6917 2017-01-12 Richard Biener <rguenther@suse.de>
6919 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
6920 and TS_TARGET_OPTION directly derive from TS_BASE.
6921 * tree-core.h (tree_optimization_option): Derive from tree_base.
6922 (tree_target_option): Likewise.
6924 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
6926 * config/i386/i386.c (memory_address_length): Increase len
6927 only when rip_relative_addr_p returns false.
6929 2017-01-11 Julia Koval <julia.koval@intel.com>
6931 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
6932 (OPTION_MASK_ISA_SGX_SET): New.
6933 (ix86_handle_option): Handle OPT_msgx.
6934 * config.gcc: Added sgxintrin.h.
6935 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
6936 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
6937 * config/i386/i386.c (ix86_target_string): Add -msgx.
6939 (ix86_option_override_internal): Handle new options.
6940 (ix86_valid_target_attribute_inner_p): Add sgx.
6941 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
6942 * config/i386/i386.opt: Add msgx.
6943 * config/i386/sgxintrin.h: New file.
6944 * config/i386/x86intrin.h: Add sgxintrin.h.
6946 2017-01-11 Jakub Jelinek <jakub@redhat.com>
6949 * fold-const.c (maybe_nonzero_address): Return 1 for function
6951 (tree_single_nonzero_warnv_p): Don't handle function local objects
6955 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
6958 2017-01-11 David Malcolm <dmalcolm@redhat.com>
6961 * opts.c: Include "spellcheck.h"
6962 (struct string_fragment): New struct.
6963 (struct edit_distance_traits<const string_fragment &>): New
6965 (get_closest_sanitizer_option): New function.
6966 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
6968 2017-01-11 Jakub Jelinek <jakub@redhat.com>
6970 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
6972 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
6973 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
6974 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
6975 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
6976 for initial die_offset if dwarf_split_debug_info.
6977 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
6978 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
6980 (output_skeleton_debug_sections): Formatting fix. Use
6981 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
6982 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
6984 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
6986 * config/arm/cortex-a53.md: Add bypasses for
6988 (cortex_a53_r2f): Only use for transfers.
6989 (cortex_a53_f2r): Likewise.
6990 (cortex_a53_r2f_cvt): Add reservation for conversions.
6991 (cortex_a53_f2r_cvt): Likewise.
6993 2017-01-11 Tamar Christina <tamar.christina@arm.com>
6995 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
6996 to all inlined functions, change static to extern.
6998 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
7001 * config/arm/arm.c (legitimize_pic_address): Handle reference to
7003 (arm_assemble_integer): Likewise.
7005 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
7007 * config.gcc: Use new awk script to check CPU, FPU and architecture
7008 parameters for --with-... options.
7009 * config/arm/parsecpu.awk: New file
7010 * config/arm/arm-cpus.in: New file.
7011 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
7013 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
7015 * config/arm/t-arm: Update dependency rules.
7016 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
7017 of processing .def files.
7018 * config/arm/genopt.sh: Deleted.
7019 * config/arm/gentune.sh: Deleted.
7020 * config/arm/arm-cores.def: Deleted.
7021 * config/arm/arm-arches.def: Deleted.
7022 * config/arm/arm-fpus.def: Deleted.
7023 * config/arm/arm-tune.md: Regenerated.
7024 * config/arm/arm-tables.opt: Regenerated.
7025 * config/arm/arm-cpu.h: New generated file.
7026 * config/arm/arm-cpu-data.h: New generated file.
7027 * config/arm/arm-cpu-cdata.h: New generated file.
7029 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
7032 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
7034 (input_varpool_node): Unpack dynamically_initialized bit.
7036 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
7038 PR rtl-optimization/79032
7039 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
7040 the alignment of the adjusted memory reference against that of MODE,
7041 instead of the alignment of the original memory reference.
7043 2017-01-11 Martin Jambor <mjambor@suse.cz>
7045 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
7047 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
7048 decorated functions.
7050 2017-01-11 Richard Biener <rguenther@suse.de>
7052 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
7053 set range/nonnull info for PHI results. Do not set it on
7054 stmts marked for removal.
7056 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
7058 * expr.c (store_field): In the bitfield case, fetch the return value
7059 from the registers before applying a single big-endian adjustment.
7060 Always do a final load for a BLKmode value not larger than a word.
7062 2017-01-10 David Malcolm <dmalcolm@redhat.com>
7065 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
7066 that we correctly handle column numbers greater than
7067 LINE_MAP_MAX_COLUMN_NUMBER.
7069 2017-01-10 Martin Sebor <msebor@redhat.com>
7072 * gimple-ssa-sprintf.c (get_destination_size): Call
7073 {init,fini}object_sizes.
7074 * tree-object-size.c (addr_object_size): Adjust.
7075 (pass_through_call): Adjust.
7076 (pass_object_sizes::execute): Adjust.
7077 * tree-object-size.h (fini_object_sizes): Declare.
7079 2017-01-10 Martin Sebor <msebor@redhat.com>
7081 PR tree-optimization/78775
7082 * builtins.c (get_size_range): Move...
7083 * calls.c: ...to here.
7084 (alloc_max_size): Accept zero argument.
7085 (operand_signed_p): Remove.
7086 (maybe_warn_alloc_args_overflow): Call get_size_range.
7087 * calls.h (get_size_range): Declare.
7089 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
7091 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
7092 from TI's devices.csv file as of September 2016.
7093 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
7095 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
7097 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
7098 * doc/invoke.texi: Likewise.
7099 * doc/md.texi: Likewise.
7100 * doc/objc.texi: Likewise.
7102 2017-01-10 Joshua Conner <joshconner@google.com>
7104 * config/arm/fuchsia-elf.h: New file.
7105 * config/fuchsia.h: New file.
7106 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
7107 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
7109 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
7111 2016-01-10 Richard Biener <rguenther@suse.de>
7113 PR tree-optimization/79034
7114 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
7115 Propagate out degenerate PHIs in the joiner.
7117 2017-01-10 Martin Liska <mliska@suse.cz>
7119 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
7120 (sort_congruence_classes_by_decl_uid): Likewise.
7121 (sort_congruence_class_groups_by_decl_uid): Likewise.
7122 (sem_item_optimizer::merge_classes): Sort class, groups in these
7123 classes and members in the groups by DECL_UID of declarations.
7124 This would make merge operations stable.
7126 2017-01-10 Martin Liska <mliska@suse.cz>
7128 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
7129 usage of m_classes_vec.
7130 (sem_item_optimizer::~sem_item_optimizer): Likewise.
7131 (sem_item_optimizer::get_group_by_hash): Likewise.
7132 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
7133 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
7134 (sem_item_optimizer::verify_classes): Likewise.
7135 (sem_item_optimizer::process_cong_reduction): Likewise.
7136 (sem_item_optimizer::dump_cong_classes): Likewise.
7137 (sem_item_optimizer::merge_classes): Likewise.
7138 * ipa-icf.h (congruence_class_hash): Rename from
7139 congruence_class_group_hash. Remove declaration of m_classes_vec.
7141 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
7143 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
7144 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
7145 * config.gcc: Add avx512vpopcntdqintrin.h.
7146 * config/i386/avx512vpopcntdqintrin.h: New.
7147 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
7148 * config/i386/i386-builtin-types.def: Add new types.
7149 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
7150 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
7151 __builtin_ia32_vpopcountq_v8di_mask): New.
7152 * config/i386/i386-c.c (ix86_target_macros_internal): Define
7153 __AVX512VPOPCNTDQ__.
7154 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
7155 (PTA_AVX512VPOPCNTDQ): Define.
7156 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
7157 TARGET_AVX512VPOPCNTDQ_P): Define.
7158 * config/i386/i386.opt: Add mavx512vpopcntdq.
7159 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
7160 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
7162 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7165 * predict.def (PRED_CALL): Set to 67.
7167 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
7169 * expr.c (store_field): In the bitfield case, if the value comes from
7170 a function call and is of an aggregate type returned in registers, do
7171 not modify the field mode; extract the value in all cases if the mode
7172 is BLKmode and the size is not larger than a word.
7174 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
7177 * config/i386/cpuid.h: Fix undefined behavior.
7179 2017-01-04 Jeff Law <law@redhat.com>
7181 PR tree-optimization/79007
7182 PR tree-optimization/67955
7183 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
7184 conservative for pt.null when flag_non_call_exceptions is on.
7186 2017-01-09 Jakub Jelinek <jakub@redhat.com>
7188 PR translation/79019
7189 PR translation/79020
7190 * params.def (PARAM_INLINE_MIN_SPEEDUP,
7191 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
7192 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
7194 * config/avr/avr.opt (maccumulate-args): Likewise.
7195 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
7196 * common.opt (freport-bug): Likewise.
7197 * cif-code.def (CIF_FINAL_ERROR): Likewise.
7198 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
7199 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
7200 translatable string.
7201 * config/i386/i386.c (function_value_32): Likewise.
7202 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
7203 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
7205 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
7206 * common/config/msp430/msp430-common.c (msp430_handle_option):
7208 * symtab.c (symtab_node::verify_base): Likewise.
7209 * opts.c (set_debug_level): Likewise.
7210 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
7211 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
7212 missing whitespace to translatable strings.
7213 * config/avr/avr.md (bswapsi2): Fix typo in comment.
7214 * config/sh/superh.h: Likewise.
7215 * config/i386/xopintrin.h: Likewise.
7216 * config/i386/znver1.md: Likewise.
7217 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
7218 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
7219 * double-int.h (struct double_int): Likewise.
7220 * double-int.c (div_and_round_double): Likewise.
7221 * wide-int.cc: Likewise.
7222 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
7223 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
7224 * cfgcleanup.c (crossjumps_occured): Renamed to ...
7225 (crossjumps_occurred): ... this.
7226 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
7229 PR tree-optimization/78899
7230 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
7231 returning bool return struct loop *, NULL for failure and the new
7233 (versionable_outer_loop_p): Don't version outer loop if it has
7234 dont_vectorized bit set.
7235 (tree_if_conversion): When versioning outer loop, ensure
7236 tree_if_conversion is performed also on the inner loop of the
7237 non-vectorizable outer loop copy.
7238 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
7239 LOOP_VECTORIZED in inner loop of the scalar outer loop and
7240 prevent vectorization of it.
7241 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
7242 the outer loop vectorization of the non-scalar version is attempted
7243 before vectorization of the inner loop in scalar version. If
7244 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
7245 vectorization of its inner loop.
7246 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
7247 has 2 inner loops, rename also on edges from bb whose single pred
7248 is outer_loop->header. Fix typo in function comment.
7250 2017-01-09 Martin Sebor <msebor@redhat.com>
7253 * asan.c (asan_emit_stack_protection): Increase local buffer size
7254 to avoid snprintf truncation warning.
7256 2017-01-09 Andrew Pinski <apinski@cavium.com>
7258 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
7259 to reference thunderx2t99 for the tuning structure
7260 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
7262 (thunderx2t99_extra_costs): This.
7263 * config/aarch64/aarch64-tune.md: Regenerate.
7264 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
7265 (vulcan_addrcost_table): This.
7266 (vulcan_regmove_cost): Rename to ...
7267 (thunderx2t99_regmove_cost): This.
7268 (vulcan_vector_cost): Rename to ...
7269 (thunderx2t99_vector_cost): this.
7270 (vulcan_branch_cost): Rename to ...
7271 (thunderx2t99_branch_cost): This.
7272 (vulcan_tunings): Rename to ...
7273 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
7274 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
7276 2017-01-09 Martin Jambor <mjambor@suse.cz>
7280 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
7281 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
7282 (propagate_vr_accross_jump_function): Use the above function for all
7283 value range computations for pass-through jump functions and type
7284 converasion from explicit value range values.
7285 (ipcp_propagate_stage): Do not attempt to deduce types of formal
7286 parameters from TYPE_ARG_TYPES.
7287 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
7288 (ipa_write_node_info): Stream type of the actual argument.
7289 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
7291 2017-01-09 Martin Liska <mliska@suse.cz>
7294 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
7295 (lookup_compiler): Do not show error message with have_E.
7297 2017-01-09 Jakub Jelinek <jakub@redhat.com>
7299 PR tree-optimization/78938
7300 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
7301 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
7302 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
7303 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
7306 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7308 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
7311 2017-01-09 Richard Biener <rguenther@suse.de>
7313 PR tree-optimization/78997
7314 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
7315 name condition properly.
7317 2017-01-09 Richard Biener <rguenther@suse.de>
7320 * dwarf2out.c (is_cxx): New overload with context.
7321 (is_naming_typedef_decl): Use it.
7323 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7325 * invoke.texi (Option Summary): Correct spacing in option lists
7326 and add line breaks to fix over-long lines.
7328 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7332 * extend.texi (Common Variable Attributes): Add xref to GCC
7333 Internals manual to explain mode attribute keywords.
7335 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7338 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
7339 and Preprocessor Options.
7340 (Options for Linking): Document -pthread here....
7341 (RS/6000 and PowerPC Options): ...not here.
7342 (Solaris 2 Options): ...or here.
7343 * doc/cppopts.texi: Document -pthread.
7345 2017-01-08 Martin Sebor <msebor@redhat.com>
7348 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
7349 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
7350 New member functions.
7351 (format_directive): Used them.
7353 (pass_sprintf_length::handle_gimple_call): Same.
7354 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
7355 to avoid truncation for any argument.
7356 (extract_affine_mul): Same.
7357 * tree.c (get_file_function_name): Same.
7359 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7362 * predict.def (PRED_INDIR_CALL): Set to 86.
7364 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
7366 PR preprocessor/54124
7367 * doc/cppopts.texi: Reformat -d subtable to list the full name
7368 of the options. Add cross-reference to the docs for the general
7369 compiler -d options.
7370 * doc/invoke.texi (Developer Options): Add cross-reference to the
7371 preprocessor-specific -d option documentation.
7373 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
7375 PR preprocessor/13498
7376 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
7377 redudant material, and reflect new command-line options.
7378 (System Headers): Likewise.
7380 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
7382 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
7383 -isystem, and -idirafter. Copy-edit.
7384 * doc/cppopts.texi: Copy-edit. Remove contradiction about
7385 default for -ftrack-macro-expansion. Delete obsolete and
7386 badly-formatted implementation details about -fdebug-cpp output.
7387 * doc/cppwarnopts.texi: Copy-edit.
7389 2017-01-07 David Malcolm <dmalcolm@redhat.com>
7392 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
7393 that the transition from a max line width >= 1<<10 to narrower
7394 lines works correctly.
7396 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
7398 * doc/options.texi (PerFunction): New.
7399 * opt-functions.awk (switch_flags): Map both Optimization and
7400 PerFunction to CL_OPTIMIZATION.
7401 * opth-gen.awk: Test for PerFunction flag along with
7403 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
7404 it only when the latter is present. Skip those that don't in
7405 the hash function generator.
7406 * common.opt (fvar-tracking): Mark as PerFunction instead of
7408 (fvar-tracking-assignments): Likewise.
7409 (fvar-tracking-assignments-toggle): Likewise.
7410 (fvar-tracking-uninit): Likewise.
7412 2017-01-07 Jakub Jelinek <jakub@redhat.com>
7414 PR translation/79018
7415 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
7418 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
7421 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
7422 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
7423 TARGET_LONG_JUMP_TABLE_OFFSETS.
7424 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
7425 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
7426 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
7427 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
7428 * config/m68k/m68k.md (tablejump expander): Likewise.
7429 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
7430 TARGET_LONG_JUMP_TABLE_OFFSETS.
7431 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
7432 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
7434 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
7435 David Holsgrove <david.holsgrove@xilinx.com>
7437 * common/config/microblaze/microblaze-common.c
7438 (TARGET_EXCEPT_UNWIND_INFO): Remove.
7439 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
7441 * config/microblaze/microblaze.c (microblaze_must_save_register)
7442 (microblaze_expand_epilogue, microblaze_return_addr): Handle
7444 (microblaze_eh_return): New function.
7445 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
7446 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
7447 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
7448 * config/microblaze/microblaze.md (eh_return): New pattern.
7450 2017-01-06 Jakub Jelinek <jakub@redhat.com>
7452 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
7453 GCC_DIAGNOSTIC_STRINGIFY): Define.
7455 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
7457 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7459 * config/arm/arm.md (<mcrr>): New.
7461 * config/arm/arm.c (arm_arch5te): New.
7462 (arm_option_override): Set arm_arch5te.
7463 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
7465 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
7466 (arm_mcrr_qualifiers): ... this. New.
7467 (MRRC_QUALIFIERS): Define to...
7468 (arm_mrrc_qualifiers): ... this. New.
7469 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
7471 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
7472 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
7473 (MRRCI, mrrc, MRRC): New.
7474 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
7475 VUNSPEC_MRRC2): New.
7477 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7479 * config/arm/arm.md (<mcr>): New.
7481 * config/arm/arm.c (arm_coproc_builtin_available): Add
7482 support for mcr, mrc, mcr2 and mrc2.
7483 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
7484 (arm_mcr_qualifiers): ... this. New.
7485 (MRC_QUALIFIERS): Define to ...
7486 (arm_mrc_qualifiers): ... this. New.
7487 (MCR_QUALIFIERS): Define to ...
7488 (arm_mcr_qualifiers): ... this. New.
7489 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
7491 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
7492 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
7493 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
7496 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7498 * config/arm/arm.md (*ldc): New.
7502 * config/arm/arm.c (arm_coproc_builtin_available): Add
7503 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
7504 (arm_coproc_ldc_stc_legitimate_address): New.
7505 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
7506 'qualifier_const_pointer'.
7507 (LDC_QUALIFIERS): Define to...
7508 (arm_ldc_qualifiers): ... this. New.
7509 (STC_QUALIFIERS): Define to...
7510 (arm_stc_qualifiers): ... this. New.
7511 * config/arm/arm-protos.h
7512 (arm_coproc_ldc_stc_legitimate_address): New.
7513 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
7514 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
7515 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
7516 stc2, stcl, stc2l): New.
7517 * config/arm/constraints.md (Uz): New.
7518 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
7519 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
7520 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
7521 VUNSPEC_STC2L): New.
7523 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7525 * config/arm/arm.md (<cdp>): New.
7526 * config/arm/arm.c (neon_const_bounds): Rename this ...
7527 (arm_const_bounds): ... this.
7528 (arm_coproc_builtin_available): New.
7529 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
7530 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
7531 (CDP_QUALIFIERS): Define to...
7532 (arm_cdp_qualifiers): ... this. New.
7534 (arm_expand_builtin_args): Add case for 6 arguments.
7535 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
7536 (arm_const_bounds): ... this.
7537 (arm_coproc_builtin_available): New.
7538 * config/arm/arm_acle.h (__arm_cdp): New.
7540 * config/arm/arm_acle_builtins.def (cdp): New.
7542 * config/arm/iterators.md (CDPI,CDP,cdp): New.
7543 * config/arm/neon.md: Rename all 'neon_const_bounds' to
7545 * config/arm/types.md (coproc): New.
7546 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
7547 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
7548 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
7549 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
7551 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7553 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
7554 (UBINOP_QUALIFIERS): New.
7556 (acle_builtin_data): New. Change comment.
7557 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
7558 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
7559 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
7560 arm_acle_builtins.def.
7561 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
7562 (arm_init_acle_builtins): New.
7563 (CRC32_BUILTIN): Remove.
7564 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
7565 crc32cb, crc32ch and crc32cw.
7566 (arm_init_crc32_builtins): Remove.
7567 (arm_init_builtins): Use arm_init_acle_builtins rather
7568 than arm_init_crc32_builtins.
7569 (arm_expand_acle_builtin): New.
7570 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
7571 * config/arm/arm_acle_builtins.def: New.
7573 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7575 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
7576 (arm_builtin_datum): ... this.
7577 (arm_init_neon_builtin): Rename to ...
7578 (arm_init_builtin): ... this. Add a new parameters PREFIX
7579 and USE_SIG_IN_NAME.
7580 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
7581 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
7582 'arm_builtin_datum'.
7583 (arm_init_vfp_builtins): Likewise.
7584 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
7585 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
7586 (arm_expand_neon_args): Rename to ...
7587 (arm_expand_builtin_args): ... this. Rename builtin_arg
7588 enum values and differentiate between ARG_BUILTIN_MEMORY
7589 and ARG_BUILTIN_NEON_MEMORY.
7590 (arm_expand_neon_builtin_1): Rename to ...
7591 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
7592 values, arm_expand_builtin_args and add bool parameter NEON.
7593 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
7594 (arm_expand_vfp_builtin): Likewise.
7595 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
7597 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7600 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
7601 * predict.c (tree_estimate_probability_bb): Reverse direction of
7602 polymorphic call predictor.
7604 2017-01-06 David Malcolm <dmalcolm@redhat.com>
7606 * passes.c (execute_one_pass): Split out pass-skipping logic into...
7607 (determine_pass_name_match): ...this new function and...
7608 (should_skip_pass_p): ...this new function.
7610 2017-01-06 Nathan Sidwell <nathan@acm.org>
7612 * ipa-visibility.c (function_and_variable_visibility): Reformat
7613 comments and long lines. Remove extrneous if.
7614 * symtab.c (symtab_node::make_decl_local): Fix code format.
7615 (symtab_node::set_section_for_node): Fix comment typo.
7617 2017-01-06 Martin Liska <mliska@suse.cz>
7620 * lra-constraints.c: Rename invariant to lra_invariant.
7621 * predict.c (set_even_probabilities): Initialize e to NULL.
7623 2017-01-05 Martin Sebor <msebor@redhat.com>
7625 PR tree-optimization/78910
7626 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
7627 (format_integer): Correct off-by-one error in the handling
7628 of precision with negative numbers in signed conversions..
7630 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
7632 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
7634 2017-01-05 Jakub Jelinek <jakub@redhat.com>
7636 PR tree-optimization/71016
7637 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
7638 factor_out_conditional_conversion. Formatting fix.
7639 (factor_out_conditional_conversion): Add cond_stmt argument.
7640 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
7641 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
7644 2017-01-05 David Malcolm <dmalcolm@redhat.com>
7646 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
7647 read-rtl-function.o, and selftest-rtl.o.
7648 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
7649 (selftest::aarch64_test_loading_full_dump): New function.
7650 (selftest::aarch64_run_selftests): New function.
7651 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
7652 selftest::aarch64_run_selftests.
7653 * config/i386/i386.c
7654 (selftest::ix86_test_loading_dump_fragment_1): New function.
7655 (selftest::ix86_test_loading_call_insn): New function.
7656 (selftest::ix86_test_loading_full_dump): New function.
7657 (selftest::ix86_test_loading_unspec): New function.
7658 (selftest::ix86_run_selftests): Call the new functions.
7659 * emit-rtl.c (maybe_set_max_label_num): New function.
7660 * emit-rtl.h (maybe_set_max_label_num): New decl.
7661 * function.c (instantiate_decls): Guard call to
7662 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
7663 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
7665 * gensupport.c (gen_reader::gen_reader): Pass "false"
7666 for new "compact" param of rtx_reader.
7667 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
7668 rather than an empty string for NULL strings.
7669 * read-md.c: Potentially include config.h rather than bconfig.h.
7670 Wrap include of errors.h with #ifdef GENERATOR_FILE.
7671 (have_error): New global, copied from errors.c.
7672 (md_reader::read_name): Rename to...
7673 (md_reader::read_name_1): ...this, adding "out_loc" param,
7674 and converting "missing name or number" to returning false, rather
7676 (md_reader::read_name): Reimplement in terms of read_name_1.
7677 (md_reader::read_name_or_nil): New function.
7678 (md_reader::read_string): Handle "(nil)" by returning NULL.
7679 (md_reader::md_reader): Add new param "compact".
7680 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
7681 (md_reader::read_file): New method.
7682 * read-md.h (md_reader::md_reader): Add new param "compact".
7683 (md_reader::read_file): New method.
7684 (md_reader::is_compact): New accessor.
7685 (md_reader::read_name): Convert return type from void to file_location.
7686 (md_reader::read_name_or_nil): New decl.
7687 (md_reader::read_name_1): New decl.
7688 (md_reader::m_compact): New field.
7689 (noop_reader::noop_reader): Pass "false" for new "compact" param
7691 (rtx_reader::rtx_reader): Add new "compact" param.
7692 (rtx_reader::read_rtx_operand): Make virtual and convert return
7693 type from void to rtx.
7694 (rtx_reader::read_until): New decl.
7695 (rtx_reader::handle_any_trailing_information): New virtual function.
7696 (rtx_reader::postprocess): New virtual function.
7697 (rtx_reader::finalize_string): New virtual function.
7698 (rtx_reader::m_in_call_function_usage): New field.
7699 (rtx_reader::m_reuse_rtx_by_id): New field.
7700 * read-rtl-function.c: New file.
7701 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
7702 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
7703 (selftest::verify_three_block_rtl_cfg): New decl.
7704 * read-rtl-function.h: New file.
7705 * read-rtl.c: Potentially include config.h rather than bconfig.h.
7706 For host, include function.h, memmodel.h, and emit-rtl.h.
7707 (one_time_initialization): New function.
7708 (struct compact_insn_name): New struct.
7709 (compact_insn_names): New array.
7710 (find_code): Handle insn codes in compact dumps.
7711 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
7712 (bind_subst_iter_and_attr): Likewise.
7713 (add_condition_to_string): Likewise.
7714 (add_condition_to_rtx): Likewise.
7715 (apply_attribute_uses): Likewise.
7716 (add_current_iterators): Likewise.
7717 (apply_iterators): Likewise.
7718 (initialize_iterators): Guard usage of apply_subst_iterator with
7719 #ifdef GENERATOR_FILE.
7720 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
7721 (md_reader::read_mapping): Likewise.
7722 (add_define_attr_for_define_subst): Likewise.
7723 (add_define_subst_attr): Likewise.
7724 (read_subst_mapping): Likewise.
7725 (check_code_iterator): Likewise.
7726 (rtx_reader::read_rtx): Likewise. Move one-time initialization
7728 (one_time_initialization): New function.
7729 (rtx_reader::read_until): New method.
7730 (read_flags): New function.
7731 (parse_reg_note_name): New function.
7732 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
7733 Handle reuse_rtx ids.
7734 Wrap iterator lookup within #ifdef GENERATOR_FILE.
7735 Add parsing support for RTL dumps, mirroring the special-cases in
7736 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
7737 values, and calling handle_any_trailing_information.
7738 (rtx_reader::read_rtx_operand): Convert return type from void
7739 to rtx, returning return_rtx. Handle case 'e'. Call
7740 finalize_string on XSTR and XTMPL fields.
7741 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
7742 "(nil)" values were omitted. Call the postprocess vfunc on the
7744 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
7745 class ctor. Initialize m_in_call_function_usage. Call
7746 one_time_initialization.
7747 * rtl-tests.c (selftest::test_uncond_jump): Call
7748 set_new_first_and_last_insn.
7749 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
7750 * selftest-rtl.c: New file.
7751 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
7752 (selftest::get_insn_by_uid): New decl.
7753 * selftest-run-tests.c (selftest::run_tests): Call
7754 read_rtl_function_c_tests.
7755 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
7756 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
7759 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
7761 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
7762 operands in a special way. Assert that pos+len <= mode precision.
7764 2017-01-05 Jakub Jelinek <jakub@redhat.com>
7766 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
7767 3 argument Alias with unlimited for the negative form.
7768 (fno-vect-cost-model): Removed.
7770 2017-01-05 Martin Liska <mliska@suse.cz>
7772 * hsa-gen.c (gen_hsa_divmod): New function.
7773 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
7775 2017-01-05 Martin Liska <mliska@suse.cz>
7778 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
7781 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7783 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
7784 small constant length operands.
7786 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7788 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
7789 between loop iterations.
7791 2017-01-05 Martin Liska <mliska@suse.cz>
7794 * gimplify.c (gimplify_decl_expr): Compare to
7795 asan_poisoned_variables instread of checking flags.
7796 (gimplify_target_expr): Likewise.
7797 (gimplify_expr): Likewise.
7798 (gimplify_function_tree): Conditionally initialize
7799 asan_poisoned_variables.
7801 2017-01-04 Jeff Law <law@redhat.com>
7803 PR tree-optimizatin/78812
7804 * rtl.h (contains_mem_rtx_p): Prototype.
7805 * ifcvt.c (containts_mem_rtx_p): Move from here to...
7806 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
7807 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
7808 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
7809 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
7811 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7813 * input.c (assert_char_at_range): Default-initialize actual_range.
7815 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7817 * df-scan.c (df_ref_create_structure): Make regno unsigned,
7818 to match the caller.
7820 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7822 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
7823 insns after final jump in test to emit dummy move.
7825 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7827 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
7828 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
7830 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
7832 * multiple_target.c (create_dispatcher_calls): Init e_next.
7833 * tree-ssa-loop-split.c (split_loop): Init border.
7834 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
7837 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
7842 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
7843 (altivec_register_operand): Do not return true if the operand
7844 contains a SUBREG mixing SImode and SFmode.
7845 (vsx_register_operand): Likewise.
7846 (vsx_reg_sfsubreg_ok): New predicate.
7847 (vfloat_operand): Do not return true if the operand contains a
7848 SUBREG mixing SImode and SFmode.
7849 (vint_operand): Likewise.
7850 (vlogical_operand): Likewise.
7851 (gpc_reg_operand): Likewise.
7852 (int_reg_operand): Likewise.
7853 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
7854 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
7855 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
7857 (rs6000_emit_move_si_sf_subreg): New helper function.
7858 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
7859 fixup SUBREGs involving SImode and SFmode.
7860 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
7861 numbers for the new peephole2 optimization.
7862 (peephole2 for SFmode unions): New peephole2 to optimize cases in
7863 the GLIBC math library that do AND/IOR/XOR operations on single
7864 precision floating point.
7865 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
7866 target macros to say whether we need to avoid SUBREGs mixing
7868 (TARGET_ALLOW_SF_SUBREG): Likewise.
7869 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
7870 (UNSPEC_SI_FROM_SF): Likewise.
7871 (iorxor): Change spacing.
7872 (and_ior_xor): New iterator for AND, IOR, and XOR.
7873 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
7874 (movdi_from_sf_zero_ext): Likewise.
7875 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
7876 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
7877 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
7878 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
7879 (fms<mode>4): Likewise.
7880 (fnma<mode>4): Likewise.
7881 (fnms<mode>4): Likewise.
7882 (nfma<mode>4): Likewise.
7883 (nfms<mode>4): Likewise.
7885 2017-01-04 Marek Polacek <polacek@redhat.com>
7888 * doc/invoke.texi: Document -Wpointer-compare.
7890 2017-01-04 Jakub Jelinek <jakub@redhat.com>
7892 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
7895 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
7896 descriptions for -gdwarf-5 and emit them as uleb128 instead of
7899 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
7902 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
7903 documentation of the powerpc_popcntb_ok attribute.
7904 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
7905 code to issue warning messages if a requested CPU configuration is
7906 not supported by the binary (assembler and loader) toolchain.
7907 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
7908 made to define a built-in function that has been disabled.
7909 (paired_init_builtins): Add assertion to prevent ICE if attempt is
7910 made to define a built-in function that has been disabled.
7911 (altivec_init_builtins): Add comment explaining why definition
7912 of the DST built-in functions is not preceded by an assertion
7913 check. Add assertions to prevent ICE if attempts are made to
7914 define an altivec predicate or an abs* built-in function that has
7916 (htm_init_builtins): Add comment explaining why definition of the
7917 htm built-in functions is not preceded by an assertion check.
7919 2017-01-04 Jeff Law <law@redhat.com>
7921 PR tree-optimizatin/67955
7922 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
7923 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
7924 the points-to solution does not include pt_null. Use DECL_PT_UID
7927 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
7929 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
7930 Use gen_int_mode instead of gen_lopwart for const_int operands.
7932 2017-01-04 Jakub Jelinek <jakub@redhat.com>
7934 PR tree-optimization/71563
7935 * match.pd: Simplify X << Y into X if Y is known to be 0 or
7936 out of range value - has low bits known to be zero.
7938 2017-01-04 Alan Modra <amodra@gmail.com>
7940 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
7941 * configure: Regenerate.
7942 * config.in: Regenerate.
7944 2017-01-04 Jakub Jelinek <jakub@redhat.com>
7947 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
7948 a substring of the message, but strcmp with the whole message. Ifdef
7949 ENABLE_NLS, translate the message first using dgettext.
7951 2017-01-03 Jeff Law <law@redhat.com>
7953 PR tree-optimizatin/78856
7954 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
7955 (mark_threaded_blocks): Remove code to truncate thread paths that
7956 cross multiple loop headers. Instead invalidate the cached loop
7957 iteration information and handle case of a thread path walking
7958 into an irreducible region.
7960 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
7963 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
7964 assertions. Add support for doing the signbit if the IEEE 128-bit
7965 floating point value is in a GPR.
7966 * config/rs6000/rs6000.md (Fsignbit): Delete.
7967 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
7968 Update the length attribute if the value is in a GPR.
7969 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
7970 the sign or zero extension instruction, since the value is always 0/1.
7971 (signbit<mode>2_dm2): Delete using <Fsignbit>.
7974 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
7975 extracting SImode to a GPR register so that we can generate a
7976 store, limit the vector to be in a traditional Altivec register
7977 for the vextuwrx instruction.
7979 2017-01-03 Ian Lance Taylor <iant@google.com>
7981 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
7983 2017-01-03 Martin Sebor <msebor@redhat.com>
7985 PR tree-optimization/78696
7986 * gimple-ssa-sprintf.c (format_floating): Correct handling of
7987 precision. Use MPFR for %f for greater fidelity. Correct handling
7989 (pass_sprintf_length::compute_format_length): Set width and precision
7990 specified by asrerisk to void_node for vararg functions.
7991 (try_substitute_return_value): Adjust dump output.
7993 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
7995 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
7997 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
7999 * doc/invoke.texi (SPARC options): Document -mlra as the default.
8000 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
8001 -mlra/-mno-lra was passed to the compiler.
8003 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
8005 PR rtl-optimization/65618
8006 * emit-rtl.c (try_split): Move initialization of "before" and
8007 "after" to just before the call to emit_insn_after_setloc.
8009 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
8011 * doc/md.texi (Standard Names): Remove reference to Java frontend.
8013 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
8015 * dwarf2out.c (gen_enumeration_type_die): When
8016 -gno-strict-dwarf, add a DW_AT_encoding attribute.
8018 2017-01-03 Jakub Jelinek <jakub@redhat.com>
8020 PR tree-optimization/78965
8021 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
8022 Change first argument from const call_info & to call_info &. For %n
8023 set info.nowrite to false.
8026 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
8027 possibly throwing calls.
8029 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
8030 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
8031 and fns handling, rather than in a separate case SSA_NAME.
8033 2017-01-02 Jeff Law <law@redhat.com>
8035 * config/darwin-driver.c (darwin_driver_init): Const-correctness
8036 fixes for first_period and second_period variables.
8038 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
8041 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
8042 (*insvqi_1): New insn pattern.
8043 (*insvqi_1_mem_rex64): Ditto.
8045 (*insvqi_3): Rename from *insvqi.
8047 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
8049 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
8051 * doc/cfg.texi (Edges): Remove reference to Java.
8052 (Maintaining the CFG): Ditto.
8054 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8057 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
8058 transparent aliases.
8060 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8063 * predict.def (PRED_CALL): Update hitrate.
8064 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
8065 * predict.c (tree_estimate_probability_bb): Split CALL predictor
8066 into direct/indirect/polymorphic variants.
8068 2017-01-01 Jakub Jelinek <jakub@redhat.com>
8070 Update copyright years.
8072 * gcc.c (process_command): Update copyright notice dates.
8073 * gcov-dump.c (print_version): Ditto.
8074 * gcov.c (print_version): Ditto.
8075 * gcov-tool.c (print_version): Ditto.
8076 * gengtype.c (create_file): Ditto.
8077 * doc/cpp.texi: Bump @copying's copyright year.
8078 * doc/cppinternals.texi: Ditto.
8079 * doc/gcc.texi: Ditto.
8080 * doc/gccint.texi: Ditto.
8081 * doc/gcov.texi: Ditto.
8082 * doc/install.texi: Ditto.
8083 * doc/invoke.texi: Ditto.
8085 Copyright (C) 2017 Free Software Foundation, Inc.
8087 Copying and distribution of this file, with or without modification,
8088 are permitted in any medium without royalty provided the copyright
8089 notice and this notice are preserved.