1 2017-05-02 Release Manager
5 2017-05-02 Richard Biener <rguenther@suse.de>
7 PR tree-optimization/80591
9 2017-04-10 Richard Biener <rguenther@suse.de>
11 * tree-ssa-structalias.c (find_func_aliases): Properly handle
14 2017-04-28 Jakub Jelinek <jakub@redhat.com>
17 * cgraph.h (symtab_node::debug_symtab): No longer inline.
18 * symtab.c (symtab_node::debug_symtab): Move definition here.
20 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
23 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
24 that the logic for permitting reciprocal estimates matches that
27 2017-04-27 Jakub Jelinek <jakub@redhat.com>
30 * tree.c (type_cache_hasher::equal): Only compare
31 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
32 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33 non-aggregate element types.
34 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
35 about the flag on ARRAY_TYPEs in the comment, formatting fix.
38 * reg-stack.c (emit_swap_insn): If i1src mentions the stack pointer,
39 punt if tmp contains autoinc of stack pointer.
42 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
43 (aarch64_function_arg_alignment): Return unsigned int again, but still
44 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
45 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
46 Don't emit -Wpsabi note.
47 (aarch64_function_arg_boundary): Likewise.
48 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
51 2017-04-25 Martin Sebor <msebor@redhat.com>
53 PR tree-optimization/80497
54 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
55 constants are representable in HOST_WIDE_INT.
56 (parse_directive): Ditto.
58 2017-04-25 Marek Polacek <polacek@redhat.com>
60 2017-04-25 Marek Polacek <polacek@redhat.com>
61 Backport from mainline
64 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
65 first argument to type.
67 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
68 Jakub Jelinek <jakub@redhat.com>
71 * config/arm/arm.c: Include gimple.h.
72 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
73 returns negative, increment ncrn only if it returned positive.
74 (arm_needs_doubleword_align): Return int instead of bool,
75 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
76 members, but if there is any such non-FIELD_DECL
77 > PARM_BOUNDARY aligned decl, return -1 instead of false.
78 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
79 returns negative, increment nregs only if it returned positive.
80 (arm_setup_incoming_varargs): Likewise.
81 (arm_function_arg_boundary): Emit -Wpsabi note if
82 arm_needs_doubleword_align returns negative, return
83 DOUBLEWORD_ALIGNMENT only if it returned positive.
85 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
87 Backport from mainline
89 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
90 type checks to test for compatibility instead of equality.
92 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
93 Jakub Jelinek <jakub@redhat.com>
96 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
98 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
99 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
100 the alignment computation, but return their maximum in warn_alignment.
101 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
102 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
104 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
105 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
108 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
110 Backport from mainline
111 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
114 * config/s390/vector.md: Split MEM->GPR vector moves for
115 non-s_operand addresses.
117 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
119 Backport from mainline
120 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
123 * config/s390/predicates.md (reload_const_wide_int_operand): New
125 * config/s390/s390.md ("movti"): Remove d/P alternative.
126 ("movti_bigconst"): New pattern definition.
128 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
130 Backport from maineline
131 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
134 * s390-protos.h (s390_expand_cs_hqi): Removed.
135 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
136 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
137 modes as well as CCZ1mode and CCZmode.
138 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
139 signature of s390_emit_compare_and_swap.
140 (s390_expand_cs_hqi): Likewise, make static.
141 (s390_expand_cs_tdsi): Generate an explicit compare before trying
142 compare-and-swap, in some cases.
143 (s390_expand_cs): Wrapper function.
144 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
146 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
147 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
148 patterns for small and large integers. Forbid symref memory operands.
149 Move expander to s390.c. Require cc register.
150 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
151 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
152 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
153 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
154 symref memory operands. Remove CC mode and call s390_match_ccmode
156 ("atomic_exchange<mode>"): Allow and implement all integer modes.
158 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
160 Backport from mainline
161 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
163 * config/s390/s390.md (define_peephole2): New peephole to help
164 combining the load-and-test pattern with volatile memory.
167 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
169 Backport from mainline
170 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
172 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
173 with CCZmode for TARGET_Z196.
175 2017-04-25 Jakub Jelinek <jakub@redhat.com>
177 PR rtl-optimization/80501
178 * combine.c (make_compound_operation_int): Set subreg_code to SET
179 even for AND with mask of the sign bit of mode.
181 PR rtl-optimization/80500
182 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
185 2017-04-24 Martin Liska <mliska@suse.cz>
187 Backport from mainline
188 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
191 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
193 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
195 Backport from mainline
196 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
198 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear
199 and lastprivate clauses in SIMT case.
201 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
203 Backport from mainline
204 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
206 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
207 uninitialized variable warning to avoid buffer overrun.
209 2017-04-20 Jakub Jelinek <jakub@redhat.com>
211 * DEV-PHASE: Set to prerelease.
213 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
215 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
216 priority .init_array and .fini_array section with SECTION_NOTYPE
219 2017-04-20 Jakub Jelinek <jakub@redhat.com>
222 * tree.h (build_array_type): Add typeless_storage default argument.
223 * tree.c (type_cache_hasher::equal): Also compare
224 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
225 (build_array_type): Add typeless_storage argument, set
226 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
228 (build_nonshared_array_type): Adjust build_array_type_1 caller.
229 (build_array_type): Likewise. Add typeless_storage argument.
231 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
232 Jakub Jelinek <jakub@redhat.com>
234 PR tree-optimization/80426
235 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
236 operation on symbolic operands, also compute the overflow for the
237 invariant part when the operation degenerates into a negation.
239 2017-04-19 Jakub Jelinek <jakub@redhat.com>
242 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
243 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
246 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
248 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
251 * config/avr/avr.c (tree.h): Include it.
252 (cgraph.h): Include it.
253 (avr_encode_section_info): Don't warn for uninitialized progmem
254 variable if it's just an alias.
256 2017-04-19 Richard Biener <rguenther@suse.de>
259 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
260 when needed by AutoPGO.
262 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
265 * doc/lto.texi: Remove an extra 'that'.
267 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
269 PR rtl-optimization/80429
270 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
271 are only used in debug insns.
273 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
274 Vladimir Makarov <vmakarov@redhat.com>
276 * config/sparc/predicates.md (input_operand): Add comment. Return
277 true for any memory operand when LRA is in progress.
278 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
280 2017-04-18 Jeff Law <law@redhat.com>
283 * mips.md ({return,simple_return}_internal): Do not overwrite
286 2017-04-18 Jakub Jelinek <jakub@redhat.com>
288 PR tree-optimization/80443
289 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
290 instead of adding 1, subtract -1 and similarly instead of subtracting
293 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
295 PR rtl-optimization/80357
296 * haifa-sched.c (tmp_bitmap): New variable.
297 (model_recompute): Handle duplicate use records.
298 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
299 (free_global_sched_pressure_data): Free it.
301 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
304 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
305 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
306 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
307 instead of SYSTEM_HEADER_DIR.
309 2017-04-18 Jeff Law <law@redhat.com>
312 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
313 predecessors after walking up the insn chain.
315 2017-04-18 Jakub Jelinek <jakub@redhat.com>
318 * dwarf2out.c (modified_type_die): Try harder not to emit internal
319 sizetype type into debug info.
321 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
324 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
325 unneeded test for TARGET_UPPER_REGS_SF.
326 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
328 2017-04-18 Jakub Jelinek <jakub@redhat.com>
331 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
332 instead of gsi_after_labels.
334 2017-04-18 Jeff Law <law@redhat.com>
336 * regcprop.c (maybe_mode_change): Avoid creating copies of the
340 2017-04-13 Jeff Law <law@redhat.com>
341 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
342 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
344 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
347 * config/avr/avr.c (intl.h): Include it.
348 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
350 2017-04-18 Martin Liska <mliska@suse.cz>
352 PR gcov-profile/78783
353 * gcov-tool.c (gcov_output_files): Validate that destination
354 file is either removed by the tool or by a user.
356 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
357 Guy Benyei <guybe@mellanox.com>
359 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
360 block, and do not negate it, the stored id is already negative.
362 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
364 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
366 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
369 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
370 masks of options that should be turned off if the VSX vector
371 options are turned off.
372 (OTHER_P8_VECTOR_MASKS): Likewise.
373 (OTHER_VSX_VECTOR_MASKS): Likewise.
374 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
375 rs6000_disable_incompatible_switches to validate no type switches
377 (rs6000_incompatible_switch): New function to disallow turning on
378 other vector options if -mno-vsx, -mno-power8-vector, or
379 -mno-power9-vector are specified.
381 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
383 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
385 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
387 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
388 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
389 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
390 (ARG_POINTER_CFA_OFFSET): Likewise.
392 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
394 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
395 conditions to take advantage of various optimizations.
397 2017-04-13 Jeff Law <law@redhat.com>
399 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
400 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
401 (zero_extendsidi2_dext): Likewise.
403 2017-04-13 Jakub Jelinek <jakub@redhat.com>
406 * fold-const.c (fold_ternary_loc): Revert
407 use op0 instead of fold_convert_loc (loc, type, arg0) part of
410 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
412 PR rtl-optimization/80343
413 * lra-remat.c (update_scratch_ops): Assign original hard reg to
416 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
419 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
420 to ubsan_encode_value.
422 2017-04-13 Jeff Law <law@redhat.com>
424 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
425 appearing in DEBUG_INSNs.
427 2017-04-13 Martin Liska <mliska@suse.cz>
429 PR gcov-profile/80413
430 * gcov-io.c (gcov_write_string): Copy to buffer just when
431 allocated size is greater than zero.
433 2017-04-13 Jakub Jelinek <jakub@redhat.com>
436 * dwarf2out.c (decls_for_scope): Ignore declarations of
437 current_function_decl in BLOCK_NONLOCALIZED_VARS.
439 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
442 * ipa-visibility.c (non_local_p): Fix typos.
443 (localize_node): When localizing symbol in same comdat group,
444 dissolve the group only when we know external symbols are going
446 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
448 2017-04-12 Jakub Jelinek <jakub@redhat.com>
450 PR tree-optimization/79390
451 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
452 order does not result in usable sequence, retry with reversed operand
458 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
459 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
460 op0 instead of fold_convert_loc (loc, type, arg0).
462 2017-04-12 Jeff Law <law@redhat.com>
464 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
465 has a delay slot in the generated code.
467 * config/cris/cris.md (cris_preferred_reload_class): Return
468 GENNONACR_REGS rather than GENERAL_REGS.
470 2017-04-12 Jakub Jelinek <jakub@redhat.com>
473 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
474 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
475 signedness of the result type.
477 2017-04-12 Richard Biener <rguenther@suse.de>
478 Jeff Law <law@redhat.com>
480 PR tree-optimization/80359
481 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
482 trim stores to TARGET_MEM_REFs.
484 2017-04-12 Richard Biener <rguenther@suse.de>
486 PR tree-optimization/79390
487 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
488 threading case even more.
490 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
493 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
494 for quad_address_p for TImode, instead of just not indexed_address.
496 2017-04-12 Richard Biener <rguenther@suse.de>
497 Bernd Edlinger <bernd.edlinger@hotmail.de>
500 * alias.c (component_uses_parent_alias_set_from): Handle
501 TYPE_TYPELESS_STORAGE.
502 (get_alias_set): Likewise.
503 * tree-core.h (tree_type_common): Add typeless_storage flag.
504 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
505 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
506 for types containing members with TYPE_TYPELESS_STORAGE.
507 (place_field): Likewise.
508 (layout_type): Likewise for ARRAY_TYPE.
509 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
510 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
511 TYPE_TYPELESS_STORAGE.
512 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
514 2017-04-12 Jakub Jelinek <jakub@redhat.com>
517 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
518 first argument to type.
520 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
524 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
525 CONST0_RTX (mode) rather than const0_rtx where appropriate.
526 (rs6000_expand_binop_builtin): Likewise.
527 (rs6000_expand_ternop_builtin): Likewise; also add missing
528 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
530 * doc/extend.texi: Document that vec_xxpermdi's third argument
533 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
535 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
536 Use shift_const cost parameter when calculating gain of STV shifts.
538 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
540 PR rtl-optimization/70478
541 * lra-constraints.c (process_alt_operands): Check memory for
542 disfavoring memory insn operand.
544 2017-04-11 Jakub Jelinek <jakub@redhat.com>
547 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
548 left shift in unsigned HOST_WIDE_INT type.
550 PR rtl-optimization/80385
551 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
552 (not (neg X)) into (plus X -1) for complex or non-integral modes.
555 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
556 if they have any depend clauses.
558 2017-04-11 Martin Liska <mliska@suse.cz>
561 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
562 * ipa-split.c (split_function): Create a local comdat symbol
563 if caller is in a comdat group.
565 2017-04-11 Martin Liska <mliska@suse.cz>
568 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
571 2017-04-11 Martin Sebor <msebor@redhat.com>
574 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
575 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
577 (directive::set_width, directive::set_precision, format_character):
579 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
582 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
585 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
586 conflict, set target->arch_name instead of target->cpu_name.
588 2017-04-11 Richard Biener <rguenther@suse.de>
590 PR tree-optimization/80374
591 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
592 build_zero_cst, remove fold_convertible_p check again.
594 2017-04-11 Martin Liska <mliska@suse.cz>
597 * ubsan.c (instrument_object_size): Do not instrument register
600 2017-04-11 Jakub Jelinek <jakub@redhat.com>
603 * config/i386/i386-builtin-types.def
604 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
605 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
606 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
607 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
608 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
609 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
610 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
611 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
612 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
613 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
614 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
615 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
616 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
617 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
618 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
619 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
620 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
621 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
622 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
623 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
624 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
625 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
626 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
627 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
628 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
629 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
630 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
631 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
632 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
633 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
634 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
635 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
636 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
637 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
638 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
639 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
640 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
641 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
642 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
643 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
644 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
645 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
646 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
647 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
648 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
649 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
651 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
652 flag to second_arg_count, handle 4 argument function type _COUNT
653 aliases, handle second_arg_count on second argument rather than last.
655 2017-04-10 Jeff Law <law@redhat.com>
657 PR tree-optimization/80374
658 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
659 record anything if we can not convert integer_zero_node to the
662 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
665 * config/rs6000/rs6000.c (rs6000_option_override_internal):
666 Enhance special handling given to the TARGET_P9_MINMAX option in
667 relation to certain other options.
669 2017-04-10 Bin Cheng <bin.cheng@arm.com>
671 PR tree-optimization/80153
672 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
673 remove POINTER_PLUS_EXPR's base part directly, rather than through
676 2017-04-10 Richard Biener <rguenther@suse.de>
677 Bin Cheng <bin.cheng@arm.com>
679 PR tree-optimization/80153
680 * tree-affine.c (aff_combination_to_tree): Get base pointer from
681 the first element of pointer type aff_tree. Build result expr in
683 (add_elt_to_tree): Convert to type unconditionally. Remove other
685 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
686 (rewrite_use_nonlinear_expr): Check invariant using iv information.
688 2017-04-10 Richard Biener <rguenther@suse.de>
690 * tree-ssa-structalias.c (find_func_aliases): Properly handle
693 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
695 PR rtl-optimization/70478
696 * lra-constraints.c (curr_small_class_check): New.
697 (update_and_check_small_class_inputs): New.
698 (process_alt_operands): Update curr_small_class_check. Disfavor
699 alternative insn memory operands. Check available regs for small
702 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
705 * config/mips/mips.opt (-mvirt): Update description.
706 * doc/invoke.texi (-mvirt): Likewise.
708 2017-04-10 Richard Biener <rguenther@suse.de>
711 * fold-const.c (fold_binary_loc): Look at unstripped ops when
712 looking for NEGATE_EXPR in -A / -B to A / B folding.
714 2017-04-10 Martin Liska <mliska@suse.cz>
716 PR gcov-profile/80224
717 * gcov.c (print_usage): Fix usage string.
718 (get_gcov_intermediate_filename): Remove.
719 (output_gcov_file): Use both for normal and intermediate format.
720 (generate_results): Do not initialize special file for
723 2017-04-10 Richard Biener <rguenther@suse.de>
725 PR tree-optimization/80304
726 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
729 2017-04-10 Nathan Sidwell <nathan@acm.org>
732 * config/rs6000/rs6000.c (rs6000_vector_type): New.
733 (rs6000_init_builtins): Use it.
735 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
737 * config/arm/arm.md (<mrc>): Add mode to SET source.
740 2017-04-10 Richard Biener <rguenther@suse.de>
743 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
745 2017-04-10 Jakub Jelinek <jakub@redhat.com>
748 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
749 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
750 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
751 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
752 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
753 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
754 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
755 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
756 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
757 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
758 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
759 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
760 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
761 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
762 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
763 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
764 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
765 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
766 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
767 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
768 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
769 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
770 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
772 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
774 PR rtl-optimization/70478
775 * lra-constraints.c: Reverse the last patch.
777 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
779 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
780 Add comment for WCHAR_T.
782 2017-04-08 Martin Liska <mliska@suse.cz>
785 2017-04-07 Martin Liska <mliska@suse.cz>
788 * ipa-split.c (split_function): Add function part to a same comdat
791 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
794 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
796 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
798 * rs6000/rs6000.c (vec_load_pendulum): Rename...
799 (vec_pairing): ...to this.
800 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
801 (rs6000_sched_init): Adjust for name change.
802 (struct rs6000_sched_context): Likewise.
803 (rs6000_init_sched_context): Likewise.
804 (rs6000_set_sched_context): Likewise.
806 2017-04-07 Jakub Jelinek <jakub@redhat.com>
812 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
814 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
815 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
816 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
818 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
820 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
822 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
824 PR rtl-optimization/70703
825 * ira-color.c (update_conflict_hard_regno_costs): Use
826 int64_t instead of HOST_WIDE_INT.
828 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
830 PR rtl-optimization/70478
831 * lra-constraints.c (process_alt_operands): Disfavor alternative
832 insn memory operands.
834 2017-04-07 Jeff Law <law@redhat.com>
836 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
837 CALL and NOTE_INSN_CALL_ARG_LOCATION.
839 2017-04-07 Martin Liska <mliska@suse.cz>
842 * config/aarch64/aarch64.c (aarch64_process_target_attr):
843 Show error message instead of an ICE.
845 2017-04-07 Martin Liska <mliska@suse.cz>
848 * ipa-split.c (split_function): Add function part to a same comdat
851 2017-04-07 Richard Biener <rguenther@suse.de>
854 * tree.c (get_unwidened): Also handle ! for_type case for
856 * convert.c (do_narrow): Split out from ...
857 (convert_to_integer_1): ... here. Do not pass final truncation
858 type to get_unwidened for TRUNC_DIV_EXPR.
860 2017-04-07 Richard Biener <rguenther@suse.de>
862 * tree-affine.c (wide_int_ext_for_comb): Take type rather
864 (aff_combination_const): Adjust.
865 (aff_combination_scale): Likewise.
866 (aff_combination_add_elt): Likewise.
867 (aff_combination_add_cst): Likewise.
868 (aff_combination_convert): Likewise.
869 (add_elt_to_tree): Likewise. Remove unused argument.
870 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
872 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
874 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
876 * config/arm/arm.c (arm_default_short_enums): Use
877 ARM_DEFAULT_SHORT_ENUMS.
878 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
880 2017-04-06 Jakub Jelinek <jakub@redhat.com>
883 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
884 members with redundant out-of-class redeclaration.
886 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
889 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
890 * config/i386/i386.md (*zero_extendsidi2):
891 Add (?*x,*x) and (?*v,*v) alternatives.
893 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
896 * config/i386/i386.c (ix86_expand_builtin)
897 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
898 mode from insn data. Convert operands to insn operand mode.
899 Copy operands that don't satisfy insn predicate to a register.
901 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
903 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
906 2017-04-06 Richard Biener <rguenther@suse.de>
908 PR tree-optimization/80334
909 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
910 preserve alignment of accesses.
912 2017-04-06 Richard Biener <rguenther@suse.de>
914 PR tree-optimization/80262
915 * tree-sra.c (build_ref_for_offset): Preserve address-space
917 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
918 Drop useless address-space information on MEM_REF offsets.
920 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
922 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
924 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
926 PR rtl-optimization/70703
927 * ira-color.c (update_conflict_hard_regno_costs): Use
928 HOST_WIDE_INT instead of long.
930 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
933 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
934 not defined for x86_64 target. Add -mmmx target option when __SSE2__
936 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
937 for x86_64 target. Handle -m3dnowa option.
939 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
941 PR rtl-optimization/70703
942 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
943 (update_conflict_hard_regno_costs): Use long instead of unsigned
944 arithmetic for cost calculation.
946 2017-04-05 Jakub Jelinek <jakub@redhat.com>
947 Bernd Edlinger <bernd.edlinger@hotmail.de>
950 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
953 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
956 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
957 ptr_mode with Pmode throughout.
958 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
959 into probe_stack_range and use DImode.
961 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
964 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
965 call_eh_return is true.
967 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
969 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
970 Initialize last_match_fntype_index.
972 2017-04-05 Jakub Jelinek <jakub@redhat.com>
975 * tree-nvr.c: Include internal-fn.h.
976 (pass_return_slot::execute): Ignore internal calls without
979 2017-04-04 Jakub Jelinek <jakub@redhat.com>
980 Richard Biener <rguenther@suse.de>
983 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
984 captures used multiple times, except for the last use.
985 * generic-match-head.c: Include gimplify.h.
987 2017-04-04 Jakub Jelinek <jakub@redhat.com>
989 PR tree-optimization/79390
990 * target.h (struct noce_if_info): Declare.
991 * targhooks.h (default_noce_conversion_profitable_p): Declare.
992 * target.def (noce_conversion_profitable_p): New target hook.
993 * ifcvt.h (struct noce_if_info): New type, moved from ...
994 * ifcvt.c (struct noce_if_info): ... here.
995 (noce_conversion_profitable_p): Renamed to ...
996 (default_noce_conversion_profitable_p): ... this. No longer
998 (noce_try_store_flag_constants, noce_try_addcc,
999 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
1000 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
1001 instead of noce_conversion_profitable_p.
1002 * config/i386/i386.c: Include ifcvt.h.
1003 (ix86_option_override_internal): Don't override
1004 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
1005 (ix86_noce_conversion_profitable_p): New function.
1006 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
1007 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
1008 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
1009 * doc/tm.texi: Regenerated.
1011 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1013 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
1016 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
1019 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
1020 instructions for small multiply cores.
1022 2017-04-04 Jeff Law <law@redhat.com>
1024 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
1026 (mips_expand_vec_perm_const): Initialize elements in orig_perm
1027 that are not set by the loop over the elements.
1029 2017-04-04 Jakub Jelinek <jakub@redhat.com>
1032 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
1033 int mode, convert_modes it to mode as unsigned, otherwise use
1034 lowpart_subreg to mode rather than SImode.
1035 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
1036 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
1037 Use DImode instead of SImode for the shift count operand.
1038 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
1041 2017-04-04 Richard Biener <rguenther@suse.de>
1044 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
1045 arithmetic done for the negate or the plus. Simplify.
1046 (A - (-B) -> A + B): Likewise.
1047 * fold-const.c (split_tree): Make sure to not negate pointers.
1049 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
1051 PR rtl-optimization/60818
1052 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
1053 a compare of comparisons with the thing compared if this results
1054 in a different machine mode.
1056 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
1058 * alias.c (base_alias_check): Fix typo in comment.
1059 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
1060 * cgraphunit.c (symbol_table::compile): Likewise.
1061 * collect2.c (maybe_run_lto_and_relink): Likewise.
1062 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
1063 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
1064 * config/avr/avr.c (avr_map_op_t): Likewise.
1065 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
1066 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
1067 * config/epiphany/epiphany.md (movcc): Likewise.
1068 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
1069 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
1071 * config/mips/mips.c (mips_save_restore_reg): Likewise.
1072 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
1073 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
1074 * config/sh/sh.c (sh_rtx_costs): Likewise.
1075 * fold-const.c (fold_truth_andor): Likewise.
1076 * genautomata.c (collapse_flag): Likewise.
1077 * gengtype.h (struct type::u::s): Likewise.
1078 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
1079 * input.c (FORMAT_AMOUNT): Likewise.
1080 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
1081 (known_aggs_to_agg_replacement_list): Likewise.
1082 * ipa-inline-analysis.c: Likewise.
1083 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
1084 * ipa-polymorphic-call.c
1085 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
1086 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
1087 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
1089 * modulo-sched.c (apply_reg_moves): Likewise.
1090 * omp-expand.c (build_omp_regions_1): Likewise.
1091 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
1092 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
1093 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
1094 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
1095 * value-prof.c: Likewise.
1096 * var-tracking.c (val_reset): Likewise.
1098 2017-04-03 Richard Biener <rguenther@suse.de>
1100 PR tree-optimization/80275
1101 * fold-const.c (split_address_to_core_and_offset): Handle
1104 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
1106 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
1107 descriptors is at least equal to that of functions.
1109 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
1111 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
1113 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
1116 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
1117 (mov<IMOD4:mode>): New expander.
1118 (*mov<IMOD4:mode>_internal): New insn and split pattern.
1120 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
1122 PR rtl-optimization/79405
1123 * fwprop.c (propagations_left): New variable.
1124 (forward_propagate_into): Decrement it.
1125 (fwprop_init): Initialize it.
1126 (fw_prop): If the variable has reached zero, stop propagating.
1127 (fwprop_addr): Ditto.
1129 2017-03-31 Jakub Jelinek <jakub@redhat.com>
1132 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
1133 a FUNCTION_DECL, pass it as decl instead of origin to
1136 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
1138 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
1141 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
1144 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
1145 TARGET_VSX_SMALL_INTEGER.
1147 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1149 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
1150 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
1152 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
1154 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
1155 extraction from odd-numbered MSA register.
1157 2017-03-31 Jakub Jelinek <jakub@redhat.com>
1160 * expmed.c (store_bit_field_1): Don't attempt to create
1161 a word subreg out of hard registers wider than word if they
1162 have HARD_REGNO_NREGS of 1 for their mode.
1165 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
1166 conversions to integer types wider than word and pointer.
1169 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
1170 (rtx_equal_for_cselib_p): Pass 0 to it.
1171 * cselib.c (cselib_hasher::equal): Likewise.
1172 (rtx_equal_for_cselib_1): Add depth argument. If depth
1173 is 128, don't look up VALUE locs and punt. Increment
1174 depth in recursive calls when walking VALUE locs.
1176 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
1178 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
1179 (make_gcov_file_name): Use the canonical path name for generating
1181 (read_line): Fix handling of files with ascii null bytes.
1183 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
1185 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
1186 to initialise a vector register instead
1187 of using a const_int.
1189 2017-03-30 Jakub Jelinek <jakub@redhat.com>
1191 PR translation/80189
1192 * gimplify.c (omp_default_clause): Use %qs instead of %s in
1193 diagnostic messages.
1195 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
1198 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
1199 (dfp_diex_<mode>): Update mode of operand 1.
1200 * doc/extend.texi (dxex, dxexq): Document change to return type.
1201 (diex, diexq): Document change to argument type.
1203 2017-03-30 Martin Jambor <mjambor@suse.cz>
1206 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
1207 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
1208 it reflects the signature changes performed at the callee side.
1209 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
1210 to cgraph_build_function_type_skip_args.
1211 (build_function_decl_skip_args): Adjust call to the above function.
1213 2017-03-30 Jakub Jelinek <jakub@redhat.com>
1216 * config/i386/sse.md
1217 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
1218 register as dest whenever it is a MEM not rtx_equal_p to the
1219 corresponding dup operand, and when forcing into reg move the
1220 reg into the memory afterwards.
1221 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
1222 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
1223 for the force_reg mode.
1224 (avx512vl_vextractf128<mode>): Use register as dest either
1225 always when a MEM, or when it is a MEM not rtx_equal_p to the
1226 corresponding dup operand, or even not when it is a CONST_VECTOR
1227 depending on the mode and lo vs. hi.
1228 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
1230 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
1231 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
1232 Likewise. Require that operands[2] is even.
1233 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
1234 Remove extraneous parens. Require that operands[2] is a multiple
1236 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
1237 operands[0] is a MEM if <mask_applied>, the predicates/constraints
1238 disallow memory then.
1240 2017-03-30 Richard Biener <rguenther@suse.de>
1242 PR tree-optimization/77498
1243 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
1244 to non-constants over backedges.
1246 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
1248 PR rtl-optimization/80233
1249 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
1250 as last_combined_insn. Do not test for BARRIER_P separately.
1252 2017-03-29 Andreas Schwab <schwab@suse.de>
1255 * calls.c (prepare_call_address): Convert funexp to Pmode before
1256 copying to temp reg.
1258 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1260 PR tree-optimization/80158
1261 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
1262 Handle possible future case of more than one alternate
1264 (replace_rhs_if_not_dup): Likewise.
1265 (replace_one_candidate): Likewise.
1267 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
1269 PR rtl-optimization/80193
1270 * ira.c (ira): Do not check allocation for LRA.
1272 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
1274 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
1275 (nvptx_output_simt_exit): Declare.
1276 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
1277 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
1278 (init_softstack_frame): Move initialization of crtl->is_leaf to...
1279 (nvptx_declare_function_name): ...here. Emit declaration of local
1280 memory space buffer for omp_simt_enter insn.
1281 (nvptx_output_unisimt_switch): New.
1282 (nvptx_output_softstack_switch): New.
1283 (nvptx_output_simt_enter): New.
1284 (nvptx_output_simt_exit): New.
1285 * config/nvptx/nvptx.h (struct machine_function): New fields
1286 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
1287 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
1288 (UNSPECV_SIMT_EXIT): Ditto.
1289 (omp_simt_enter_insn): New insn.
1290 (omp_simt_enter): New expansion.
1291 (omp_simt_exit): New insn.
1292 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
1294 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
1295 (expand_GOMP_SIMT_ENTER_ALLOC): New.
1296 (expand_GOMP_SIMT_EXIT): New.
1297 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
1298 (GOMP_SIMT_ENTER_ALLOC): Ditto.
1299 (GOMP_SIMT_EXIT): Ditto.
1300 * target-insns.def (omp_simt_enter): New insn.
1301 (omp_simt_exit): Ditto.
1302 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
1304 (lower_rec_simd_input_clauses): Implement SIMT privatization.
1305 (lower_rec_input_clauses): Likewise.
1306 (lower_lastprivate_clauses): Handle SIMT privatization.
1308 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
1309 (ompdevlow_adjust_simt_enter): New.
1310 (find_simtpriv_var_op): New.
1311 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
1312 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
1314 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
1315 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
1316 (copy_decl_for_dup_finish): Ditto.
1318 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
1320 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
1323 * config/i386/i386.c (ix86_option_override_internal): Always
1324 allow -mpreferred-stack-boundary=3 for 64-bit targets.
1326 2017-03-28 Bin Cheng <bin.cheng@arm.com>
1328 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
1330 2017-03-28 Bin Cheng <bin.cheng@arm.com>
1332 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
1333 mark new edge's irreducible flag accordign to it.
1334 (vect_do_peeling): Check loop preheader edge's irreducible flag
1335 and pass it to function slpeel_add_loop_guard.
1337 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
1339 PR tree-optimization/80218
1340 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
1341 Update block frequencies and counts.
1343 2017-03-28 Richard Biener <rguenther@suse.de>
1345 PR tree-optimization/78644
1346 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
1347 of a simplification result we may not use it at all.
1349 2017-03-28 Richard Biener <rguenther@suse.de>
1352 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
1353 without arguments, generate default definition of a SSA name.
1355 2017-03-28 Richard Biener <rguenther@suse.de>
1358 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
1359 TYPE_REF_CAN_ALIAS_ALL references.
1360 * fold-const.c (fold_indirect_ref_1): Likewise.
1362 2017-03-28 Martin Liska <mliska@suse.cz>
1365 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
1366 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
1368 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
1369 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
1371 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
1372 (EXTRA_SPECS): Define.
1373 (SUBTARGET_EXTRA_SPECS): Likewise.
1374 (SUBTARGET_CPP_SPEC): Likewise.
1375 * config/arc/elf.h (EXTRA_SPECS): Renamed to
1376 SUBTARGET_EXTRA_SPECS.
1377 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
1379 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
1381 * config/arc/simdext.md (vst64_insn): Update pattern.
1382 (vld32wh_insn): Likewise.
1383 (vld32wl_insn): Likewise.
1384 (vld64_insn): Likewise.
1385 (vld32_insn): Likewise.
1387 2017-03-28 Marek Polacek <polacek@redhat.com>
1390 * fold-const.c (fold_comparison): Use protected_set_expr_location
1391 instead of SET_EXPR_LOCATION.
1393 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
1395 * tree.c (add_expr): Avoid name lookup warning.
1397 2017-03-27 Jeff Law <law@redhat.com>
1399 PR tree-optimization/80216
1400 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
1401 function name. Limit recursion depth.
1402 (record_temporary_equivalences): Corresponding changes.
1404 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
1406 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
1409 2017-03-27 Jakub Jelinek <jakub@redhat.com>
1412 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
1414 * cfgcleanup.c (reg_note_cfa_p): New array.
1415 (insns_have_identical_cfa_notes): New function.
1416 (old_insns_match_p): Don't cross-jump in between /f
1417 and non-/f instructions. If both i1 and i2 are frame related,
1418 verify all CFA notes, their order and content.
1420 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1423 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
1424 HImode and SImode with zero extend to DImode to one insn.
1425 (bswap<mode>2_extenddi): Likewise.
1426 (bswapsi2_extenddi): Likewise.
1427 (bswaphi2_extendsi): Likewise.
1428 (bswaphi2): Combine bswap HImode and SImode into one insn.
1429 Separate memory insns from swapping register.
1430 (bswapsi2): Likewise.
1431 (bswap<mode>2): Likewise.
1432 (bswaphi2_internal): Delete, no longer used.
1433 (bswapsi2_internal): Likewise.
1434 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
1435 store, and gpr<-gpr swap insns.
1436 (bswap<mode>2_store): Likewise.
1437 (bswaphi2_reg): Register only splitter, combine with the splitter.
1438 (bswaphi2 splitter): Likewise.
1439 (bswapsi2_reg): Likewise.
1440 (bswapsi2 splitter): Likewise.
1441 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
1442 the insns into load, store, and register/register insns.
1443 (bswapdi2_ldbrx): Likewise.
1444 (bswapdi2_load): Likewise.
1445 (bswapdi2_store): Likewise.
1446 (bswapdi2_reg): Likewise.
1448 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
1450 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
1451 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
1453 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1456 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
1458 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1459 special handling for target option conflicts between dform
1460 options (-mpower9-dform, -mpower9-dform-vector,
1461 -mpower9-dform-scalar) and -mno-direct-move.
1463 2017-03-27 Richard Biener <rguenther@suse.de>
1465 PR tree-optimization/80181
1466 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
1468 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1470 * config/arc/predicates.md (move_double_src_operand): Replace the
1471 call to move_double_src_operand with a call to address_operand.
1473 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1475 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
1476 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
1477 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
1479 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1481 * config/arc/predicates.md (long_immediate_loadstore_operand):
1482 Consider scaled addresses cases.
1484 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
1486 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
1487 restored when in interrupt.
1488 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
1489 doesn't have delay slot.
1491 2017-03-27 Richard Biener <rguenther@suse.de>
1494 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
1495 inlined thunk clones.
1497 2017-03-27 Jakub Jelinek <jakub@redhat.com>
1500 * asan.c (instrument_derefs): Copy over last operand from
1501 original COMPONENT_REF to the new COMPONENT_REF with
1502 DECL_BIT_FIELD_REPRESENTATIVE.
1503 * ubsan.c (instrument_object_size): Likewise.
1505 2017-03-27 Richard Biener <rguenther@suse.de>
1507 PR tree-optimization/80170
1508 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
1509 sure DR/SCEV didnt fold in constants we do not see when looking
1510 at the reference base alignment.
1512 2017-03-27 Richard Biener <rguenther@suse.de>
1515 * gimple-fold.c (fold_ctor_reference): Properly guard against
1516 NULL return value from canonicalize_constructor_val.
1518 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
1521 * config/i386/i386.c (ix86_expand_builtin)
1522 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
1523 flags reg setting and flags reg using instructions.
1524 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
1525 clobbering instructions to zero extend op2.
1527 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
1529 * doc/install.texi (Configuration) <--with-aix-soname>:
1530 Update link to AIX ld.
1532 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
1534 PR rtl-optimization/80160
1535 PR rtl-optimization/80159
1536 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
1537 reg_alternate_class into account.
1539 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
1542 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
1543 to consider in curr_insn_transform.
1545 2017-03-24 Jakub Jelinek <jakub@redhat.com>
1547 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
1548 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
1549 and emit_mode_inner.
1551 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1553 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
1554 argument to the overloaded builtin variants. Use the new flag to
1555 deprecate certain builtin variants.
1556 * config/s390/s390-builtin-types.def: Add new builtin types.
1557 * config/s390/s390-builtins.h: Support new flags field for
1558 overloaded builtins.
1559 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
1560 (s390_macro_to_expand): Enable vector float data type.
1561 (s390_cpu_cpp_builtins_internal): Indicate support of the new
1562 builtins by incrementing the __VEC__ version number.
1563 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
1565 (s390_resolve_overloaded_builtin): Emit error messages depending
1566 on the builtin flags.
1567 * config/s390/s390.c (s390_expand_builtin): Support additional
1568 flags argument. Change error message to match the messages
1569 emitted in s390-c.c.
1570 * config/s390/s390.md: New UNSPEC_* constants.
1571 (op_type): Add new instruction types.
1572 * config/s390/vecintrin.h: Add new builtins and test data class
1574 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
1575 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
1576 (VEC_INEXACT, VEC_NOINEXACT): New constants.
1577 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
1578 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
1579 ("vec_mergel<mode>"): V_HW -> VEC_HW.
1581 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
1582 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
1583 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
1584 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
1586 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
1587 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
1588 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
1589 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
1591 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
1592 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
1593 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
1594 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
1595 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
1596 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
1597 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
1599 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
1600 ("vec_scatter_element<V_HW_4:mode>_DI")
1601 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
1602 ("vec_fpint<mode>", "vflls")
1603 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
1604 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
1605 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
1606 ("*vec_cmphe<mode>_cc"): ... these.
1608 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
1609 mode constant instead of magic value.
1611 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1613 * config/s390/s390.c (s390_expand_vec_compare): Support other
1614 vector floating point modes than just V2DF.
1615 (s390_expand_vcond): Likewise.
1616 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
1617 (s390_cannot_change_mode_class): Prevent mode changes between TF
1618 and V1TF in vector registers.
1619 * config/s390/s390.md (DF, SF): New mode attributes.
1620 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
1621 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
1622 SFmode support for VRs.
1623 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
1625 (VFT, VF_HW): New mode iterators.
1626 (vw, sdx): New mode attributes.
1627 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
1628 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
1629 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
1630 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
1631 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
1632 also the new vector floating point modes. Renaming to ...
1634 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
1635 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
1636 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
1637 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
1638 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
1639 ("vec_unordered<mode>"): ... these.
1641 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
1642 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
1643 ("*vec_extendv2df"): New insn definitions.
1645 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1647 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
1648 ("mulditi3_2", "*muldi3_sign"): New patterns.
1649 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
1650 rename the pattern definition.
1652 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1654 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
1656 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
1658 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1660 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
1661 instruction if possible.
1662 * config/s390/vector.md (vec_halfnumelts): New mode
1664 ("*vec_vllezlf<mode>"): New pattern.
1666 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1668 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
1669 ("popcountv4si2", "popcountv2di2"): Rename to ...
1670 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
1671 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
1673 ("popcount<mode>2_vxe"): New pattern.
1675 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1677 * common/config/s390/s390-common.c (processor_flags_table): Add
1679 * config.gcc: Add arch12.
1680 * config/s390/driver-native.c (s390_host_detect_local_cpu):
1681 Default to arch12 for unknown CPU model numbers.
1682 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
1683 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
1684 PROCESSOR_max sanity check.
1685 * config/s390/s390-opts.h (enum processor_type): Add
1687 * config/s390/s390.c (processor_table): Add arch12.
1688 (s390_expand_builtin): Add check for B_VXE flag.
1689 (s390_issue_rate): Add PROCESSOR_ARCH12.
1690 (s390_get_sched_attrmask): Likewise.
1691 (s390_get_unit_mask): Likewise.
1692 (s390_sched_score): Enable z13 scheduling for arch12.
1693 (s390_sched_reorder): Likewise.
1694 (s390_sched_variable_issue): Likewise.
1695 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
1697 (s390_tune_attr): Use z13 scheduling also for arch12.
1698 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
1699 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
1700 (TARGET_VXE_P): New macros.
1701 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
1702 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
1703 * config/s390/s390.opt: Add arch12 as processor_type.
1705 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1707 * config/s390/s390.md
1708 ("fixuns_truncdddi2", "fixuns_trunctddi2")
1709 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
1710 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
1712 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
1713 Rename expanders to ...
1715 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
1716 ("fixuns_truncdddi2_emu"): ... these.
1718 ("fixuns_trunc<mode>si2_emu"): New expander.
1720 ("*fixuns_truncdfdi2_z13"): Rename to ...
1721 ("*fixuns_truncdfdi2_vx"): ... this.
1723 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1725 * config/s390/2964.md: Remove the single element vector compare
1726 instructions which are no longer used.
1727 * config/s390/s390.c (s390_select_ccmode): Remove handling of
1729 (s390_canonicalize_comparison): Remove handling of DFmode
1731 (s390_expand_vec_compare_scalar): Remove function.
1732 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
1733 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
1735 ("*cmp<mode>_ccs"): Add wfcdb instruction.
1737 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1739 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
1741 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
1742 will anyway by matched by mov<mode>_64dfp.
1744 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1746 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
1747 vlef/vstef. Add missing operand to vleif.
1749 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1751 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
1752 pair for all vector types with 64 bit elements.
1753 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
1754 * config/s390/vector.md (V_HW_64): ... here.
1755 (V_128_NOSINGLE): New mode iterator.
1756 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
1757 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
1758 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
1759 ("*vec_load_pairv2di"): Change to ...
1760 ("*vec_load_pair<mode>"): ... this one.
1762 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1764 * config/s390/constraints.md: Add comments.
1765 (jKK): Reject element sizes > 8 bytes.
1766 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
1768 * config/s390/s390.md: Add the s_operand checks formerly in
1769 s390_split_ok_p to various splitters where they are still
1771 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
1772 for 128 bit vectors. Plus two splitters.
1774 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1776 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
1779 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1782 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
1783 error if the boundary argument is not constant.
1785 2017-03-24 Jakub Jelinek <jakub@redhat.com>
1787 PR rtl-optimization/80112
1788 * loop-doloop.c (doloop_condition_get): Don't check condition
1789 if cmp isn't SET with IF_THEN_ELSE src.
1791 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1793 PR tree-optimization/80158
1794 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
1795 replacing a candidate statement, also replace it for the
1796 candidate's alternate interpretation.
1797 (replace_rhs_if_not_dup): Likewise.
1798 (replace_one_candidate): Likewise.
1800 2017-03-24 Richard Biener <rguenther@suse.de>
1802 PR tree-optimization/80167
1803 * graphite-isl-ast-to-gimple.c
1804 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
1806 (translate_isl_ast_to_gimple::get_rename): Likewise.
1808 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
1810 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
1811 handling of certain combinations of target options, including the
1812 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
1813 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
1815 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1818 * config/arm/arm.md (*load_multiple): Add reload_completed to
1821 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1822 Richard Biener <rguenth@suse.de>
1824 PR tree-optimization/79908
1825 PR tree-optimization/80136
1826 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
1827 been cast away, gimplify_and_add suffices.
1829 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
1831 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
1833 2017-03-23 Richard Biener <rguenther@suse.de>
1835 PR tree-optimization/80032
1836 * gimplify.c (gimple_push_cleanup): Forced unconditional
1837 cleanups still have to go to the conditional_cleanups
1840 2017-03-22 Jakub Jelinek <jakub@redhat.com>
1842 PR tree-optimization/80072
1843 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
1845 (next_operand_entry_id): Change type to unsigned int.
1846 (sort_by_operand_rank): Make sure to return the right return value
1847 even if unsigned fields are bigger than INT_MAX.
1848 (struct oecount): Change cnt and id type to unsigned int.
1849 (oecount_hasher::equal): Formatting fix.
1850 (oecount_cmp): Make sure to return the right return value
1851 even if unsigned fields are bigger than INT_MAX.
1852 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
1855 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
1856 TREE_READONLY on result if writing it more than once.
1859 * doc/invoke.texi (-fsanitize=thread): Document that with
1860 -fnon-call-exceptions atomics are not able to throw
1864 * tsan.c: Include tree-eh.h.
1865 (instrument_builtin_call): Call maybe_clean_eh_stmt or
1866 maybe_clean_or_replace_eh_stmt where needed.
1867 (instrument_memory_accesses): Add cfg_changed argument.
1868 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
1869 if it returned true.
1870 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
1872 PR rtl-optimization/63191
1873 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
1874 wrapper function, moved the whole old content into ...
1875 (ix86_delegitimize_address_1): ... this. New inline function.
1876 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
1877 true as last argument instead of ix86_delegitimize_address.
1879 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
1881 * config/aarch64/aarch64.c (generic_branch_cost): Copy
1882 cortexa57_branch_cost.
1884 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
1886 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
1888 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1891 * doc/md.texi (Constraints): Document wA constraint.
1892 * config/rs6000/constraints.md (wA): New.
1893 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
1894 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
1895 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
1896 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
1898 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
1901 * gimplify.c (is_oacc_declared): New function.
1902 (oacc_default_clause): Use it to set default flags for acc declared
1903 variables inside parallel regions.
1904 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
1906 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
1907 declare attribute to any decl as necessary.
1909 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
1912 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
1913 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
1914 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
1915 (arm_arch_lpae): This.
1916 * config/arm/arm.c (arm_arch7ve): Rename into ...
1917 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
1918 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
1921 2017-03-22 Martin Liska <mliska@suse.cz>
1924 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
1925 error message instead of an ICE.
1927 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1929 * doc/extend.texi (6.11 Additional Floating Types): Revise.
1931 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
1933 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
1935 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
1938 2017-03-21 Martin Sebor <msebor@redhat.com>
1940 * doc/extend.texi: Use "cannot" instead of "can't."
1941 * doc/hostconfig.texi: Same.
1942 * doc/install.texi: Same.
1943 * doc/invoke.texi: Same.
1944 * doc/loop.texi: Same.
1945 * doc/md.texi: Same.
1946 * doc/objc.texi: Same.
1947 * doc/rtl.texi: Same.
1948 * doc/tm.texi: Same.
1949 * doc/tm.texi.in: Same.
1950 * doc/trouble.texi: Same.
1952 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
1955 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
1956 (collect_checksum_attributes): Set it.
1957 (die_checksum_ordered): Use it.
1959 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1961 PR tree-optimization/79908
1962 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
1963 change: For a VA_ARG whose LHS has been cast away, use
1964 force_gimple_operand to construct the side effects.
1966 2017-03-21 David Malcolm <dmalcolm@redhat.com>
1968 PR translation/80001
1969 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
1970 more amenable to translation.
1971 (oacc_loop_auto_partitions): Likewise.
1973 2017-03-21 Marek Polacek <polacek@redhat.com>
1974 Martin Sebor <msebor@redhat.com>
1976 PR tree-optimization/80109
1977 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
1980 2017-03-21 Jakub Jelinek <jakub@redhat.com>
1981 Segher Boessenkool <segher@kernel.crashing.org>
1984 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
1985 check reg_used_between_p between insn and one of succ or succ2
1986 depending on if succ is artificial insn not inserted into insn
1989 2017-03-21 Martin Liska <mliska@suse.cz>
1991 PR gcov-profile/80081
1992 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
1993 * doc/gcc.texi: Include gcov-dump stuff.
1994 * doc/gcov-dump.texi: New file.
1996 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
1998 PR rtl-optimization/79150
1999 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
2000 conditional jump, if the jump is the last insn of the loop.
2002 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2003 Richard Biener <rguenth@suse.de>
2005 PR tree-optimization/79908
2006 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
2007 been cast away, use force_gimple_operand to construct the side
2010 2017-03-21 Martin Liska <mliska@suse.cz>
2013 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
2016 2017-03-21 Brad Spengler <spender@grsecurity.net>
2019 * plugin.c (htab_hash_plugin): New function.
2020 (add_new_plugin): Use it and adjust.
2021 (parse_plugin_arg_opt): Adjust.
2022 (init_one_plugin): Likewise.
2024 2017-03-21 Richard Biener <rguenther@suse.de>
2026 PR tree-optimization/80032
2027 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
2028 if set force the cleanup to happen unconditionally.
2029 (gimplify_target_expr): Push inserted clobbers with force_uncond
2030 to avoid them being removed by control-dependent DCE.
2032 2017-03-21 Richard Biener <rguenther@suse.de>
2034 PR tree-optimization/80122
2035 * tree-inline.c (copy_bb): Do not expans va-arg packs or
2036 va_arg_pack_len when the inlined call stmt requires pack
2038 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
2040 2017-03-21 Jakub Jelinek <jakub@redhat.com>
2043 * tsan.c (instrument_builtin_call): If the memory model argument
2044 is not a constant, assume it is valid.
2047 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
2050 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
2052 PR rtl-optimization/79910
2053 * combine.c (can_combine_p): Do not allow combining an I0 or I1
2054 if its dest is used by an insn before I2 (other than the combined
2055 insns themselves, which are properly handled already).
2057 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
2060 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
2062 * combine.c (record_used_regs): New static function.
2063 (try_combine): Handle situations where there is an additional
2064 instruction between I2 and I3 which needs to have a LOG_LINK
2068 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
2070 * combine.c (try_combine): Delete redundant i1 test. Call
2071 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
2073 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2076 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
2079 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2081 PR tree-optimization/80054
2082 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
2083 the optimization if a PHI or any of its arguments is not dominated
2084 by the candidate's basis. Use gphi* rather than gimple* as
2086 (replace_profitable_candidates): Clean up a gimple* variable that
2087 should be a gphi* variable.
2089 2017-03-20 Martin Sebor <msebor@redhat.com>
2092 * doc/extend.texi (attribute constructor): Document present limitation.
2094 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
2097 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
2098 __POWER9_VECTOR__ #ifdef control, change template definition to
2099 use Power9-specific built-in function.
2100 (vec_any_eq): Likewise.
2101 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
2102 to control outcomes from this test.
2103 (vector_ae_<mode>p): For VEC_F modes, likewise.
2105 2017-03-20 Ian Lance Taylor <iant@google.com>
2107 * config/i386/i386.c (ix86_function_regparm): Save an extra
2108 register for -fsplit-stack with DECL_STATIC_CHAIN.
2110 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
2113 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
2114 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
2116 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
2118 * config/riscv/riscv.c (riscv_print_operand): Use "fence
2120 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
2123 2017-03-20 Marek Polacek <polacek@redhat.com>
2126 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
2128 2017-03-20 Richard Biener <rguenther@suse.de>
2130 PR tree-optimization/80113
2131 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
2132 allocate extra SSA name for PHI def.
2133 (add_close_phis_to_outer_loops): Likewise.
2134 (add_close_phis_to_merge_points): Likewise.
2135 (copy_loop_close_phi_args): Likewise.
2136 (copy_cond_phi_nodes): Likewise.
2138 2017-03-20 Martin Liska <mliska@suse.cz>
2141 * tree-chkp.c (chkp_build_returned_bound): Do not build
2142 returned bounds for a LHS that's not a BOUNDED_P type.
2144 2017-03-20 Martin Liska <mliska@suse.cz>
2148 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
2149 COMPLEX_CST and VECTOR_CST.
2151 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2154 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
2155 target operand. A new splitter adds the clobber statement in case
2156 the target operand is dead anyway.
2158 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
2160 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
2161 to age-old versions of binutils and glibc.
2163 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
2165 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
2167 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
2169 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
2171 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
2173 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
2174 requirement for binutils 2.13.
2176 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
2178 * combine.c (try_combine): Delete redundant i1 test. Call
2179 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
2181 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
2183 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
2184 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
2186 <riscv64-*-elf>: Re-arrange section
2187 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
2188 <riscv32-*-linux>: Likewise.
2189 <riscv64-*-elf>: Likewise
2190 <riscv64-*-linux>: Likewise.
2192 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
2195 * aarch64.opt(verbose-cost-dump): Fix typo.
2197 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
2200 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
2201 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
2203 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
2205 * reload.c (find_reloads): When reloading a nonoffsettable address,
2206 use RELOAD_OTHER for it and its address reloads.
2208 PR rtl-optimization/79910
2209 * combine.c (record_used_regs): New static function.
2210 (try_combine): Handle situations where there is an additional
2211 instruction between I2 and I3 which needs to have a LOG_LINK
2214 2017-03-17 Jeff Law <law@redhat.com>
2216 PR tree-optimization/71437
2217 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
2218 conditional in the hash table first.
2219 (vrp_dom_walker::before_dom_children): Extract condition from
2220 ASSERT_EXPR. Record condition, its inverion and any implied
2223 2017-03-17 Marek Polacek <polacek@redhat.com>
2224 Markus Trippelsdorf <markus@trippelsdorf.de>
2226 PR tree-optimization/80079
2227 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
2230 2017-03-17 Richard Biener <rguenther@suse.de>
2233 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
2234 Properly verify the LHS before the RHS possibly claims to be
2236 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
2239 2017-03-17 Martin Jambor <mjambor@suse.cz>
2241 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
2242 (List of -O2 options): Likewise.
2243 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
2246 2017-03-17 Tom de Vries <tom@codesourcery.com>
2248 * gcov-dump.c (print_usage): Print bug_report_url.
2250 2017-03-17 Richard Biener <rguenther@suse.de>
2253 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
2254 (parser::peek): Likewise.
2256 2017-03-17 Richard Biener <rguenther@suse.de>
2258 PR tree-optimization/80048
2259 * sese.c (free_sese_info): Properly release rename_map and
2260 copied_bb_map elements.
2262 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
2264 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
2265 Add linked-list forward and backlinks. Insert on
2266 construction, remove on destruction.
2267 (class pass_store_merging): Add m_stores_head field.
2268 (pass_store_merging::terminate_and_process_all_chains):
2269 Iterate over m_stores_head list.
2270 (pass_store_merging::terminate_all_aliasing_chains):
2272 (pass_store_merging::execute): Check for debug stmts first.
2273 Push new chains onto the m_stores_head stack.
2275 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
2278 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
2279 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
2280 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
2282 2017-03-16 Jeff Law <law@redhat.com>
2284 PR tree-optimization/71437
2285 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
2286 member function. Implementation moved into after_dom_children
2287 member function and into the threader's thread_outgoing_edges
2289 (dom_opt_dom_walker::after_dom_children): Simplify by moving
2290 some code into new thread_outgoing_edges.
2291 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
2292 definition. Simplify marker handling (do it here). Assume we always
2293 have the available expression and the const/copies tables.
2294 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
2296 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
2297 * tree-vrp.c (equiv_stack): No longer file scoped.
2298 (vrp_dom_walker): New class.
2299 (vrp_dom_walker::before_dom_children): New member function.
2300 (vrp_dom_walker::after_dom_children): Likewise.
2301 (identify_jump_threads): Setup domwalker. Use it rather than
2302 walking edges in a random order by hand. Simplify setup/finalization.
2303 (finalize_jump_threads): Remove.
2304 (vrp_finalize): Do not call identify_jump_threads here.
2305 (execute_vrp): Do it here instead and call thread_through_all_blocks
2308 PR tree-optimization/71437
2309 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
2311 (simplify_stmt_for_jump_threading): Add basic_block argument. All
2313 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
2314 (dom_opt_dom_walker::thread_across_edge): Remove
2315 handle_dominating_asserts argument. All callers changed.
2316 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
2317 changes. Remove calls to lhs_of_dominating_assert. Other
2318 uses of handle_dominating_asserts turn into unconditional code
2319 (simplify_control_stmt_condition_1): Likewise.
2320 (simplify_control_stmt_condition): Likewise.
2321 (thread_through_normal_block, thread_across_edge): Likewise.
2322 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
2323 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
2324 object if it is not an SSA_NAME.
2325 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
2326 before calling into the VRP specific simplifiers.
2327 (identify_jump_threads): Remove handle_dominating_asserts
2330 2017-03-16 Jakub Jelinek <jakub@redhat.com>
2333 * tree-diagnostic.c (default_tree_printer): No longer static.
2334 * tree-diagnostic.h (default_tree_printer): New prototype.
2336 2017-03-16 Tamar Christina <tamar.christina@arm.com>
2338 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
2339 Change ins into fmov.
2341 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2343 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
2344 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
2345 Use h_con constraint for operand 1.
2346 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
2347 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
2349 2017-03-15 Jeff Law <law@redhat.com>
2351 PR tree-optimization/71437
2352 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
2353 (record_temporary_equivalences): Use it.
2355 PR tree-optimization/71437
2356 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
2357 tree-ssa-scopedtables.
2358 (lookup_avail_expr, build_and_record_new_cond): Likewise.
2359 (record_conditions, record_cond, vuse_eq): Likewise.
2360 (record_edge_info): Adjust to API tweak of record_conditions.
2361 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
2362 (record_temporary_equivalences, optimize_stmt): Likewise.
2363 (eliminate_redundant_computations): Likewise.
2364 (record_equivalences_from_stmt): Likewise.
2365 * tree-ssa-scopedtables.c: Include options.h and params.h.
2366 (vuse_eq): New function, moved from tree-ssa-dom.c
2367 (build_and_record_new_cond): Likewise.
2368 (record_conditions): Likewise. Accept vector of conditions rather
2369 than edge_equivalence structure for first argument.
2370 for the first argument.
2371 (avail_exprs_stack::lookup_avail_expr): New member function, moved
2372 from tree-ssa-dom.c.
2373 (avail_exprs_stack::record_cond): Likewise.
2374 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
2375 from tree-ssa-dom.c.
2376 (avail_exprs_stack): Add new member functions lookup_avail_expr
2378 (record_conditions): Declare.
2380 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
2383 * lra-constraints.c (process_alt_operands): Increase reject for
2384 reloading an input/output operand.
2386 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
2389 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
2390 insns to convert from signed/unsigned char/short to IEEE 128-bit
2392 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
2394 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
2397 * config/i386/i386.c (ix86_vector_duplicate_value): Create
2398 subreg of inner mode for values already in registers.
2400 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
2402 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
2403 iteration reg is used after the loop.
2405 2017-03-14 Martin Sebor <msebor@redhat.com>
2407 PR tree-optimization/79800
2408 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
2409 precision in negative-positive range.
2410 (format_floating): Call non-const overload with adjusted precision.
2412 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2415 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
2418 2017-03-14 Martin Sebor <msebor@redhat.com>
2421 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
2422 * builtins.def (aligned_alloc): Use it.
2425 * Makefile.in (GTFILES): Add calls.c.
2426 * calls.c: Include "gt-calls.h".
2428 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
2430 PR rtl-optimization/79728
2431 * regs.h (struct target_regs): New field
2432 x_contains_allocatable_regs_of_mode.
2433 (contains_allocatable_regs_of_mode): New macro.
2434 * reginfo.c (init_reg_sets_1): Initialize it, and change
2435 contains_reg_of_mode so it includes global regs as well.
2436 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
2437 rather than contains_regs_of_mode.
2439 2017-03-14 Martin Liska <mliska@suse.cz>
2441 * doc/invoke.texi: Document options that can't be combined with
2442 -fcheck-pointer-bounds.
2444 2017-03-14 Martin Liska <mliska@suse.cz>
2447 * doc/invoke.texi (-Wchkp): Document the option.
2449 2017-03-14 Martin Liska <mliska@suse.cz>
2451 * Makefile.in: Install gcov-dump.
2453 2017-03-14 Martin Liska <mliska@suse.cz>
2455 * multiple_target.c (expand_target_clones): Bail out for
2456 an invalid attribute.
2458 2017-03-14 Richard Biener <rguenther@suse.de>
2460 * alias.c (struct alias_set_entry): Pack properly.
2461 * cfgloop.h (struct loop): Likewise.
2462 * cse.c (struct set): Likewise.
2463 * ipa-utils.c (struct searchc_env): Likewise.
2464 * loop-invariant.c (struct invariant): Likewise.
2465 * lra-remat.c (struct cand): Likewise.
2466 * recog.c (struct change_t): Likewise.
2467 * rtl.h (struct address_info): Likewise.
2468 * symbol-summary.h (function_summary): Likewise.
2469 * tree-loop-distribution.c (struct partition): Likewise.
2470 * tree-object-size.c (struct object_size_info): Likewise.
2471 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
2472 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
2473 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
2474 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
2475 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
2476 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
2477 (struct _stmt_vec_info): Likewise.
2479 2017-03-14 Martin Liska <mliska@suse.cz>
2482 * multiple_target.c (create_dispatcher_calls): Check that
2483 a target can create a function dispatcher.
2485 2017-03-14 Martin Liska <mliska@suse.cz>
2488 * multiple_target.c (expand_target_clones): Drop local.local
2489 flag for default implementation.
2491 2017-03-14 Richard Biener <rguenther@suse.de>
2493 PR tree-optimization/80030
2494 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
2496 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
2498 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
2499 gcc_fallthrough() instead of __attribute__((fallthrough));
2501 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
2503 * doc/gcc.texi: Remove "up" link to (DIR).
2504 * doc/gccint.texi: Ditto.
2506 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
2508 * doc/install.texi (Specific) <avr>: Remove reference to
2511 2017-03-13 Jeff Law <law@redhat.com>
2513 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
2514 attribute rather than comments.
2516 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
2517 match_scratch operand is highest.
2519 2017-03-13 Martin Liska <mliska@suse.cz>
2522 * ipa-pure-const.c (warn_function_noreturn): If the declarations
2523 is a CHKP clone, use original declaration.
2525 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2527 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
2528 (arc_conditional_register_usage): Use a different allocation order
2529 when optimizing for size.
2530 * common/config/arc/arc-common.c (arc_option_optimization_table):
2531 Section anchors default on when optimizing for size.
2533 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2535 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
2537 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2539 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
2540 * config/arc/arc.md (cpu_facility): Add cd variant.
2541 (*movqi_insn): Add code density variant.
2542 (*movhi_insn): Likewise.
2543 (*movqi_insn): Likewise.
2544 (*addsi3_mixed): Likewise.
2545 (subsi3_insn): Likewise.
2547 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2549 * config/arc/arc.md (movsi_cond_exec): Update constraint.
2551 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
2553 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
2554 expressions with MINUS and UNARY ops.
2556 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2559 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
2561 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
2562 between vec_select and vector argument.
2563 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
2564 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
2565 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
2566 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
2567 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
2568 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
2570 2017-03-13 Richard Biener <rguenther@suse.de>
2573 * params.def (vect-max-peeling-for-alignment): Fix typo.
2575 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2577 * doc/install.texi (Specific) <mips-*-*>: Remove description of
2578 issue that only occurred with binutils below 2.18.
2580 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2582 * doc/install.texi (Specific) <cris-axis-elf>: No longer
2583 refer to binutils 2.11/2.12 minimum.
2585 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
2587 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
2588 ftp.kernel.org and simplify binutils requirement.
2590 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
2592 * doc/invoke.texi (Warning Options): Fix spelling of link-time
2594 (Optimize Options): Ditto. Also remove redundancy.
2596 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2598 PR translation/79848
2599 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
2601 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
2602 to G_ to avoid double translation.
2604 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2606 PR translation/79923
2607 * auto-profile.c (get_combined_location): Convert leading
2608 character of diagnostics to lower case and remove trailing period.
2609 (read_profile): Likewise for various diagnostics.
2610 * config/arm/arm.c (arm_option_override): Remove trailing period
2611 from various diagnostics.
2612 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
2613 (msp430_expand_delay_cycles): Likewise.
2615 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2618 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
2619 full command-line argument, rather than just "str".
2620 (aarch64_validate_march): Likewise.
2621 (aarch64_validate_mtune): Likewise.
2623 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
2625 PR rtl-optimization/78911
2626 * lra-assigns.c (must_not_spill_p): New function.
2627 (spill_for): Use it.
2629 2017-03-10 Jakub Jelinek <jakub@redhat.com>
2631 PR tree-optimization/79981
2632 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
2633 ATOMIC_COMPARE_EXCHANGE ifn result.
2634 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
2635 IFN_ATOMIC_COMPARE_EXCHANGE.
2637 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2640 * opts.c (parse_sanitizer_options): Add missing question mark to
2641 "did you mean" message.
2643 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2645 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
2647 (VMULEUH_UNS): Likewise.
2648 (VMULOUB_UNS): Likewise.
2649 (VMULOUH_UNS): Likewise.
2650 * config/rs6000/rs6000.c (builtin_function_type): Remove
2651 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
2653 2017-03-10 David Malcolm <dmalcolm@redhat.com>
2656 * read-rtl-function.c (function_reader::read_rtx_operand): Update
2657 x with result of extra_parsing_for_operand_code_0.
2658 (function_reader::extra_parsing_for_operand_code_0): Convert
2659 return type from void to rtx, returning x. When reading
2660 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
2661 larger size containing struct block_symbol.
2663 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
2665 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
2666 -mfloat128-hardware without -m64.
2668 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
2671 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
2672 entries to the case statement that marks unsigned arguments to
2673 overloaded functions.
2675 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
2677 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
2678 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
2680 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
2683 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
2684 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
2686 2017-03-10 Martin Liska <mliska@suse.cz>
2690 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
2691 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
2694 2017-03-10 Olivier Hainque <hainque@adacore.com>
2696 * tree-switch-conversion (array_value_type): Start by resetting
2697 candidate type to it's main variant.
2699 2017-03-10 Jakub Jelinek <jakub@redhat.com>
2701 PR rtl-optimization/79909
2702 * combine.c (try_combine): Use simplify_replace_rtx on individual
2703 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
2704 of the whole CALL_INSN_FUNCTION_USAGE.
2706 PR tree-optimization/79972
2707 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
2708 get_range_info on SSA_NAMEs. Formatting fixes.
2710 2017-03-10 Richard Biener <rguenther@suse.de>
2711 Jakub Jelinek <jakub@redhat.com>
2713 PR tree-optimization/77975
2714 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
2715 edge to be constant.
2716 (get_val_for): For constant x return it. Formatting fix.
2717 (loop_niter_by_eval): Avoid pointless looping if the next iteration
2718 would use the same bases as the current one.
2720 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2722 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
2723 instead of vec_select for V1TImode.
2724 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
2726 (VSX_LE_128): Add V1TI to this mode iterator.
2727 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
2728 (*vsx_le_perm_store_<mode>): Likewise.
2729 (pre-reload splitter for VSX stores): Likewise.
2730 (post-reload splitter for VSX stores): Likewise.
2731 (*vsx_xxpermdi2_le_<mode>): Likewise.
2732 (*vsx_lxvd2x2_le_<mode>): Likewise.
2733 (*vsx_stxvd2x2_le_<mode>): Likewise.
2735 2017-03-09 Michael Eager <eager@eagercon.com>
2737 Correct failures with --enable-checking=yes,rtl.
2739 * config/microblaze/microblaze.c (microblaze_expand_shift):
2740 Replace GET_CODE test with CONST_INT_P and INTVAL test with
2741 test for const0_rtx.
2742 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
2743 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
2745 2017-03-09 Richard Biener <rguenther@suse.de>
2747 PR tree-optimization/79977
2748 * graphite-scop-detection.c (scop_detection::merge_sese):
2749 Handle the case of extra exits to blocks dominating the entry.
2751 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
2753 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
2756 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
2758 PR rtl-optimization/79949
2759 * lra-constraints.c (process_alt_operands): Check memory when
2760 trying to predict a cycle. Print about the overall increase.
2762 2017-03-09 Richard Biener <rguenther@suse.de>
2765 * gimple-expr.c (useless_type_conversion_p): Preserve
2766 TYPE_SATURATING for fixed-point types.
2768 2017-03-09 Richard Biener <rguenther@suse.de>
2771 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
2772 alignment of BLKmode params.
2774 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2777 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
2778 (VALL_NO_V2Q): Likewise.
2780 * config/aarch64/aarch64-simd.md
2781 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
2783 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
2784 VALL_NO_V2Q mode iterator.
2785 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
2787 2017-03-09 Martin Liska <mliska@suse.cz>
2789 PR tree-optimization/79631
2790 * tree-chkp-opt.c (chkp_is_constant_addr): Call
2791 tree_int_cst_sign_bit just for INTEGER constants.
2793 2017-03-09 Martin Liska <mliska@suse.cz>
2797 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
2800 2017-03-09 Marek Polacek <polacek@redhat.com>
2803 * tree.c (inchash::add_expr): Handle TREE_VEC.
2805 2017-03-09 Martin Liska <mliska@suse.cz>
2808 (chkp_narrow_size_and_offset): New function.
2809 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
2810 (void chkp_parse_bit_field_ref): New function.
2811 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
2812 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
2814 2017-03-09 Martin Liska <mliska@suse.cz>
2817 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
2818 (chkp_find_bounds_1): Remove gcc_unreachable.
2820 2017-03-09 Jakub Jelinek <jakub@redhat.com>
2823 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
2824 BUILT_IN_SYNC*, determine the access type from the size suffix and
2825 always build a MEM_REF with that type. Handle forgotten
2826 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
2829 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
2830 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
2831 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
2832 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
2833 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
2834 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
2835 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
2836 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
2837 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
2838 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
2839 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
2840 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
2841 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
2842 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
2843 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
2844 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
2845 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
2846 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
2847 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
2848 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
2849 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
2850 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
2851 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
2852 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
2853 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
2854 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
2855 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
2856 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
2857 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
2858 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
2859 definitions outside of __OPTIMIZE__ guarded section.
2862 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
2863 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
2864 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
2865 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
2868 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2870 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
2871 ("vfenez<mode>"): Add missing constraints.
2873 2017-03-08 Martin Sebor <msebor@redhat.com>
2876 * config/nds32/nds32.c (nds32_option_override):
2877 Fix misspelled diagnostic.
2879 2017-03-08 Jakub Jelinek <jakub@redhat.com>
2882 * gimplify.c (gimplify_omp_for): Replace index var in outer
2883 taskloop statement with an artificial variable and add
2884 OMP_CLAUSE_PRIVATE clause for it.
2886 2017-03-08 Richard Biener <rguenther@suse.de>
2888 PR tree-optimization/79955
2889 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
2890 for accesses that are completely outside of the variable.
2892 2017-03-08 Andrew Haley <aph@redhat.com>
2894 PR tree-optimization/79943
2895 * tree-ssa-loop-split.c (compute_new_first_bound): When
2896 calculating the new upper bound, (END-BEG) should be added, not
2899 2017-03-08 Jakub Jelinek <jakub@redhat.com>
2901 * config/avr/avr.md (setmemhi): Make sure match_dup
2902 operand number comes before match_scratch.
2904 2017-03-08 Richard Biener <rguenther@suse.de>
2906 PR tree-optimization/79920
2907 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
2908 with ncopies == 1 to ...
2909 (vect_transform_slp_perm_load): ... here. Properly compute
2910 all element loads by iterating VF times over the group. Do
2911 not handle ncopies (computed in a broken way) in
2912 vect_create_mask_and_perm.
2914 2017-03-08 Jakub Jelinek <jakub@redhat.com>
2917 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
2918 is a uniform vector, use uniform_vector_p return value instead of
2919 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
2921 2017-03-07 Marek Polacek <polacek@redhat.com>
2924 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
2925 (alloca_call_type): Likewise.
2927 2017-03-07 Martin Liska <mliska@suse.cz>
2929 * gcov.c (process_args): Put comment to correct location.
2931 2017-03-07 Martin Liska <mliska@suse.cz>
2934 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
2935 Use array_at_struct_end_p instead of DECL_CHAIN (field).
2936 (chkp_narrow_bounds_for_field): Likewise.
2937 (chkp_parse_array_and_component_ref): Pass one more argument to
2940 2017-03-07 Richard Biener <rguenther@suse.de>
2942 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
2945 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
2947 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
2948 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
2950 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2953 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
2954 to end of description.
2955 (PARAM_MAX_STORES_TO_MERGE): Likewise.
2957 2017-03-07 Jakub Jelinek <jakub@redhat.com>
2959 PR rtl-optimization/79901
2960 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
2962 (*avx512f_<code><mode>3<mask_name>): ... this.
2963 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
2964 iterator instead of VI8_AVX2_AVX512BW.
2966 PR rtl-optimization/79901
2967 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
2968 min/max expander, expand it using expand_vec_cond_expr.
2971 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
2974 2017-03-06 Jakub Jelinek <jakub@redhat.com>
2977 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
2978 to void * for PCH reasons.
2979 * dwarf2out.c (output_loc_operands, output_die): Cast
2980 v.val_vec.array to unsigned char *.
2982 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
2985 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
2988 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
2990 PR rtl-optimization/79571
2991 * lra-constraints.c (process_alt_operands): Calculate static
2992 reject and subtract it from overall when only addresses will be
2995 2017-03-06 Julia Koval <julia.koval@intel.com>
2998 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
2999 incoming stack boundary to 128 for 64-bit targets.
3001 2017-03-06 Richard Biener <rguenther@suse.de>
3003 PR tree-optimization/79894
3004 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
3005 to NULL after folding it.
3007 2017-03-06 Richard Biener <rguenther@suse.de>
3009 PR tree-optimization/79824
3010 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
3011 check disabling peeling for gaps.
3013 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
3015 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
3016 attributes): Document gettimeofday.
3018 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
3020 * config/s390/s390.c (s390_option_override_internal): Set
3021 PARAM_MIN_VECT_LOOP_BOUND
3023 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
3025 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
3026 * config/s390/s390.md: Likewise.
3028 2017-03-06 Jakub Jelinek <jakub@redhat.com>
3031 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
3032 (<avx2_avx512>_perm<mode>): Rename to ...
3033 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
3035 (<avx512>_perm<mode>_mask): Rename to ...
3036 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
3038 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
3039 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
3040 instead of VI8F_256_512.
3041 (avx512f_perm<mode>): New define_expand.
3042 (avx512f_perm<mode>_mask): Likewise.
3043 (avx512f_perm<mode>_1<mask_name>): New define_insn.
3044 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
3046 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
3048 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
3049 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
3051 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
3053 2017-03-06 Martin Liska <mliska@suse.cz>
3056 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
3057 when having a SSA NAME w/o VAR_DECL assigned to it.
3059 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
3061 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
3062 msa_dpsub_<su>_d): Fix MODE for vec_select.
3064 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
3066 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
3068 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
3070 2017-03-06 Richard Biener <rguenther@suse.de>
3072 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
3073 * plugin.c (register_plugin_info): Likewise.
3074 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
3076 2017-03-05 Jakub Jelinek <jakub@redhat.com>
3078 * config/i386/sse.md (sse_storehps, sse_storelps,
3079 avx_<castmode><avxsizesuffix>_<castmode>,
3080 avx512f_<castmode><avxsizesuffix>_<castmode>,
3081 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
3082 in condition that at least one operand is not a MEM.
3084 2017-03-03 Jakub Jelinek <jakub@redhat.com>
3087 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
3088 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
3090 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
3091 gimple_call_nothrow_p flag based on whether original builtin can throw.
3092 If it can, emit following stmts on the fallthrough edge.
3093 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
3094 don't create new bb if inserting just debug stmts on the edge, try to
3095 insert them on the fallthru bb or just reset debug stmts.
3097 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
3100 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
3101 restore recog_data (including the operand rtxes inside it) around
3102 the call to get_insn_template.
3104 2017-03-03 Martin Sebor <msebor@redhat.com>
3106 PR tree-optimization/79699
3107 * context.c (context::~context): Free MPFR caches to avoid
3108 a memory leak on program exit.
3110 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3112 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
3113 Use wide_int::ulow () instead of .elt (0).
3115 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
3117 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
3118 (*pushxf): Limit oF constraint to 32bit targets and add oC
3119 constraint for 64bit targets.
3120 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
3121 (*pushdf): Change rmF constraint to rmC.
3123 2017-03-03 Martin Liska <mliska@suse.cz>
3125 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
3126 Remove unused variable.
3128 2017-03-03 Jakub Jelinek <jakub@redhat.com>
3131 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
3132 is a memory operand, increase num_memory.
3133 (ix86_expand_args_builtin): Likewise.
3135 2017-03-03 Jan Hubicka <jh@suse.cz>
3138 * ipa-devirt.c (maybe_record_node): Properly handle
3139 __cxa_pure_virtual visibility.
3141 2017-03-03 Martin Liska <mliska@suse.cz>
3143 PR tree-optimization/79803
3144 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
3146 (pass_loop_prefetch::execute): Disabled optimization if an
3147 assumption about L1 cache size is not met.
3149 2017-03-03 Martin Liska <mliska@suse.cz>
3151 PR rtl-optimization/79574
3152 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
3153 (hash_scan_set): Likewise.
3154 (dump_hash_table): Likewise.
3155 (hoist_code): Likewise.
3157 2017-03-03 Richard Biener <rguenther@suse.de>
3159 * fixed-value.c (fixed_from_string): Restore use of elt (1)
3160 in place of uhigh ().
3161 (fixed_convert_from_real): Likewise.
3163 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
3166 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
3168 2017-03-03 Richard Biener <rguenther@suse.de>
3171 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
3172 TYPE_OVERFLOW_UNDEFINED check.
3174 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3176 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
3178 (vector_ae_<mode>_p): Likewise.
3179 (vector_nez_<mode>_p): Likewise.
3180 (vector_ne_v2di_p): Likewise.
3181 (vector_ae_v2di_p): Likewise.
3182 (vector_ne_<mode>_p): Likewise.
3183 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
3185 (vsx_tsqrt<mode>2_fe): Likewise.
3187 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
3190 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
3192 2017-03-02 Jakub Jelinek <jakub@redhat.com>
3194 PR rtl-optimization/79780
3195 * cprop.c (one_cprop_pass): When second and further conditional trap
3196 in a single basic block is turned into an unconditional trap, turn it
3197 into a deleted note to avoid RTL verification failures.
3199 2017-03-02 Richard Biener <rguenther@suse.de>
3201 * fold-const.c (const_binop): Use ulow () instead of elt (0).
3203 2017-03-02 Richard Biener <rguenther@suse.de>
3205 PR tree-optimization/79345
3207 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
3208 param and abort the walk, returning -1 if it is hit.
3209 (walk_aliased_vdefs): Take a limit param and pass it on.
3210 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
3211 defaulting to 0 and return a signed int.
3212 * tree-ssa-uninit.c (struct check_defs_data): New struct.
3213 (check_defs): New helper.
3214 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
3215 about uninitialized memory.
3216 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
3217 bogus uninitialized warning.
3218 (fixed_convert_from_real): Likewise.
3220 2017-03-02 Bin Cheng <bin.cheng@arm.com>
3222 PR tree-optimization/66768
3223 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
3224 iv_use if base object can't be determined.
3226 2017-03-02 Jakub Jelinek <jakub@redhat.com>
3228 PR tree-optimization/79345
3229 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
3230 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
3231 (get_pattern_stats): Initialize it.
3232 * genemit.c (gen_expand): Verify match_scratch numbers come after
3233 match_operand/match_dup numbers.
3234 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
3235 match_scratch numbers.
3236 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
3238 * config/s390/s390.md (trunctdsd2): Likewise.
3240 2017-03-02 Richard Biener <rguenther@suse.de>
3242 * wide-int.h (wide_int_storage::operator=): Implement in terms
3245 2017-03-02 Richard Biener <rguenther@suse.de>
3247 PR tree-optimization/79777
3248 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
3249 the to insert expression to sth existing.
3251 2017-03-01 Martin Sebor <msebor@redhat.com>
3254 * gimple-ssa-sprintf.c
3255 (directive::known_width_and_precision): New function.
3256 (format_integer): Use it.
3257 (get_mpfr_format_length): Consider the full range of precision
3258 when computing %g output with the # flag. Set the likely byte
3259 count to 3 rather than 1 when precision is indeterminate.
3260 (format_floating): Correct the lower bound of precision.
3262 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3264 * doc/invoke.texi: Document default code model for 64-bit Linux.
3266 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3269 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
3270 udiv rather than div since input pattern is unsigned.
3272 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
3274 * config/i386/i386.c (print_reg): Warn for values of
3275 unsupported size in integer register.
3277 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
3280 * config/rs6000/predicates.md (current_file_function_operand): Do
3281 not allow self calls to be local if the function is replaceable.
3283 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
3286 * config/rs6000/altivec.h (vec_ctz and others): Change the
3287 preprocessor macro that controls conditional compilation from
3288 _ARCH_PWR9 to __POWER9_VECTOR__.
3289 (vec_all_ne): Change parameterization of __altivec_scalar_pred
3290 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
3291 control (instead of _ARCH_PWR9 control) so that template
3292 definition uses power9-specific function.
3293 (vec_any_eq): Likewise.
3294 (vec_all_ne): Change macro definition to use a power9-specific
3295 expansion under #ifdef __POWER9_VECTOR__ control (instead of
3296 _ARCH_PWR9 control).
3297 (vec_any_eq) Likewise.
3298 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
3299 expansion for CMPNEF to remove support for xvcmpnesp instruction.
3300 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
3301 support for xvcmpnedp instruction.
3302 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
3303 macro expansion so that Power9 implementation of vec_all_ne does
3304 not use the AltiVec predicate framework.
3305 (VCMPNEH_P): Likewise.
3306 (VCMPNEW_P): Likewise.
3307 (VCMPNED_P): Likewise.
3308 (VCMPNEFP_P): Likewise.
3309 (VCMPNEDP_P): Likewise.
3310 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
3311 implementation of vec_any_eq to not use AltiVec predicate
3313 (VCMPAEH_P): Likewise.
3314 (VCMPAEW_P): Likewise.
3315 (VCMPAED_P): Likewise.
3316 (VCMPAEFP_P): Likewise.
3317 (VCMPAEDP_P): Likewise.
3318 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
3319 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
3320 not use the AltiVec predicate framework.
3321 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
3322 of vec_any_eq to not use AltiVec predicate framework.
3323 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
3324 support for predefined __POWER9_VECTOR__ macro to indicate that
3325 Power9 instruction selection is enabled.
3326 (altivec_overloaded_builtins): Remove extraneous
3327 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
3328 function argument types RS6000_BTI_bool_V16QI and
3329 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
3330 entry for overloaded function argument types RS6000_BTI_bool_V4SI
3331 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
3332 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
3333 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
3334 Power9 for implementations of vec_cmpne. Change the signature for
3335 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
3336 (representing vec_all_ne) to remove the previously described first
3337 argument of type RS6000_BTI_INTSI, as this was an artifact of
3338 reliance on the AltiVec predicate framework, which is no longer
3339 used in the implementation of these functions. Add
3340 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
3341 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
3342 since, unlike the AltiVec predicate framework implementation, we
3343 do not share function descriptors between vec_alle and vec_anyeq.
3344 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
3345 set of modes that receive special treatment even when
3346 TARGET_P9_VECTOR is true. The special treatment emits code that
3347 does not depend on Power9 instructions.
3348 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
3349 define_expand to not rely on AltiVec predicate framework.
3350 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
3352 (vector_ne_v2di_p): Change this define_expand to not rely on
3353 AltiVec predicate framework.
3354 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
3356 (vector_ne_<mode>_p): Change this define_expand to not rely on
3357 AltiVec predicate framework.
3358 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
3360 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
3361 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
3362 define_insn pattern.
3363 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
3364 define_insn pattern because the xvcmpne<VSs>. instruction is not
3366 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
3367 instruction is not supported.
3369 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3371 * config/nvptx/nvptx.c: Include intl.h.
3373 2017-03-01 Martin Jambor <mjambor@suse.cz>
3376 * ipa-prop.h (ipa_bits): Removed field known.
3377 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
3378 to pointers. Adjusted their comments to warn about their sharing.
3379 (ipcp_transformation_summary): Change bits to a vector of pointers.
3380 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
3381 (ipa_get_ipa_bits_for_value): Declare.
3382 * tree-vrp.h (value_range): Mark as GTY((for_user)).
3383 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
3384 (ipa_bits_hash_table): Likewise.
3385 (ipa_vr_ggc_hash_traits): Likewise.
3386 (ipa_vr_hash_table): Likewise.
3387 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
3388 being pointers and vr_known being removed.
3389 (ipa_set_jf_unknown): Likewise.
3390 (ipa_get_ipa_bits_for_value): New function.
3391 (ipa_set_jfunc_bits): Likewise.
3392 (ipa_get_value_range): New overloaded functions.
3393 (ipa_set_jfunc_vr): Likewise.
3394 (ipa_compute_jump_functions_for_edge): Use the above functions to
3395 construct bits and vr parts of jump functions.
3396 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
3397 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
3399 (ipcp_grow_transformations_if_necessary): Also allocate
3400 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
3402 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
3403 them. Fix too long lines.
3404 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
3405 vr_known being removed.
3406 (ipa_read_jump_function): Use new setter functions to construct bits
3407 and vr parts of jump functions or set them to NULL.
3408 (write_ipcp_transformation_info): Adjust for bits being pointers.
3409 (read_ipcp_transformation_info): Likewise.
3410 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
3412 Include gt-ipa-prop.h.
3413 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
3415 (ipcp_store_bits_results): Likewise.
3416 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
3417 Do not write to existing jump functions but use a temporary instead.
3419 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3422 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
3423 attempt to use its first operand as BIT_FIELD_REF base.
3425 2017-03-01 Richard Biener <rguenther@suse.de>
3428 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
3429 interpolating formula in wrapping arithmetic.
3430 (chrec_apply): Convert chrec_evaluate return value to wanted type.
3432 2017-03-01 Jakub Jelinek <jakub@redhat.com>
3434 PR tree-optimization/79734
3435 * tree-vect-generic.c (expand_vector_condition): Optimize
3436 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
3437 Handle VEC_COND_EXPR where comparison has different inner width from
3440 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
3442 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
3443 markup, and similar issues. Remove @opindex entries for things
3444 that aren't options. Add missing -mmpy-option entries.
3446 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3448 PR tree-optimization/79737
3449 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
3450 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
3451 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
3452 instead of byte_size. Formatting fix.
3453 (shift_bytes_in_array_right): Formatting fix.
3455 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
3458 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
3459 condition on optimize for the leaf function test.
3461 2017-02-28 Martin Liska <mliska@suse.cz>
3464 * read-rtl-function.c (function_reader::handle_unknown_directive):
3465 Bail out when one uses -flto.
3467 2017-02-28 Martin Liska <mliska@suse.cz>
3469 * common.opt: Replace space with tabular for options of <number>
3471 * config/i386/i386.opt: Show <number> value for
3472 -mlarge-data-threshold.
3473 * opts.c (print_filtered_help): Do not display number in hexadecimal
3476 2017-02-28 Martin Liska <mliska@suse.cz>
3478 * common.opt: Fix --help=option -Q for options which are of
3481 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
3483 * config/i386/i386.c (print_reg): Error out for values
3484 of 8-bit size in invalid integer register.
3486 2017-02-28 Martin Sebor <msebor@redhat.com>
3488 PR tree-optimization/79691
3489 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
3491 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3494 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
3495 gcc_unreachable with output_operand_lossage.
3497 2017-02-28 Richard Biener <rguenther@suse.de>
3499 PR tree-optimization/79740
3500 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
3502 (visit_nary_op): Insert the nary into the hashtable if we
3503 pattern-matched sth.
3504 * tree-ssa-pre.c (eliminate_insert): Robustify.
3506 2017-02-28 Richard Biener <rguenther@suse.de>
3509 * fold-const.c (decode_field_reference): Reject out-of-bound
3512 2017-02-28 Jakub Jelinek <jakub@redhat.com>
3514 * config/i386/i386.c: Include intl.h.
3515 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
3516 instead of just cond ? "..." : "...".
3517 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
3518 * coverage.c (read_counts_file): Likewise.
3519 * omp-offload.c: Include intl.h.
3520 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
3521 of just cond ? "..." : "...".
3522 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
3523 of just cond ? "..." : "...".
3525 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
3528 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
3530 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
3531 'tune for' CPU name.
3532 * config/arm/arm-cpu-data.h: Regenerated.
3534 2017-02-28 Richard Biener <rguenther@suse.de>
3536 PR tree-optimization/79732
3537 * tree-inline.c (expand_call_inline): Do not shadow var.
3539 2017-02-28 Richard Biener <rguenther@suse.de>
3541 PR tree-optimization/79723
3542 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
3543 address-space properly.
3545 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
3547 * doc/optinfo.texi (Optimization groups): Fix option used for
3549 * doc/invoke.texi (-fopt-info): Document "omp".
3550 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
3551 (OPTGROUP_ALL): Add OPTGROUP_OMP.
3552 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
3553 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
3554 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
3556 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
3558 * dumpfile.c (optgroup_options): Instead of "openmp", associate
3559 OPTGROUP_OMP with "omp".
3561 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
3564 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
3565 for arithmetic shift of unsigned V2DI.
3567 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
3569 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
3570 arc/linux.h headers.
3571 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
3572 (LINK_SPEC): Likewise.
3573 (ARC_TLS_EXTRA_START_SPEC): Likewise.
3574 (EXTRA_SPECS): Likewise.
3575 (STARTFILE_SPEC): Likewise.
3576 (ENDFILE_SPEC): Likewise.
3577 (LIB_SPEC): Likewise.
3578 (TARGET_SDATA_DEFAULT): Likewise.
3579 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
3580 (MULTILIB_DEFAULTS): Likewise.
3581 (DWARF2_UNWIND_INFO): Likewise.
3582 * config/arc/big.h: New file.
3583 * config/arc/elf.h: Likewise.
3584 * config/arc/linux.h: Likewise.
3585 * config/arc/t-uClibc: Remove.
3587 2017-02-27 Bin Cheng <bin.cheng@arm.com>
3589 PR tree-optimization/77536
3590 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
3591 (tree_transform_and_unroll_loop): Use above function to compute the
3592 estimated niter of unrolled loop and use it when scaling profile.
3593 Also use count info rather than frequency if it's non-zero.
3594 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
3595 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
3596 (vect_transform_loop): Call above function.
3598 2017-02-27 Richard Biener <rguenther@suse.de>
3600 PR tree-optimization/45397
3601 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
3602 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
3603 (visit_nary_op): Add pattern matching for CSEing sign-changed
3604 or truncated operations with wider ones.
3606 2017-02-27 Richard Biener <rguenther@suse.de>
3608 PR tree-optimization/79690
3609 * tree-vect-stmts.c (vectorizable_store): Use vector type
3610 built from the DR with address-space.
3612 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
3614 * doc/invoke.texi (Optimize Options): Refine the description
3615 of asan-use-after-return.
3617 2017-02-25 Alan Modra <amodra@gmail.com>
3619 PR rtl-optimization/79584
3620 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
3621 base, not ad->base_term, the reg within base. Remove assertion
3622 that ad->base == ad->base_term. Replace gen_int_mode using
3623 bogus mode with const0_rtx.
3625 2017-02-25 Jakub Jelinek <jakub@redhat.com>
3628 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
3629 FMA_EXPR like tcc_binary or tcc_unary.
3631 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
3634 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
3636 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
3637 (output_loc_operands): Handle DW_OP_call_ref and
3638 DW_OP_GNU_variable_value.
3639 (struct variable_value_struct): New type.
3640 (struct variable_value_hasher): Likewise.
3641 (variable_value_hash): New variable.
3642 (string_types): Remove.
3643 (copy_loc_descr): New function.
3644 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
3645 (prepend_loc_descr_to_each): New function.
3646 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
3647 instead of add_loc_descr_to_each if the first argument is single
3648 location list and the second has multiple.
3649 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
3650 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
3651 when looking for variable value which doesn't have other location info.
3652 (loc_list_from_tree): Formatting fix.
3653 (gen_array_type_die): Simplify DW_AT_string_length handling.
3654 (adjust_string_types): Remove.
3655 (gen_subprogram_die): Don't call adjust_string_types nor test/set
3656 string_types. Call resolve_variable_values.
3657 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
3658 (resolve_addr_in_expr): Likewise. Add A argument.
3659 (copy_deref_exprloc): Remove deref argument. Adjust for the
3660 original expression being DW_OP_GNU_variable_value with optionally
3661 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
3662 optionally after it.
3663 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
3664 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
3665 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
3666 (variable_value_hasher::hash, variable_value_hasher::equal): New
3668 (resolve_variable_value_in_expr, resolve_variable_value,
3669 resolve_variable_values, note_variable_value_in_expr,
3670 note_variable_value): New functions.
3671 (dwarf2out_early_finish): Call note_variable_value on all toplevel
3674 2017-02-24 Jakub Jelinek <jakub@redhat.com>
3677 * opts.h (handle_generated_option): Add GENERATED_P argument.
3678 * opts-common.c (handle_option): Adjust function comment.
3679 (handle_generated_option): Add GENERATED_P argument, pass it to
3681 (control_warning_option): Pass false to handle_generated_option
3683 * opts.c (maybe_default_option): Pass true to handle_generated_option
3685 * optc-gen.awk: Likewise.
3687 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3689 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
3690 a REG, look at the REG it is a SUBREG of.
3691 (splitter for cmpeqsi_t): Ditto.
3693 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3695 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
3696 the special USEs with the pattern of the insn, not the insn itself.
3698 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
3701 * doc/invoke.texi: Document -mload-store-pairs.
3703 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3704 Sandra Loosemore <sandra@codesourcery.com>
3706 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
3707 argument isn't a CONST_INT.
3708 (nios2_alternate_compare_const): Assert op is a CONST_INT.
3709 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
3710 (nios2_validate_compare): Bypass alternate compare logic if *op2
3712 (ldstwm_operation_p): Return false if first_base is not a REG or
3713 if first_offset is not a CONST_INT.
3715 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3717 * config/cris/cris.md: Use correct operand in a define_peephole2.
3719 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3721 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
3723 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
3725 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
3726 this_insn if it is an INSN or JUMP_INSN.
3727 (force_offsettable): Look at base, not at addr.
3728 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
3729 on things that aren't necessarily CONST_INTs.
3731 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
3733 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
3734 -mfpmath=sse is the default also for x86-32 targets with SSE2
3735 instruction set when @option{-ffast-math} is enabled
3737 2017-02-24 Jeff Law <law@redhat.com>
3739 PR rtl-optimizatoin/79286
3740 * ira.c (update_equiv_regs): Drop may_trap_p exception to
3743 2017-02-24 Richard Biener <rguenther@suse.de>
3745 PR tree-optimization/79389
3746 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
3749 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
3751 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
3752 function comment to reflect reality.
3753 (loop_exits_before_overflow): Fix typo in function description.
3755 2017-02-24 Richard Biener <rguenther@suse.de>
3757 PR tree-optimization/79389
3758 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
3759 properly that a threading opportunity exists. Detect conditional
3760 copy/constant propagation opportunities.
3762 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
3764 * config/visium/visium.md (type): Add trap.
3765 (b): New mode attribute.
3766 (*btst): Rename into...
3767 (*btst<mode>): ...this and adjust.
3768 (*cbranchsi4_btst_insn): Rename into...
3769 (*cbranch<mode>4_btst_insn): ...this and adjust.
3770 (trap): New define_insn.
3772 2017-02-23 Jakub Jelinek <jakub@redhat.com>
3774 PR tree-optimization/79389
3775 * ifcvt.c (struct noce_if_info): Add rev_cond field.
3776 (noce_reversed_cond_code): New function.
3777 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
3778 reversed_comparison_code. Formatting fix.
3779 (noce_try_store_flag): Test rev_cond != NULL in addition to
3780 reversed_comparison_code.
3781 (noce_try_store_flag_constants): Likewise.
3782 (noce_try_store_flag_mask): Likewise.
3783 (noce_try_addcc): Use rev_cond if non-NULL instead of
3784 reversed_comparison_code.
3785 (noce_try_cmove_arith): Likewise. Formatting fixes.
3786 (noce_try_minmax, noce_try_abs): Clear rev_cond.
3787 (noce_find_if_block): Initialize rev_cond.
3788 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
3789 instead of false as last argument never attempt to reverse it
3792 2017-02-23 Bin Cheng <bin.cheng@arm.com>
3794 PR tree-optimization/79663
3795 * tree-predcom.c (combine_chains): Process refs in reverse order
3796 only for ZERO length chains, and add explaining comment.
3798 2017-02-23 Jeff Law <law@redhat.com>
3800 PR tree-optimization/79578
3801 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
3802 in call to operand_equal_p.
3804 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
3807 * config/i386/cpuid.h: Fix another undefined behavior.
3809 2017-02-23 Richard Biener <rguenther@suse.de>
3811 PR tree-optimization/79683
3812 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
3813 vector types for data-refs.
3815 2017-02-23 Martin Liska <mliska@suse.cz>
3817 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
3819 2017-02-23 Jakub Jelinek <jakub@redhat.com>
3822 * internal-fn.c (get_range_pos_neg): Moved to ...
3823 * tree.c (get_range_pos_neg): ... here. No longer static.
3824 * tree.h (get_range_pos_neg): New prototype.
3825 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
3826 are known to be in between 0 and signed maximum inclusive, try to
3827 expand both unsigned and signed divmod and use the cheaper one from
3830 2017-02-22 Jeff Law <law@redhat.com>
3832 PR tree-optimization/79578
3833 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
3834 to compare base operands.
3836 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
3839 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
3840 gpc_reg_operand instead of fpr_reg_operand.
3842 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
3844 * config/mips/mips.c (mips_return_in_memory): Force FP
3845 vector types to be returned in memory for o32 ABI.
3847 2017-02-22 Jakub Jelinek <jakub@redhat.com>
3849 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
3850 instead of DW_TAG_member for static data member declarations and don't
3851 set no_linkage_name for static inline data members.
3852 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
3855 2017-02-22 Martin Liska <mliska@suse.cz>
3857 * doc/invoke.texi: Replace inequality signs with square brackets
3860 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3862 PR tree-optimization/68644
3863 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
3865 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
3868 * lra-constraints.c (simplify_operand_subreg): Handle
3869 WORD_REGISTER_OPERATIONS targets.
3871 2017-02-22 Jakub Jelinek <jakub@redhat.com>
3874 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
3875 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
3876 elimination by swapping fld*.
3878 2017-02-22 Richard Biener <rguenther@suse.de>
3880 PR tree-optimization/79673
3881 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
3882 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
3883 irrelevant address-space qualifiers and avoiding a
3884 ADDR_SPACE_CONVERT_EXPR from fold_convert.
3886 2017-02-22 Richard Biener <rguenther@suse.de>
3888 PR tree-optimization/79666
3889 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
3890 to not symbolically negate if that may introduce undefined
3893 2017-02-22 Martin Liska <mliska@suse.cz>
3896 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
3897 * data-streamer-out.c (streamer_write_gcov_count_stream):
3899 * value-prof.c (stream_out_histogram_value): Make assert more
3900 precise based on type of counter.
3902 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
3905 * config/i386/i386.md (standard_x87sse_constant_load splitter):
3906 Use nonimmediate_operand instead of memory_operand for operand 1.
3907 (float-extend standard_x87sse_constant_load splitter): Ditto.
3909 2017-02-21 Jeff Law <law@redhat.com>
3911 PR tree-optimization/79621
3912 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
3913 blocks with edges to themselves.
3915 2017-02-21 Jakub Jelinek <jakub@redhat.com>
3918 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
3919 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
3920 Use gimple_call_builtin_p.
3923 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
3924 on temporarily removed DEBUG_INSNs.
3926 PR tree-optimization/79649
3927 * tree-loop-distribution.c (classify_partition): Give up on
3928 non-generic address space loads/stores.
3930 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
3932 * doc/loop.texi (Loop manipulation): Remove nonexistent
3933 tree_ssa_loop_version from the documentation.
3934 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
3936 2017-02-21 Jakub Jelinek <jakub@redhat.com>
3939 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
3940 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
3941 * config/rs6000/rs6000.c: Include except.h.
3942 (rs6000_expand_split_stack_prologue): Call
3943 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
3945 2017-02-21 Martin Jambor <mjambor@suse.cz>
3948 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
3951 2017-02-21 Martin Jambor <mjambor@suse.cz>
3953 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
3954 for backward compatibility only.
3955 * doc/invoke.texi (Option Summary): Remove all references to
3958 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
3962 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
3964 * lra-constraints.c (curr_insn_transform): Handle
3965 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
3967 2017-02-21 Martin Liska <mliska@suse.cz>
3969 * config/i386/i386.opt: Replace -masm-dialect with -masm.
3971 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
3973 PR translation/79638
3974 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
3976 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
3979 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
3980 (arm_function_ok_for_sibcall): Return false for an indirect call by
3981 descriptor if all the argument registers are used.
3982 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
3983 alignment of the function.
3985 2017-02-21 Jakub Jelinek <jakub@redhat.com>
3987 PR tree-optimization/61441
3988 * simplify-rtx.c (simplify_const_unary_operation): For
3989 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
3990 the sNaN unmodified.
3992 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
3994 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
3995 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
3996 instead of SYSTEM_HEADER_DIR.
3998 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
3999 Martin Liška <mliska@suse.cz>
4001 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
4002 Fix typos and grammar, use active voice, and clarify.
4004 2017-02-20 Marek Polacek <polacek@redhat.com>
4007 * gimplify.c (gimplify_expr): Handle unused *&&L;.
4010 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
4012 2017-02-20 Jakub Jelinek <jakub@redhat.com>
4015 * config/i386/i386.c (ix86_expand_builtin): Handle
4016 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
4017 ix86_builtins_isa[fcode].isa as a requirement of those
4018 flags and any other flag in the bitmask.
4019 (ix86_init_mmx_sse_builtins): Use 0 instead of
4020 ~OPTION_MASK_ISA_64BIT as mask.
4021 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
4022 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
4023 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
4024 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
4026 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4029 * lra-constraints.c (split_reg): Check requested split mode
4030 is supported by the register.
4032 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4034 * lra-constraints.c (simplify_operand_subreg): Remove early
4037 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4040 * lra-constraints.c (curr_insn_transform): Tighten condition
4041 for converting SUBREG reloads from OP_OUT to OP_INOUT.
4043 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
4046 * lra-constraints.c (curr_insn_transform): Handle
4047 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
4049 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
4052 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
4054 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
4056 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
4059 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
4062 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4064 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
4067 2017-02-18 Jakub Jelinek <jakub@redhat.com>
4069 * final.c (last_columnnum, override_columnnum): New variables.
4070 (final_start_function): Set last_columnnum, pass it to begin_prologue
4071 hook and pass 0 to dwarf2out_begin_prologue.
4072 (final_scan_insn): Update override_columnnum. Pass last_columnnum
4073 to source_line debug hook.
4074 (notice_source_line): Compute last_columnnum and for debug_column_info
4075 return true on column changes.
4076 * debug.h (struct gcc_debug_hooks): Add column argument to
4077 source_line and begin_prologue hooks.
4078 (debug_nothing_int_charstar_int_bool): Remove prototype.
4079 (debug_nothing_int_int_charstar,
4080 debug_nothing_int_int_charstar_int_bool): New prototypes.
4081 (dwarf2out_begin_prologue): Add column argument.
4082 * debug.c (do_nothing_debug_hooks): Adjust source_line and
4083 begin_prologue hooks.
4084 (debug_nothing_int_charstar_int_bool): Remove.
4085 (debug_nothing_int_int_charstar,
4086 debug_nothing_int_int_charstar_int_bool): New functions.
4087 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
4088 through to dwarf2out_source_line.
4089 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
4090 (dwarf2out_source_line): Add column argument, emit it if requested.
4091 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
4093 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
4094 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
4095 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
4096 through to dwarf2out_begin_prologue.
4097 (vmsdbgout_source_line): Add column argument, pass it through to
4098 dwarf2out_source_line.
4099 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
4100 dbxout_source_line caller.
4101 (dbxout_source_line): Add column argument.
4103 * common.opt (gno-column-info, gcolumn-info): New options.
4104 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
4105 (check_die): Also test for multiple DW_AT_decl_column attributes.
4106 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
4107 DW_AT_decl_column if requested.
4108 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
4110 (gen_variable_die): Likewise.
4111 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
4112 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
4115 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
4116 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
4117 (ix86_handle_option): Handle OPT_m3dnowa.
4118 * doc/invoke.texi (-m3dnowa): Document.
4119 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
4120 -m3dnowa instead of -m3dnow -march=athlon.
4123 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
4124 instead of gcc_assert for K, r and R code checks. Formatting fixes.
4126 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4129 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
4130 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
4131 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
4132 generator for vsx_xxpermdi_<mode>_be.
4133 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
4134 force big-endian semantics.
4135 (vsx_xxpermdi_<mode>_be): New define_expand with same
4136 implementation as previous version of vsx_xxpermdi_<mode>.
4138 2017-02-17 Jakub Jelinek <jakub@redhat.com>
4140 PR tree-optimization/79327
4141 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
4142 variable, its initialization and use.
4144 2017-02-17 Julia Koval <julia.koval@intel.com>
4146 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
4147 (OPTION_MASK_ISA_PKU_UNSET): New.
4148 (ix86_handle_option): Handle -mrdpid.
4149 * config/i386/cpuid.h (bit_RDPID): New.
4150 * config/i386/driver-i386.c (host_detect_local_cpu):
4151 Detect RDPID feature.
4152 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
4153 * config/i386/i386-c.c (ix86_target_macros_internal):
4155 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
4156 (ix86_valid_target_attribute_inner_p): Add "rdpid".
4157 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
4158 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
4159 * config/i386/i386.md (define_insn "rdpid"): New.
4160 * config/i386/i386.opt Add -mrdpid.
4161 * config/i386/immintrin.h (_rdpid_u32): New.
4163 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
4165 PR rtl-optimization/79541
4166 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
4167 instead of transforming it into USE.
4169 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
4171 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
4172 If HONOR_SNANS (SFmode) force the input to a register.
4173 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
4174 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
4175 an frsp or similar insn.
4177 2017-02-17 Martin Liska <mliska@suse.cz>
4179 PR rtl-optimization/79577
4180 * params.def (selsched-max-sched-times): Increase minimum to 1.
4182 2017-02-17 Martin Liska <mliska@suse.cz>
4184 PR rtl-optimization/79574
4185 * gcse.c (want_to_gcse_p): Prevent integer overflow.
4187 2017-02-17 Martin Liska <mliska@suse.cz>
4189 PR tree-optimization/79529
4190 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
4191 ssa_defined_default_def_p to handle cases which are implicitly
4193 * tree-ssa.c (ssa_defined_default_def_p): New function.
4194 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
4195 which are implicitly defined.
4196 * tree-ssa.h (ssa_defined_default_def_p): Declare.
4198 2017-02-17 Richard Biener <rguenther@suse.de>
4201 * params.def (max-ssa-name-query-depth): Limit to 10.
4203 2017-02-17 Richard Biener <rguenther@suse.de>
4205 PR tree-optimization/79552
4206 * tree-ssa-structalias.c (visit_loadstore): Properly verify
4209 2017-02-17 Richard Biener <rguenther@suse.de>
4212 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
4214 2017-02-17 Marek Polacek <polacek@redhat.com>
4217 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
4218 (fold_negate_expr): New wrapper.
4220 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
4222 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
4223 Correct terminology and de-emphasize pre-standard behavior.
4225 2017-02-16 Alan Modra <amodra@gmail.com>
4227 PR rtl-optimization/79286
4228 * ira.c (def_dominates_uses): New function.
4229 (update_equiv_regs): Don't create an equivalence for insns that
4230 may trap where the register def does not dominate the use.
4232 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
4234 PR rtl-optimization/78127
4235 * lra.c (lra): Call lra_eliminate before finish the loop after
4238 2017-02-16 Richard Biener <rguenther@suse.de>
4240 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
4242 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
4243 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
4244 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
4245 (isl_val_int_from_wi): New function.
4246 (extract_affine_gmp): Rename to ...
4247 (extract_affine_wi): ... this, take a widest_int.
4248 (extract_affine_int): Just wrap extract_affine_wi.
4249 (add_param_constraints): Use isl_val_int_from_wi.
4250 (add_loop_constraints): Likewise, and extract_affine_wi.
4252 2017-02-15 Jeff Law <law@redhat.com>
4255 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
4256 ira_init_register_move_cost_if_necessary.
4258 2017-02-15 Martin Sebor <msebor@redhat.com>
4261 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
4262 removed in a prior commit.
4264 2017-02-15 Bin Cheng <bin.cheng@arm.com>
4266 PR tree-optimization/79347
4267 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
4268 counters during peeling.
4270 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
4272 * Makefile.in (site.exp): Remove "set ISLVER".
4274 2017-02-15 Jakub Jelinek <jakub@redhat.com>
4277 * real.c (real_from_integer): Call real_convert even for decimal.
4279 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
4282 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
4284 2017-02-14 Andrew Pinski <apinski@cavium.com>
4286 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
4287 cores and change the partno/implementer to be correct.
4288 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
4289 the 'B" as the implementer.
4290 * config/aarch64/aarch64-tune.md: Regenerate.
4292 2017-02-14 Carl Love <cel@us.ibm.com>
4294 * config/rs6000/rs6000.c: Add case statement entry to make the
4295 xvcvuxdsp built-in argument unsigned.
4296 * config/rs6000/vsx.md: Fix the source and return operand types so they
4297 match the instruction definitions from the ISA document. Fix typo
4298 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
4301 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
4304 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
4305 member early_clobber_alts.
4306 * lra-lives.c (reg_early_clobber_p): New.
4307 (process_bb_lives): Use it.
4308 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
4309 (debug_operand_data): Initialize early_clobber_alts.
4310 (setup_operand_alternative): Set up early_clobber_alts.
4311 (collect_non_operand_hard_regs): Ditto. Pass early clobber
4312 alternatives to new_insn_reg.
4313 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
4315 (lra_update_insn_regno_info): Pass the new arg.
4317 2017-02-14 Jakub Jelinek <jakub@redhat.com>
4320 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
4321 (new_oacc_loop_raw): Don't clear already cleared fields.
4324 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
4325 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
4326 _mm512_prefetch_i64gather_ps): New inline functions and macros.
4328 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
4331 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
4333 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
4336 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
4337 the extra instruction to the right place to store 128-bit constant
4340 2017-02-14 Martin Sebor <msebor@redhat.com>
4343 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
4344 warning for strings of unknown length.
4346 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
4348 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
4350 2017-02-14 Jeff Law <law@redhat.com>
4353 * ira-costs.c (scan_one_insn): Initialize register move costs
4354 for pseudos seen in USE/CLOBBER insns.
4356 PR tree-optimization/79095
4357 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
4358 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
4359 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
4360 if the operands are known to be not equal, then the resulting range
4362 (intersect_ranges): If the new range is ~[0,0] and the old range is
4363 wide, then prefer ~[0,0].
4364 * tree-vrp.c (overflow_comparison_p_1): New function.
4365 (overflow_comparison_p): New function.
4366 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
4367 if NAME is used in an overflow test.
4368 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
4369 overflow check that can be expressed as an equality test, then adjust
4370 ops to be that equality test.
4372 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4374 * config/s390/s390-builtin-types.def: Remove flags argument.
4375 * config/s390/s390.c (s390_init_builtins): Likewise.
4377 2017-02-14 Martin Liska <mliska@suse.cz>
4379 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
4380 vector. Fix trailing white spaces.
4382 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
4384 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
4387 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4389 PR rtl-optimization/68664
4390 * config/arm/arm.c (arm_sched_can_speculate_insn):
4391 New function. Declare prototype.
4392 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
4394 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4396 PR rtl-optimization/68664
4397 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
4399 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
4401 2017-02-14 Amit Pawar <amit.pawar@amd.com>
4403 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
4404 max skip bytes for function, loop and jump.
4406 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4408 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
4409 ABS_EXPR for gimple dump.
4411 2017-02-14 Jakub Jelinek <jakub@redhat.com>
4414 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
4416 PR tree-optimization/79408
4417 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
4418 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
4419 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
4420 also if rhs1 is INTEGER_CST.
4422 2017-02-14 Richard Biener <rguenther@suse.de>
4425 * tree-into-ssa.c (insert_phi_nodes): When the function can
4426 have abnormal edges rewrite SSA names with broken use-def
4427 dominance out of SSA and register them for PHI insertion.
4429 2017-02-13 Martin Sebor <msebor@redhat.com>
4432 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
4433 clearing info.nowrite flag when snprintf size argument is a range.
4435 2017-02-13 Jakub Jelinek <jakub@redhat.com>
4437 * cprop.c (cprop_jump): Add missing space in string literal.
4438 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
4439 (get_constraint_for_component_ref): Likewise.
4440 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
4441 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
4442 * lra-constraints.c (process_alt_operands): Likewise.
4443 * ipa-inline.c (inline_small_functions): Likewise.
4444 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
4445 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
4446 * trans-mem.c (diagnose_tm_1_op): Likewise.
4447 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
4448 (grid_parallel_clauses_gridifiable): Likewise.
4450 * config/nvptx/mkoffload.c (process): Add space in between
4453 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
4454 "MOD4_SSE_REGS" and "ALL_REGS".
4456 * spellcheck.c (test_data): Add , in between "foo" and "food".
4458 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4461 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
4462 boundary crossing check and subsequent code generation agree.
4464 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4466 * config/aarch64/aarch64.c (has_memory_op): Delete.
4467 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
4470 2017-02-13 Jakub Jelinek <jakub@redhat.com>
4472 PR rtl-optimization/79388
4473 PR rtl-optimization/79450
4474 * combine.c (distribute_notes): When removing TEM_INSN for which
4475 corresponding dest has last value recorded, invalidate that last
4478 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4480 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
4481 of explicit '@'. Add missing assembly comment marker on branch costs
4484 2017-02-13 Nathan Sidwell <nathan@acm.org>
4486 * gengtype-lex.l (<in_struct>): Add '/'.
4488 2017-02-13 Martin Liska <mliska@suse.cz>
4491 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
4493 2017-02-13 Richard Biener <rguenther@suse.de>
4495 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
4497 * configure: Re-generate.
4498 * config.in: Likewise.
4499 * graphite-dependences.c: Simplify as if
4500 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
4501 * graphite-isl-ast-to-gimple.c: Likewise.
4502 * graphite-optimize-isl.c: Likewise.
4503 * graphite-poly.c: Likewise.
4504 * graphite-sese-to-poly.c: Likewise.
4505 * graphite.h: Likewise.
4506 * toplev.c: Include isl/version.h and use isl_version () for
4507 printing the ISL version.
4508 * doc/install.texi: Update ISL requirement.
4510 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
4512 * doc/standards.texi (Standards): Update reference to
4515 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
4517 * doc/extend.texi (Named Address Spaces): sourceware.org now
4519 * doc/install.texi (Binaries): Ditto.
4522 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
4524 * doc/cpp.texi: Replace "stringify"/"stringification" with C
4525 standard terminology "stringize"/"stringizing" throughout.
4526 * doc/cppinternals.texi: Likewise.
4528 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
4530 * doc/extend.texi: Fix some spelling mistakes and typos.
4531 * doc/invoke.texi: Likewise.
4533 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
4536 * params.def (inline-min-speedup) Change from 10 to 8.
4538 2017-02-11 Jakub Jelinek <jakub@redhat.com>
4540 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
4543 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
4546 * ipa-inline-analysis.c (get_minimal_bb): New function.
4547 (record_modified): Use it.
4548 (remap_edge_change_prob): Handle also ancestor functions.
4550 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
4552 * doc/contrib.texi (Contributors): Remove broken link into
4553 the Mauve CVS repository.
4555 2017-02-11 Jakub Jelinek <jakub@redhat.com>
4558 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
4559 result computation whenever lhs doesn't have vector mode, not
4560 just when it has BLKmode.
4562 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
4564 * doc/makefile.texi (profiledbootstrap): Refer to the
4565 installation instructions only in textual form.
4567 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4570 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
4572 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
4574 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
4575 (Specific): Update mingw-w64 reference.
4577 (Specific): Remove broken link to Renesas RX processor.
4579 2017-02-10 Richard Biener <rguenther@suse.de>
4581 * toplev.c (process_options): Do not mention obsolete graphite
4582 options when printing sorry message about missing graphite support.
4583 Mention -floop-nest-optimize.
4585 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
4587 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
4588 (vtst_p16): Likewise.
4589 (vtstq_p8): Likewise.
4590 (vtstq_p16): Likewise.
4592 (vtstq_p64): Likewise.
4593 * config/arm/arm_neon.h (vgetq_lane_p64): New.
4594 (vset_lane_p64): New.
4595 (vsetq_lane_p64): New.
4597 2017-02-10 Jakub Jelinek <jakub@redhat.com>
4599 PR tree-optimization/79411
4600 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
4601 stmt operands are SSA_NAMEs used in abnormal phis.
4602 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
4605 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
4608 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
4611 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
4613 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
4615 2017-02-09 Jakub Jelinek <jakub@redhat.com>
4617 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
4621 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
4622 not arbitrary TREE_CONSTANT.
4625 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
4626 "omp declare target link" attribute unless is_global_var.
4627 * omp-offload.c (find_link_var_op): Likewise.
4629 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
4630 Chung-Lin Tang <cltang@codesourcery.com>
4632 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
4634 (gimplify_adjust_omp_clauses): Don't delete TILE.
4635 (gimplify_omp_for): Deal with TILE.
4636 * internal-fn.c (expand_GOACC_TILE): New function.
4637 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
4639 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
4640 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
4642 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
4643 avoid DIV for outermost collapse var.
4644 (expand_oacc_for): Insert tile element loop as needed. Adjust.
4645 Remove out of date comments, fix whitespace.
4646 * omp-general.c (omp_extract_for_data): Deal with tiling.
4647 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
4648 adjust OLF_DIM_BASE value.
4649 (struct omp_for_data): Add tiling field.
4650 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
4651 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
4652 for auto loops. Remove default auto determining, moved to
4653 oacc_loop_fixed_partitions.
4654 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
4655 stmts, add e_mask field.
4656 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
4657 (oacc_thread_numbers): Use oacc_dim_call.
4658 (oacc_xform_tile): New.
4659 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
4660 (finish_oacc_loop): Adjust for ifns vector.
4661 (oacc_loop_discover_walk): Append loop abstraction sites to list,
4662 add case for GOACC_TILE fns.
4663 (oacc_loop_xform_loop): Delete.
4664 (oacc_loop_process): Iterate over call list directly, and add
4665 handling for GOACC_TILE fns.
4666 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
4668 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
4669 vector partitioning to outer loops. Assign 2 partitions to loops
4670 when available. Add TILE handling.
4671 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
4672 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
4673 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
4674 * tree.c (omp_clause_num_ops): Adjust TILE ops.
4675 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
4677 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
4679 * configure.ac (ACX_BUGURL): Update.
4680 * configure: Regenerate.
4682 2017-02-09 Richard Biener <rguenther@suse.de>
4684 PR tree-optimization/69823
4685 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
4686 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
4688 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
4690 * config/arc/arc-c.def: Add __NPS400__ definition.
4691 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
4692 (TARGET_NPS400): Define.
4694 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
4696 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
4698 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
4700 (arc_cpu_types): Fill the arch_info field with a pointer into the
4701 arc_arch_types table.
4702 (arc_selected_cpu): Declare.
4703 * config/arc/arc.c (arc_selected_cpu): Make global.
4704 (arc_selected_arch): Delete.
4705 (arc_base_cpu): Delete.
4706 (arc_override_options): Remove references to deleted variables,
4707 update access to arch information.
4708 (ARC_OPT): Update access to arch information.
4709 (ARC_OPTX): Likewise.
4710 * config/arc/arc.h (arc_base_cpu): Remove declaration.
4711 (TARGET_ARC600): Update access to arch information.
4712 (TARGET_ARC601): Likewise.
4713 (TARGET_ARC700): Likewise.
4714 (TARGET_EM): Likewise.
4715 (TARGET_HS): Likewise.
4716 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
4719 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
4722 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
4723 condition/operands for integer GE/LE/GEU/LEU operations.
4725 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
4727 PR translation/79397
4728 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
4731 2017-02-08 Martin Jambor <mjambor@suse.cz>
4734 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
4735 whether allocation happened.
4736 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
4737 nothing was allocated.
4739 2017-02-08 Jakub Jelinek <jakub@redhat.com>
4741 PR tree-optimization/79408
4742 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
4743 constant, but SSA_NAME with a known integer range, use the minimum
4744 of that range instead of op1 to determine if modulo can be replaced
4745 with its first operand.
4747 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4749 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
4751 2017-02-08 Richard Biener <rguenther@suse.de>
4753 PR tree-optimization/71824
4754 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
4755 Check all loops contained in the merged region.
4757 2017-02-07 Andrew Pinski <apinski@cavium.com>
4759 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
4761 2017-02-07 Andrew Pinski <apinski@cavium.com>
4763 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
4764 (thunderxt88): Likewise.
4765 (thunderxt81): Disable LSE and change v8.1 to v8.
4766 (thunderxt83): Likewise.
4768 2017-02-07 Jakub Jelinek <jakub@redhat.com>
4769 Richard Biener <rguenther@suse.de>
4772 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
4773 type from int to size_t.
4774 * ira-costs.c (struct_costs_size): Change type from int to size_t.
4776 2017-02-07 Jakub Jelinek <jakub@redhat.com>
4778 PR rtl-optimization/79386
4779 * cprop.c (bypass_conditional_jumps): Initialize
4780 bypass_last_basic_block already before splitting bbs after
4781 unconditional traps...
4782 (bypass_conditional_jumps): ... rather than here.
4785 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
4786 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
4787 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
4788 fix -masm=intel patterns.
4790 2017-02-07 Richard Biener <rguenther@suse.de>
4792 PR tree-optimization/79256
4794 * builtins.c (get_object_alignment_2): Use min_align_of_type
4795 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
4796 and ADJUST_FIELD_ALIGN.
4798 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
4800 * doc/tm.texi: Regenerate.
4801 * stor-layout.c (layout_decl): Adjust.
4802 (update_alignment_for_field): Likewise.
4803 (place_field): Likewise.
4804 (min_align_of_type): Likewise.
4805 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
4806 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
4807 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
4808 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
4809 * config/frv/frv.c (frv_adjust_field_align): Likewise.
4810 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
4811 * config/i386/i386.c (x86_field_alignment): Likewise.
4812 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
4813 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
4814 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
4815 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
4816 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
4817 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
4821 2017-01-30 Richard Biener <rguenther@suse.de>
4823 PR tree-optimization/79256
4824 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
4825 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
4828 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
4830 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
4831 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
4832 builtins to SImode and emit a zero-extend, if necessary.
4834 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
4836 * docs/invoke.texi (RISC-V Options): Alphabetize.
4838 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
4840 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
4843 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
4845 * config/riscv/riscv.c: New file.
4846 * gcc/common/config/riscv/riscv-common.c: Likewise.
4847 * config.gcc: Likewise.
4848 * config/riscv/constraints.md: Likewise.
4849 * config/riscv/elf.h: Likewise.
4850 * config/riscv/generic.md: Likewise.
4851 * config/riscv/linux.h: Likewise.
4852 * config/riscv/multilib-generator: Likewise.
4853 * config/riscv/peephole.md: Likewise.
4854 * config/riscv/pic.md: Likewise.
4855 * config/riscv/predicates.md: Likewise.
4856 * config/riscv/riscv-builtins.c: Likewise.
4857 * config/riscv/riscv-c.c: Likewise.
4858 * config/riscv/riscv-ftypes.def: Likewise.
4859 * config/riscv/riscv-modes.def: Likewise.
4860 * config/riscv/riscv-opts.h: Likewise.
4861 * config/riscv/riscv-protos.h: Likewise.
4862 * config/riscv/riscv.h: Likewise.
4863 * config/riscv/riscv.md: Likewise.
4864 * config/riscv/riscv.opt: Likewise.
4865 * config/riscv/sync.md: Likewise.
4866 * config/riscv/t-elf-multilib: Likewise.
4867 * config/riscv/t-linux: Likewise.
4868 * config/riscv/t-linux-multilib: Likewise.
4869 * config/riscv/t-riscv: Likewise.
4870 * configure.ac: Likewise.
4871 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
4872 Waterman as RISC-V maintainers.
4873 * doc/install.texi: Add RISC-V entries.
4874 * doc/invoke.texi: Add RISC-V options section.
4875 * doc/md.texi: Add RISC-V constraints section.
4876 * configure: Regenerated.
4878 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
4881 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
4882 false values to be constant vectors with all 0 or all 1 bits set.
4883 (vcondu<mode><mode>): Likewise.
4884 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
4886 (fpmask_comparison_operator): Update comment.
4887 (vecint_comparison_operator): New predicate.
4888 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
4889 vector conditionals when the true and false values are constant
4890 vectors with all 0 bits or all 1 bits set.
4892 2017-02-06 Martin Sebor <msebor@redhat.com>
4894 PR tree-optimization/79376
4895 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
4897 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
4899 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
4900 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
4901 to simplify split condition.
4903 2017-02-06 Jakub Jelinek <jakub@redhat.com>
4905 * omp-expand.c (oxpand_omp_atomic_fetch_op,
4906 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
4909 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
4911 PR rtl-optimization/68664
4912 * target.def (can_speculate_insn): New hook.
4913 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
4914 * doc/tm.texi: Regenerate.
4915 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
4916 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
4917 (rs6000_sched_can_speculate_insn): New function.
4919 2017-02-06 Jakub Jelinek <jakub@redhat.com>
4921 PR tree-optimization/79284
4922 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
4923 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
4924 vectorizable_mask_load_store, vectorizable_operation,
4925 vect_is_simple_cond, get_same_sized_vectype): Use it instead
4926 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
4927 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
4928 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
4929 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
4930 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
4931 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
4932 is_gimple_assign (stmt). Replace another such test with
4933 is_gimple_assign (stmt).
4935 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
4938 * config/avr/avr.c (rtl-iter.h): Include it.
4939 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
4940 (avr_legitimate_combined_insn): ...and implementation.
4942 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
4944 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
4945 * config/s390/s390.c (s390_const_operand_ok)
4946 (s390_canonicalize_comparison, s390_extract_part)
4947 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
4948 (s390_contiguous_bitmask_p, s390_rtx_costs)
4949 (legitimize_pic_address): Likewise.
4950 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
4951 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
4952 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
4953 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
4954 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
4956 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
4958 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
4959 REGNO($0) == REGNO($1).
4961 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4963 * config/s390/linux.h(SIZE_TYPE): Add comment.
4965 2017-02-06 Julian Brown <julian@codesourcery.com>
4966 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
4967 Virendra Pathak <virendra.pathak@broadcom.com>
4969 * config/aarch64/aarch64-cores.def: Change the scheduler
4971 * config/aarch64/aarch64.md: Include thunderx2t99.md.
4972 * config/aarch64/thunderx2t99.md: New file.
4974 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
4976 * doc/standards.texi (Go Language): Update link to language
4979 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
4981 * tree-eh.c (lower_resx): Sanitize profile.
4982 (cleanup_empty_eh_move_lp): Likewise.
4984 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
4987 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
4989 * cfgloopmanip.h (loop_version): Update prototype.
4990 * modulo-sched.c (sms_schedule): Update call of loop_version.
4991 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
4992 * tree-parloops.c (gen_parallel_loop): Likewise.
4993 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
4994 * tree-ssa-loop-split.c (split_loop): Likewise.
4995 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
4996 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
4998 2017-02-05 Martin Liska <mliska@suse.cz>
5001 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
5003 (print_operand_address): Initialize a struct to zero.
5005 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
5007 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
5008 garbage collector only in textual form.
5010 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
5012 * doc/extend.texi (x86 specific memory model extensions for
5013 transactional memory): Simplify a phrase.
5015 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
5018 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
5019 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
5020 (atomic_storedi_1): Likewise.
5022 2017-02-04 Jakub Jelinek <jakub@redhat.com>
5024 PR tree-optimization/79338
5025 * tree-parloops.c (gather_scalar_reductions): Don't call
5026 vect_analyze_loop_form for loop->inner before destroying loop's
5029 2017-02-03 Martin Sebor <msebor@redhat.com>
5031 PR tree-optimization/79327
5032 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
5033 when precision has resulted in leading zeros.
5034 (format_integer): Adjust the likely counter to assume an unknown
5035 argument that may be zero is non-zero.
5037 2017-02-03 Jason Merrill <jason@redhat.com>
5040 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
5041 avoid copying non-taken branch.
5043 2017-02-03 Jakub Jelinek <jakub@redhat.com>
5045 PR tree-optimization/79340
5046 * tree-vect-loop.c (vectorizable_reduction): Release
5047 vec_defs elements after safe_splicing them into other vectors.
5050 PR tree-optimization/79327
5051 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
5052 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
5054 (format_integer): Use wide_int_to_tree instead of build_int_cst
5055 + to_?hwi. If argmin is NULL, just set argmin and argmax to
5056 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
5057 of shortest and longest sequence.
5059 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
5061 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
5062 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
5064 2017-02-03 Walter Lee <walt@tilera.com>
5067 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
5068 after initial stackframe link reg save.
5069 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
5071 2017-02-03 Jakub Jelinek <jakub@redhat.com>
5074 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
5075 wu for stxssp alternative.
5077 2017-02-03 Martin Sebor <msebor@redhat.com>
5079 PR tree-optimization/79352
5080 * gimple-fold.c (get_range_strlen): Add argument.
5081 (get_range_strlen): Change return type to bool.
5082 (get_maxval_strlen): Pass in a dummy argument.
5083 * gimple-fold.h (get_range_strlen): Change return type to bool.
5084 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
5085 * tree.h (array_at_struct_end_p): Add argument.
5086 * tree.c (array_at_struct_end_p): Handle it.
5088 2017-02-03 Martin Liska <mliska@suse.cz>
5091 * multiple_target.c (create_dispatcher_calls): Redirect edge
5092 from a caller of a dispatcher.
5093 (expand_target_clones): Make the clones local.
5094 (ipa_target_clone): Do both target clones and resolvers.
5095 (ipa_dispatcher_calls): Remove the pass.
5096 (pass_dispatcher_calls::gate): Likewise.
5097 (make_pass_dispatcher_calls): Likewise.
5098 * passes.def (pass_target_clone): Put as very first IPA early
5101 2017-02-03 Martin Liska <mliska@suse.cz>
5103 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
5104 in case of a function with ifunc attribute.
5106 2017-02-03 Martin Liska <mliska@suse.cz>
5108 * cgraph.c (cgraph_node::dump): Dump function version info.
5109 * symtab.c (symtab_node::dump_base): Add missing new line.
5111 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
5113 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
5114 (ifcombine_ifandif): Use it.
5116 2017-02-03 Martin Liska <mliska@suse.cz>
5118 * doc/invoke.texi: Document default value for
5119 use-after-scope-direct-emission-threshold.
5121 2017-02-03 Martin Liska <mliska@suse.cz>
5123 PR tree-optimization/79339
5124 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
5125 (format_floating): Likewise.
5127 2017-02-03 Martin Liska <mliska@suse.cz>
5130 * ipa-prop.c (ipa_node_params_t::insert): Remove current
5132 (ipa_node_params_t::remove): Likewise.
5133 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
5134 initialization from removed ipa_node_params_t::insert.
5135 (ipa_node_params::~ipa_node_params): Move from removed
5136 ipa_node_params_t::release.
5137 * symbol-summary.h (symbol_summary::m_released): New member.
5138 Do not release a summary twice. Do not allow to call finalizer
5139 for types of a summary that live in GGC memory.
5141 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5143 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
5146 2017-02-02 Martin Sebor <msebor@redhat.com>
5149 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
5150 (format_string): Tighten up the range of output for non-constant
5151 strings and correct the expected range for wide non-constant strings.
5153 2017-02-02 Martin Sebor <msebor@redhat.com>
5155 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
5158 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
5160 (-fdump-tree-@var): Add to index and document how to come up
5161 with pass-specific option and dump file names.
5162 (-fdump-passes): Clarify where to look for output.
5164 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
5167 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
5168 statistics of the analyzed path; allow threading for speed when
5169 any of BBs along the path are optimized for speed.
5171 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
5174 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
5175 settings of the virtual registers.
5178 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
5180 * explow.c (get_dynamic_stack_size): Take known alignment of stack
5181 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
5184 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5186 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
5187 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
5189 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5191 * config/s390/s390.md: Add missing comments with the expanded
5193 * config/s390/vector.md: Likewise.
5194 * config/s390/vx-builtins.md: Likewise.
5196 2017-02-02 Jakub Jelinek <jakub@redhat.com>
5199 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
5200 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
5201 conditions on a single line.
5203 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5205 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
5206 __S390_VX__ to __VX__.
5208 2017-02-01 Andrew Pinski <apinski@cavium.com>
5210 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
5211 stmt_info to record_stmt_cost.
5212 (vect_get_known_peeling_cost): Pass stmt_info if known to
5214 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
5215 cpu_vector_cost field into
5216 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
5217 field into vec_int_stmt_cost and vec_fp_stmt_cost.
5218 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
5219 splitting of scalar_stmt_cost and vec_stmt_cost.
5220 (thunderx_vector_cost): Likewise.
5221 (cortexa57_vector_cost): LIkewise.
5222 (exynosm1_vector_cost): Likewise.
5223 (xgene1_vector_cost): Likewise.
5224 (thunderx2t99_vector_cost): Improve after the splitting of the two
5226 (aarch64_builtin_vectorization_cost): Update for the splitting of
5227 scalar_stmt_cost and vec_stmt_cost.
5229 2017-02-01 Torvald Riegel <triegel@redhat.com>
5230 Richard Henderson <rth@redhat.com>
5232 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
5233 conditional on existance of a fast atomic load.
5234 * optabs-query.c (can_atomic_load_p): New function.
5235 * optabs-query.h (can_atomic_load_p): Declare it.
5236 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
5237 no fast atomic load is available for the particular size of access.
5238 (expand_atomic_compare_and_swap): Likewise.
5239 (expand_atomic_load): Likewise.
5240 (expand_atomic_store): Likewise.
5241 (expand_atomic_fetch_op): Likewise.
5242 * testsuite/lib/target-supports.exp
5243 (check_effective_target_sync_int_128): Remove x86 because it provides
5244 no fast atomic load.
5245 (check_effective_target_sync_int_128_runtime): Likewise.
5247 2017-02-01 Richard Biener <rguenther@suse.de>
5249 * graphite.c: Include tree-vectorizer.h for find_loop_location.
5250 (graphite_transform_loops): Provide opt-info for optimized nests.
5251 * tree-parloop.c (parallelize_loops): Provide opt-info for
5254 2017-02-01 Richard Biener <rguenther@suse.de>
5257 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
5260 2017-02-01 Richard Biener <rguenther@suse.de>
5262 PR tree-optimization/71824
5263 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
5264 Verify the loops are valid in the merged SESE region.
5265 (scop_detection::can_represent_loop_1): Check analyzing the
5266 evolution of the number of iterations in the region succeeds.
5268 2017-01-31 Ian Lance Taylor <iant@golang.org>
5270 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
5271 REG_ARGS_SIZE note to 32-bit push insns and call insn.
5273 2017-01-31 David Malcolm <dmalcolm@redhat.com>
5275 PR preprocessor/79210
5276 * input.c (get_substring_ranges_for_loc): Replace line_width
5277 assertion with error-handling.
5279 2017-01-31 Richard Biener <rguenther@suse.de>
5281 PR tree-optimization/77318
5282 * graphite-sese-to-poly.c (extract_affine): Fix assert.
5283 (create_pw_aff_from_tree): Take loop parameter.
5284 (add_condition_to_pbb): Pass loop of the condition to
5285 create_pw_aff_from_tree.
5287 2017-01-31 Jakub Jelinek <jakub@redhat.com>
5289 * config/s390/s390.c (s390_asan_shadow_offset): New function.
5290 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
5292 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
5296 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
5298 (convert_int_to_float128): Likewise.
5299 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
5300 (convert_int_to_float128): Likewise.
5301 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
5302 (UNSPEC_IEEE128_CONVERT): Likewise.
5303 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
5304 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
5305 Use local variables for IBM extended format.
5306 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
5307 (fix_trunc<mode>si2_fprs): Likewise.
5308 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
5309 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
5310 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
5311 to know that we can now have integers of all sizes in vector
5313 (fix<uns>_<mode>di2_hw): Likewise.
5314 (float<uns>_<mode>si2_hw): Likewise.
5315 (fix_<mode>si2_hw): Likewise.
5316 (fixuns_<mode>si2_hw): Likewise.
5317 (float<uns>_<mode>di2_hw): Likewise.
5318 (float_<mode>di2_hw): Likewise.
5319 (float_<mode>si2_hw): Likewise.
5320 (floatuns_<mode>di2_hw): Likewise.
5321 (floatuns_<mode>si2_hw): Likewise.
5322 (xscvqp<su>wz_<mode>): Delete, no longer used.
5323 (xscvqp<su>dz_<mode>): Likewise.
5324 (xscv<su>dqp_<mode>): Likewise.
5325 (ieee128_mfvsrd_64bit): Likewise.
5326 (ieee128_mfvsrd_32bit): Likewise.
5327 (ieee128_mfvsrwz): Likewise.
5328 (ieee128_mtvsrw): Likewise.
5329 (ieee128_mtvsrd_64bit): Likewise.
5330 (ieee128_mtvsrd_32bit): Likewise.
5332 2017-01-31 Martin Liska <mliska@suse.cz>
5335 * ipa-prop.c (ipa_free_all_node_params): Call release method
5336 instead of ~sumbol_summary to not to trigger double times
5339 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
5341 PR tree-optimization/71691
5342 * bitmap.h (class auto_bitmap): New.
5343 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
5344 is_maybe_undefined instead of ssa_undefined_value_p.
5346 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5348 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
5349 __S390_ARCH_LEVEL__ to __ARCH__.
5351 2017-01-31 Jakub Jelinek <jakub@redhat.com>
5353 PR tree-optimization/79267
5354 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
5355 if should_remove_lhs_p is true.
5357 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
5360 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
5361 (add_alignment_attribute): New.
5362 (base_type_die): Add alignment attribute.
5363 (subrange_type_die): Likewise.
5364 (modified_type_die): Likewise.
5365 (gen_array_type_die): Likewise.
5366 (gen_descr_array_type_die: Likewise.
5367 (gen_enumeration_type_die): Likewise.
5368 (gen_subprogram_die): Likewise.
5369 (gen_variable_die): Likewise.
5370 (gen_field_die): Likewise.
5371 (gen_ptr_to_mbr_type_die): Likewise.
5372 (gen_struct_or_union_type_die): Likewise.
5373 (gen_subroutine_type_die): Likewise.
5374 (gen_typedef_die): Likewise.
5375 (base_type_cmp): Compare alignment attribute.
5377 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
5380 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
5381 (setb_unsigned) New pattern for setb with CCUNS.
5382 * config/rs6000/rs6000.c (expand_block_compare): Use a different
5383 subfc./subfe sequence to avoid overflow problems. Generate a
5384 shorter sequence with cmpld/setb for power9.
5385 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
5386 for generating subfc. instruction.
5387 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
5388 now uses this instruction.
5390 2017-01-30 Ian Lance Taylor <iant@google.com>
5393 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
5394 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
5396 2017-01-30 Martin Sebor <msebor@redhat.com>
5398 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
5399 Move constant to the right of a relational operator.
5400 (get_mpfr_format_length, format_character, format_string): Ditto.
5401 (should_warn_p, maybe_warn): Same.
5403 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
5405 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
5408 * asan.c (get_translation_unit_decl): Remove function.
5409 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
5411 2017-01-30 Martin Liska <mliska@suse.cz>
5413 PR gcov-profile/79259
5414 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
5417 2017-01-30 Martin Liska <mliska@suse.cz>
5420 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
5421 Initialize variables with NULL value.
5423 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
5426 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
5428 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
5430 2017-01-30 Richard Biener <rguenther@suse.de>
5432 PR tree-optimization/79276
5433 * tree-vrp.c (process_assert_insertions): Properly adjust common
5434 when removing a duplicate.
5436 * gcc.dg/torture/pr79276.c: New testcase.
5438 2017-01-30 Richard Biener <rguenther@suse.de>
5440 PR tree-optimization/79256
5441 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
5442 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
5444 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
5446 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
5449 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
5450 ("*r<noxa>sbg_<mode>_sll_bitmask")
5451 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
5452 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
5453 Use contiguous_bitmask_nowrap_operand.
5455 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5458 * config/rs6000/altivec.h (vec_xl): Revise #define.
5459 (vec_xst): Likewise.
5461 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
5463 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
5465 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
5467 PR rtl-optimization/79194
5468 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
5469 traps before call to bypass_conditional_jumps.
5471 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
5473 PR tree-optimization/71374
5474 * lra-constraints.c (check_conflict_input_operands): New.
5475 (match_reload): Use it.
5477 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
5480 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
5481 account to calculate conflict_set.
5483 2017-01-27 Bin Cheng <bin.cheng@arm.com>
5485 PR rtl-optimization/78559
5486 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
5487 other_insn in combine.
5489 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
5491 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
5492 uint16_type_node for BT_UINT16.
5494 2017-01-27 David Malcolm <dmalcolm@redhat.com>
5496 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
5497 "RTL Tests" to menu.
5498 (GIMPLE Tests): New node.
5499 (RTL Tests): New node.
5501 2017-01-27 Richard Biener <rguenther@suse.de>
5503 PR tree-optimization/79245
5504 * tree-loop-distribution.c (distribute_loop): Apply cost
5505 modeling also to detected patterns.
5507 2017-01-27 Richard Biener <rguenther@suse.de>
5509 PR tree-optimization/71433
5510 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
5511 (compare_assert_loc): New function.
5512 (process_assert_insertions): Sort and optimize assert locations
5513 to remove duplicates and push down identical assertions on
5514 edges to their destination block.
5516 2017-01-27 Richard Biener <rguenther@suse.de>
5518 PR tree-optimization/79244
5519 * tree-vrp.c (remove_range_assertions): Forcefully propagate
5520 out SSA names even if abnormal.
5522 2017-01-27 Jakub Jelinek <jakub@redhat.com>
5524 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
5525 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
5526 instead of MPFR_RNDN.
5528 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
5531 * arm.c (arm_option_override): Don't call build_target_option_node
5532 until after doing all option overrides.
5533 (arm_valid_target_attribute_tree): Likewise.
5535 2017-01-27 Martin Liska <mliska@suse.cz>
5537 * doc/invoke.texi (-fprofile-arcs): Document profiling support
5538 for {cd}tors and C++ {cd}tors.
5540 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
5542 * config/s390/s390.md ("*setmem_long_and")
5543 ("*setmem_long_and_31z"): Use zero_extend instead of and.
5545 2017-01-26 Martin Sebor <msebor@redhat.com>
5547 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
5550 2017-01-26 Martin Sebor <msebor@redhat.com>
5552 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
5553 HAVE_DFmode before using XFmode or DFmode.
5554 (parse_directive): Avoid using the z length modifier to avoid
5555 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
5558 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
5559 to accept adjustment as an array.
5560 (get_int_range): New function.
5561 (struct directive): Make width and prec arrays.
5562 (directive::set_width, directive::set_precision): Call get_int_range.
5563 (format_integer, format_floating): Handle width and precision ranges.
5564 (format_string, parse_directive): Same.
5566 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5569 * dwarf2out.c (generate_skeleton_bottom_up): For children with
5570 comdat_type_p set, just clone them, but keep the children in the
5574 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
5575 which have direct callers with -fvar-tracking-assignments enabled
5577 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
5578 inside of type units.
5580 2017-01-26 Martin Sebor <msebor@redhat.com>
5583 * gimple-ssa-sprintf.c (struct result_range): Add likely and
5585 (struct format_result): Replace number_chars, number_chars_min,
5586 and number_chars_max with a single member of struct result_range.
5588 (format_result::operator+=): Adjust.
5589 (struct fmtresult): Remove bounded. Handle likely and unlikely
5591 (fmtresult::adjust_for_width_or_precision): New function.
5592 (fmtresult:type_max_digits): New function.
5593 (bytes_remaining): Handle likely and unlikely counters.
5594 (min_bytes_remaining): Remove.
5595 (format_percent): Simplify.
5596 (format_integer, format_floating): Set likely and unlikely counters.
5597 (get_string_length, format_character, format_string): Same.
5598 (format_plain, should_warn_p): New function.
5599 (maybe_warn): Call should_warn_p. Update diagnostic messages
5600 and handle those for all directives, including plain strings.
5601 (format_directive): Handle likely and unlikely counters.
5602 Remove unnecessary quoting from diagnostics. Add an informational
5604 (add_bytes): Remove.
5605 (pass_sprintf_length::compute_format_length): Simplify.
5606 (try_substitute_return_value): Handle likely and unlikely counters.
5608 2017-01-26 Carl Love <cel@us.ibm.com>
5610 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
5611 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
5613 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
5616 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
5617 endianess for subregs into account.
5618 * lra-constraints.c (lra_constraints): Do risky transformations
5619 always on the first iteration.
5620 * lra-lives.c (check_pseudos_live_through_calls): Add arg
5621 last_call_used_reg_set.
5622 (process_bb_lives): Define and use last_call_used_reg_set.
5623 * lra.c (lra): Always continue after lra_constraints on the first
5626 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
5628 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
5630 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
5632 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5634 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
5635 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
5636 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
5637 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
5638 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
5639 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
5640 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
5641 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
5642 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
5644 2017-01-26 Marek Polacek <polacek@redhat.com>
5647 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
5648 for the third operand.
5650 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5653 * omp-low.c (struct omp_context): Add simt_stmt field.
5654 (scan_omp_for): Return omp_context *.
5655 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
5656 context to the _simt_ SIMD stmt.
5657 (lower_omp_for): For combined SIMD with sibling _simt_
5658 SIMD, make sure to use the same decls in _looptemp_
5659 clauses as in the sibling.
5661 2017-01-26 David Sherwood <david.sherwood@arm.com>
5664 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
5667 2017-01-26 Jakub Jelinek <jakub@redhat.com>
5670 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
5671 emit fld b; fld a; if possible.
5673 * brig-builtins.def: Update copyright years.
5674 * config/arm/arm_acle_builtins.def: Update copyright years.
5676 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
5679 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
5680 constraint instead of o for the stxsd instruction.
5682 2017-01-25 Carl Love <cel@us.ibm.com>
5684 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
5685 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
5687 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
5689 * doc/invoke.texi (C++ Dialect Options): Fix typo.
5691 2017-01-25 Richard Biener <rguenther@suse.de>
5693 PR tree-optimization/69264
5694 * target.def (vector_alignment_reachable): Improve documentation.
5695 * doc/tm.texi: Regenerate.
5696 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
5698 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
5699 earlier changes with respect to TYPE_USER_ALIGN.
5700 (vector_alignment_reachable_p): Likewise. Improve dumping.
5702 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5705 * config/arm/arm.md (xordi3): Force constant operand into a register
5708 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5710 * doc/invoke.texi (-fstore-merging): Correct default optimization
5711 levels at which it is enabled.
5712 (-O): Move -fstore-merging from list to...
5715 2017-01-25 Richard Biener <rguenther@suse.de>
5718 * omp-expand.c: Include debug.h.
5719 (expand_omp_taskreg): Make sure to generate early debug before
5720 outlining anything from a function.
5721 (expand_omp_target): Likewise.
5722 (grid_expand_target_grid_body): Likewise.
5724 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
5727 * asan.c (get_translation_unit_decl): New function.
5728 (asan_add_global): Extract modules file name from globals
5729 TRANSLATION_UNIT_DECL name.
5731 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
5734 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
5735 for long calls with APCS frame and VFP.
5737 2017-01-24 David Malcolm <dmalcolm@redhat.com>
5739 * cfg.c (original_copy_tables_initialized_p): New function.
5740 * cfg.h (original_copy_tables_initialized_p): New decl.
5741 * cfgrtl.c (relink_block_chain): Guard the call to
5742 free_original_copy_tables with a call to
5743 original_copy_tables_initialized_p.
5744 * cgraph.h (symtab_node::native_rtl_p): New decl.
5745 * cgraphunit.c (symtab_node::native_rtl_p): New function.
5746 (symtab_node::needed_p): Don't assert for early assembly output
5747 for __RTL functions.
5748 (cgraph_node::finalize_function): Set "force_output" for __RTL
5750 (cgraph_node::analyze): Bail out early for __RTL functions.
5751 (analyze_functions): Update assertion to support __RTL functions.
5752 (cgraph_node::expand): Bail out early for __RTL functions.
5753 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
5755 * function.h (struct function): Update comment for field
5757 * gimple-expr.c: Include "tree-pass.h".
5758 (gimple_has_body_p): Return false for __RTL functions.
5759 * Makefile.in (OBJS): Add run-rtl-passes.o.
5760 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
5762 (gcc::pass_manager::get_clean_slate): New accessor.
5763 * passes.c: Include "insn-addr.h".
5764 (should_skip_pass_p): Add logging. Update logic for running
5765 "expand" to be compatible with both __GIMPLE and __RTL. Guard
5766 property-provider override so it is only done for gimple passes.
5768 (skip_pass): New function.
5769 (execute_one_pass): Call skip_pass when skipping passes.
5770 * read-md.c (md_reader::read_char): Support filtering
5771 the input to a subset of line numbers.
5772 (md_reader::md_reader): Initialize fields
5773 m_first_line and m_last_line.
5774 (md_reader::read_file_fragment): New function.
5775 * read-md.h (md_reader::read_file_fragment): New decl.
5776 (md_reader::m_first_line): New field.
5777 (md_reader::m_last_line): New field.
5778 * read-rtl-function.c (function_reader::create_function): Only
5779 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
5780 curr_properties. Set DECL_INITIAL to a dummy block.
5781 (read_rtl_function_body_from_file_range): New function.
5782 * read-rtl-function.h (read_rtl_function_body_from_file_range):
5784 * run-rtl-passes.c: New file.
5785 * run-rtl-passes.h: New file.
5787 2017-01-24 Jeff Law <law@redhat.com>
5789 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
5792 2017-01-24 Bin Cheng <bin.cheng@arm.com>
5794 PR tree-optimization/79159
5795 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
5796 (record_nonwrapping_iv): Improve boundary using above function if no
5797 value range information.
5799 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
5800 Martin Jambor <mjambor@suse.cz>
5802 * brig-builtins.def: New file.
5803 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
5804 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
5805 (DEF_HSAIL_SAT_BUILTIN): Likewise.
5806 (DEF_HSAIL_INTR_BUILTIN): Likewise.
5807 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
5808 * builtin-types.def (BT_INT8): New.
5809 (BT_INT16): Likewise.
5810 (BT_UINT8): Likewise.
5811 (BT_UINT16): Likewise.
5812 (BT_FN_ULONG): Likewise.
5813 (BT_FN_UINT_INT): Likewise.
5814 (BT_FN_UINT_ULONG): Likewise.
5815 (BT_FN_UINT_LONG): Likewise.
5816 (BT_FN_UINT_PTR): Likewise.
5817 (BT_FN_ULONG_PTR): Likewise.
5818 (BT_FN_INT8_FLOAT): Likewise.
5819 (BT_FN_INT16_FLOAT): Likewise.
5820 (BT_FN_UINT32_FLOAT): Likewise.
5821 (BT_FN_UINT16_FLOAT): Likewise.
5822 (BT_FN_UINT8_FLOAT): Likewise.
5823 (BT_FN_UINT64_FLOAT): Likewise.
5824 (BT_FN_UINT16_UINT32): Likewise.
5825 (BT_FN_UINT32_UINT16): Likewise.
5826 (BT_FN_UINT16_UINT16_UINT16): Likewise.
5827 (BT_FN_INT_PTR_INT): Likewise.
5828 (BT_FN_UINT_PTR_UINT): Likewise.
5829 (BT_FN_LONG_PTR_LONG): Likewise.
5830 (BT_FN_ULONG_PTR_ULONG): Likewise.
5831 (BT_FN_VOID_UINT64_UINT64): Likewise.
5832 (BT_FN_UINT8_UINT8_UINT8): Likewise.
5833 (BT_FN_INT8_INT8_INT8): Likewise.
5834 (BT_FN_INT16_INT16_INT16): Likewise.
5835 (BT_FN_INT_INT_INT): Likewise.
5836 (BT_FN_UINT_FLOAT_UINT): Likewise.
5837 (BT_FN_FLOAT_UINT_UINT): Likewise.
5838 (BT_FN_ULONG_UINT_UINT): Likewise.
5839 (BT_FN_ULONG_UINT_PTR): Likewise.
5840 (BT_FN_ULONG_ULONG_ULONG): Likewise.
5841 (BT_FN_UINT_UINT_UINT): Likewise.
5842 (BT_FN_VOID_UINT_PTR): Likewise.
5843 (BT_FN_UINT_UINT_PTR: Likewise.
5844 (BT_FN_UINT32_UINT64_PTR): Likewise.
5845 (BT_FN_INT_INT_UINT_UINT): Likewise.
5846 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
5847 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
5848 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
5849 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
5850 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
5851 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
5852 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
5853 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
5854 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
5855 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
5856 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
5857 * doc/frontends.texi: List BRIG FE.
5858 * doc/install.texi (Testing): Add BRIG tesring requirements.
5859 * doc/invoke.texi (Overall Options): Mention BRIG.
5860 * doc/standards.texi (Standards): Doucment BRIG HSA version.
5862 2017-01-24 Richard Biener <rguenther@suse.de>
5864 PR translation/79208
5865 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
5867 2017-01-24 Martin Jambor <mjambor@suse.cz>
5870 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
5871 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
5874 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
5877 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
5878 casts from signed to unsigned really don't have a range.
5880 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
5882 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
5883 GMP_RNDx for compatiblity.
5885 2017-01-24 Martin Liska <mliska@suse.cz>
5888 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
5889 that would prevent us to call alloca with -1 as argument.
5891 2017-01-24 Jakub Jelinek <jakub@redhat.com>
5893 * dwarf2out.c (output_compilation_unit_header, output_file_names):
5894 Avoid -Wformat-security warning.
5896 2017-01-23 Andrew Pinski <apinski@cavium.com>
5898 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
5901 2017-01-23 Martin Sebor <msebor@redhat.com>
5904 * gimple-ssa-sprintf.c (warn_level): New global.
5905 (format_integer): Use it here and throughout the rest of the file.
5906 Use the same switch to compute sign as base.
5907 (maybe_warn): New function.
5908 (format_directive): Factor out warnings into maybe_warn.
5909 Add debugging output. Use warn_level.
5910 (add_bytes): Use warn_level.
5911 (pass_sprintf_length::compute_format_length): Add debugging output.
5912 (try_substitute_return_value): Same.
5913 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
5916 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
5917 (struct fmtresult, format_integer, format_floating): Adjust.
5918 (fmtresult::fmtresult): Set max correctly in two argument ctor.
5919 (get_string_length, format_string,format_directive): Same.
5920 (pass_sprintf_length::compute_format_length): Same.
5921 (try_substitute_return_value): Simplify slightly.
5924 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
5925 (fmtresult::operator+=): Outlined.
5926 (struct fmtresult): Add ctors.
5927 (struct conversion_spec): Rename...
5928 (struct directive): ...to this. Add and remove data members.
5929 (directive::set_width, directive::set_precision): New functions.
5930 (format_percent): Use fmtresult ctor.
5931 (get_width_and_precision): Remove.
5932 (format_integer): Make naming changes. Avoid computing width and
5934 (format_floating): Same. Adjust indentation.
5935 (format_character, format_none): New functions.
5936 (format_string): Moved character handling to format_character.
5937 (format_directive): Remove arguments, change return type.
5938 (parse_directive): New function.
5939 (pass_sprintf_length::compute_format_length): Move directive
5940 parsing to parse_directive.
5942 2017-01-23 Jakub Jelinek <jakub@redhat.com>
5944 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
5945 (assign_assembler_name_if_needed): ... this.
5946 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
5947 (assign_assembler_name_if_needed): ... this.
5948 (free_lang_data_in_cgraph): Adjust callers.
5949 * cgraphunit.c (cgraph_node::analyze): Likewise.
5950 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
5953 2017-01-23 Richard Biener <rguenther@suse.de>
5955 PR tree-optimization/79088
5956 PR tree-optimization/79188
5957 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
5958 resetting loop bounds after last path deletion. Reset loop
5959 bounds of the target loop, make code match the comments.
5960 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
5961 Make sure loops need no fixups.
5963 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
5965 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
5966 exponent support with double type for first argument.
5967 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
5968 type returned by __builtin_vec_extract_sig,
5969 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
5970 functions from "vector int" to "vector unsigned int" or from
5971 "vector long long int" to "vector unsigned long long int".
5972 Changed type returned by __builtin_vec_extract_exp,
5973 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
5974 functions from "vector int" to "vector unsigned int" or from
5975 "vector long long int" to "vector unsigned long long int".
5976 Changed return type of __builtin_vec_test_data_class,
5977 __builtin_vec_test_data_class_sp, and
5978 __builtin_vec_test_data_class_dp from "vector int" to
5979 "vector bool int" or from "vector long long int" to "vector bool
5980 long long int" and changed second argument type from "unsigned
5981 int" to "int". Added new overloaded function forms "vector float
5982 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
5983 "vector float __builtin_vec_insert_exp_sp (vector float, vector
5984 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
5985 double, vector unsigned long long int)" and "vector double
5986 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
5987 long int)". Changed return type of
5988 __builtin_scalar_test_data_class and
5989 __builtin_scalar_test_data_class_sp and
5990 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
5991 int" and changed second argument from "unsigned int" to "int".
5992 Changed type returned by __builtin_scalar_test_neg,
5993 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
5994 from "int" to "bool int". Added new overloaded function form
5995 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
5996 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
5997 exponent double-precision with floating point first argument.
5998 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
5999 documentation of scalar_test_data_class, scalar_test_neg,
6000 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
6001 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
6002 vec_test_data_class built-in functions to reflect refinements in
6003 their type signatures.
6005 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
6007 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
6009 (aarch64_elf_asm_destructor): Likewise.
6011 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
6013 PR rtl-optimization/78634
6014 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
6015 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
6016 * ifcvt.c (noce_try_cmove): Add missing cost check.
6018 PR rtl-optimization/71724
6019 * combine.c (if_then_else_cond): Look for situations where it is
6020 beneficial to undo the work of one of the recursive calls.
6022 2017-01-23 Bin Cheng <bin.cheng@arm.com>
6024 PR tree-optimization/70754
6025 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
6026 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
6027 combined stmt before it if not NULL.
6028 (combine_chains): Process refs reversely and compute dominance point
6031 2017-01-23 Martin Liska <mliska@suse.cz>
6033 PR tree-optimization/79196
6034 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
6035 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
6037 (strlen_optimize_stmt): Call the renamed function.
6039 2017-01-23 Michael Matz <matz@suse.de>
6041 PR tree-optimization/78384
6042 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
6044 2017-01-23 Richard Biener <rguenther@suse.de>
6046 PR tree-optimization/79186
6047 * tree-vrp.c (register_new_assert_for): Make sure we've seen
6048 both incoming edges before moving an assert.
6050 2017-01-23 Martin Jambor <mjambor@suse.cz>
6052 * ipa-prop.c (load_from_param_1): Removed.
6053 (load_from_unmodified_param): Bits from load_from_param_1 put back
6055 (load_from_param): Removed.
6056 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
6057 with stmt. Reverted back to use of load_from_unmodified_param.
6059 2017-01-23 Martin Jambor <mjambor@suse.cz>
6062 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
6063 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
6064 field a pointer to garbage collected vector, mark lattices and
6065 ipcp_orig_node with GTY((skip)).
6066 (ipa_get_param_count): Adjust to descriptors being a pointer.
6067 (ipa_get_param): Likewise.
6068 (ipa_get_type): Likewise.
6069 (ipa_get_param_move_cost): Likewise.
6070 (ipa_set_param_used): Likewise.
6071 (ipa_get_controlled_uses): Likewise.
6072 (ipa_set_controlled_uses): Likewise.
6073 (ipa_is_param_used): Likewise.
6074 (ipa_node_params_t): Move into garbage collector. New methods insert
6076 (ipa_node_params_sum): Annotate wth GTY(()).
6077 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
6079 (ipa_load_from_parm_agg): Adjust declaration.
6080 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
6081 * ipa-profile.c (ipa_profile): Likewise.
6082 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
6083 (ipa_populate_param_decls): Make descriptors parameter garbage
6085 (ipa_dump_param): Adjust to descriptors being a pointer.
6086 (ipa_alloc_node_params): Likewise.
6087 (ipa_initialize_node_params): Likewise.
6088 (load_from_param_1): Make descriptors parameter garbage collected.
6089 (load_from_unmodified_param): Likewise.
6090 (load_from_param): Likewise.
6091 (ipa_load_from_parm_agg): Likewise.
6092 (ipa_node_params::~ipa_node_params): Removed.
6093 (ipa_free_all_node_params): Remove call to delete operator.
6094 (ipa_node_params_t::insert): New.
6095 (ipa_node_params_t::remove): Likewise.
6096 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
6097 copy known_csts and known_contexts vectors.
6098 (ipa_read_node_info): Adjust to descriptors being a pointer.
6099 (ipcp_modif_dom_walker): Make m_descriptors field garbage
6101 (ipcp_transform_function): Make descriptors variable garbage
6104 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
6106 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
6107 * config/i386/avx512dqintrin.h: Ditto.
6108 * config/i386/avx512fintrin.h: Ditto.
6109 * gcc/config/i386/i386.c: Handle new builtins.
6110 * config/i386/i386-builtin.def: Add new builtins.
6111 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
6112 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
6114 2017-01-23 Jakub Jelinek <jakub@redhat.com>
6115 Martin Liska <mliska@suse.cz>
6117 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
6118 * asan.c (asan_expand_poison_ifn): Support stores and use
6119 appropriate ASAN report function.
6120 * internal-fn.c (expand_ASAN_POISON_USE): New function.
6121 * internal-fn.def (ASAN_POISON_USE): Declare.
6122 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
6123 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
6124 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
6125 ASAN_POISON calls w/o LHS.
6126 * tree-ssa.c (execute_update_addresses_taken): Create clobber
6127 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
6128 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
6129 * gimplify.c (asan_poison_variables): Add attribute
6130 use_after_scope_memory to variables that really needs to live
6132 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
6133 having the attribute.
6135 2017-01-23 Martin Liska <mliska@suse.cz>
6137 * asan.c (create_asan_shadow_var): New function.
6138 (asan_expand_poison_ifn): Likewise.
6139 * asan.h (asan_expand_poison_ifn): New declaration.
6140 * internal-fn.c (expand_ASAN_POISON): Likewise.
6141 * internal-fn.def (ASAN_POISON): New builtin.
6142 * sanopt.c (pass_sanopt::execute): Expand
6143 asan_expand_poison_ifn.
6144 * tree-inline.c (copy_decl_for_dup_finish): Make function
6146 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
6147 * tree-ssa.c (is_asan_mark_p): New function.
6148 (execute_update_addresses_taken): Rewrite local variables
6149 (identified just by use-after-scope as addressable) into SSA.
6151 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
6153 * doc/install.texi (Specific): opensource.apple.com uses https
6154 now. Remove trailing slash.
6156 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
6158 * README.Portability: Remove note on an Irix compatibility issue.
6160 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
6162 * gcov.c (INCLUDE_ALGORITHM): Define.
6163 (INCLUDE_VECTOR): Define.
6164 No longer include <vector> and <algorithm> directly.
6166 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
6168 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
6170 * doc/invoke.texi (Code Gen Options): Ditto.
6172 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
6175 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
6177 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
6179 rtl-optimization/79125
6180 * cprop.c (local_cprop_pass): Handle cases where we make an
6183 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
6187 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
6188 read from, for big endian.
6190 2017-01-20 Jiong Wang <jiong.wang@arm.com>
6192 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
6193 register pauth builtins for LP64 only.
6195 2017-01-20 Marek Polacek <polacek@redhat.com>
6198 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
6201 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
6203 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
6205 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
6206 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
6207 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
6209 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6212 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
6213 in big-endian mode when they are not a single duplicated value.
6215 2017-01-20 Richard Biener <rguenther@suse.de>
6217 * BASE-VER: Bump to 7.0.1.
6219 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
6221 * omp-low.c (omplow_simd_context): New struct. Use it...
6222 (lower_rec_simd_input_clauses): ...here and...
6223 (lower_rec_input_clauses): ...here to hold common data. Adjust all
6224 references to idx, lane, max_vf, is_simt.
6226 2017-01-20 Graham Markall <graham.markall@embecosm.com>
6228 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
6231 2017-01-20 Martin Jambor <mjambor@suse.cz>
6233 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
6234 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
6236 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
6237 (GTFILES): Rename hsa.c to hsa-common.c.
6238 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
6239 * hsa-dump.c: Likewise.
6240 * hsa-gen.c: Likewise.
6241 * hsa-regalloc.c: Likewise.
6242 * ipa-hsa.c: Likewise.
6243 * omp-expand.c: Likewise.
6244 * omp-low.c: Likewise.
6245 * toplev.c: Likewise.
6247 2017-01-20 Marek Polacek <polacek@redhat.com>
6250 * doc/invoke.texi: Document -Wduplicated-branches.
6251 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
6252 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
6253 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
6254 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
6255 return 0 only when not OEP_LEXICOGRAPHIC.
6256 (fold_build_cleanup_point_expr): Use the expression
6257 location when building CLEANUP_POINT_EXPR.
6258 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
6259 * tree.c (add_expr): Handle error_mark_node.
6261 2017-01-20 Martin Liska <mliska@suse.cz>
6264 * tree-profile.c (init_ic_make_global_vars): Do not call
6266 (gimple_init_gcov_profiler): Likewise.
6268 2017-01-20 Martin Liska <mliska@suse.cz>
6271 * cgraph.h (maybe_create_reference): Remove argument and
6273 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
6275 * ipa-cp.c (create_specialized_node): Likewise.
6276 * symtab.c (symtab_node::maybe_create_reference): Handle
6277 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
6279 2017-01-20 Martin Liska <mliska@suse.cz>
6281 * read-rtl-function.c (function_reader::create_function): Use
6282 build_decl instread of build_decl_stat.
6284 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
6286 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
6287 * config/i386/avx512dqintrin.h: Ditto.
6288 * config/i386/avx512fintrin.h: Ditto.
6289 * config/i386/i386-builtin-types.def: Add new types.
6290 * gcc/config/i386/i386.c: Handle new types.
6291 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
6292 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
6293 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
6294 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
6295 (__builtin_ia32_kshiftridi): New.
6296 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
6298 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
6302 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
6303 define to rs6000_init_stack_protect_guard.
6304 (rs6000_init_stack_protect_guard): New function.
6306 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
6307 Yunqiang Su <yunqiang.su@imgtec.com>
6309 * config.gcc (supported_defaults): Add madd4.
6310 (with_madd4): Add validation.
6311 (all_defaults): Add madd4.
6312 * config/mips/mips.opt (mmadd4): New option.
6313 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
6315 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
6317 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
6318 (ISA_HAS_FUSED_MADD4): Likewise.
6319 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
6320 * gcc/doc/install.texi (--with-madd4): Document the new option.
6322 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6324 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
6325 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
6326 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
6327 (aarch64_init_pauth_hint_builtins): New.
6328 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
6329 (aarch64_expand_builtin): Expand new builtins.
6331 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6333 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
6334 * combine-stack-adj.c (no_unhandled_cfa): Handle
6335 REG_CFA_TOGGLE_RA_MANGLE.
6336 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
6337 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
6338 info for return address signing.
6339 (aarch64_expand_epilogue): Likewise.
6341 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6343 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
6344 * config/aarch64/aarch64-protos.h
6345 (aarch64_return_address_signing_enabled): New declaration.
6346 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
6348 (aarch64_expand_prologue): Sign return address before it's pushed onto
6350 (aarch64_expand_epilogue): Authenticate return address fetched from
6352 (aarch64_override_options): Sanity check for ILP32 and ISA level.
6353 (aarch64_attributes): New function attributes for "sign-return-address".
6354 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
6355 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
6356 ("*do_return"): Generate combined instructions according to key index.
6357 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
6358 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
6360 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
6361 * config/aarch64/aarch64.opt (msign-return-address=): New.
6362 * doc/extend.texi (AArch64 Function Attributes): Documents
6363 "sign-return-address=".
6364 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
6366 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
6368 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
6369 overall option summary.
6371 2017-01-19 Jiong Wang <jiong.wang@arm.com>
6373 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
6374 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
6375 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
6376 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
6378 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
6380 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
6381 -mpower9-minmax by default for -mcpu=power9.
6382 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
6383 128-bit floating point.
6385 2017-01-20 Alan Modra <amodra@gmail.com>
6387 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
6388 optimizing for size.
6390 2017-01-20 Alan Modra <amodra@gmail.com>
6393 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
6394 for strcmp and strncmp from corresponding builtin decl.
6396 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
6398 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
6399 instead of i386/rtems-64.h.
6400 * config/i386/rtems-64.h: Remove.
6402 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
6406 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
6408 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
6410 2017-01-19 Tamar Christina <tamar.christina@arm.com>
6412 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
6413 Change int to HOST_WIDE_INT.
6414 * config/aarch64/aarch64-protos.h
6415 (aarch64_simd_gen_const_vector_dup): Likewise.
6416 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
6418 2017-01-19 David Malcolm <dmalcolm@redhat.com>
6420 * langhooks-def.h (lhd_type_for_size): New decl.
6421 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
6422 * langhooks.c (lhd_type_for_size): New function, taken from
6425 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
6427 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
6428 define_bypass for CR latency.
6429 (power9-cracked-alu): Update bypass latency and remove power9-branch.
6430 (power9-alu2): Add define_bypass for CR latency.
6432 (power9-mul): Update insn latency.
6433 (power9-mul-compare): Update insn latency, bypass latency and remove
6436 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6438 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
6440 * config/aarch64/aarch64.md
6441 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
6442 aarch64_nopcrelative_literal_loads.
6443 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
6445 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
6447 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
6449 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
6451 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
6454 * config.gcc (supported_defaults): Add lxc1-sxc1.
6455 (with_lxc1_sxc1): Add validation.
6456 (all_defaults): Add lxc1-sxc1.
6457 * config/mips/mips.opt (mlxc1-sxc1): New option.
6458 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
6460 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
6461 __mips_no_lxc1_sxc1.
6462 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
6463 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
6464 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
6466 2017-01-19 Richard Biener <rguenther@suse.de>
6468 PR tree-optimization/72488
6469 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
6470 sure to restore SSA info.
6471 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
6473 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
6475 PR rtl-optimization/79121
6476 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
6477 of the inner type when shifting an extended value.
6479 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
6482 * symtab.c (symtab_node::equal_address_to): Fix comparing of
6483 interposable aliases.
6485 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
6488 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
6489 Use the evmergelohi instruction.
6490 (mov_si<mode>_e500_subreg4_2_le): Likewise.
6491 (mov_sitf_e500_subreg8_2_be): Likewise.
6492 (mov_sitf_e500_subreg12_2_le): Likewise.
6493 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
6494 (mov_si<mode>_e500_subreg4_2_be): Likewise.
6495 (mov_sitf_e500_subreg8_2_le): Likewise.
6496 (mov_sitf_e500_subreg12_2_be): Likewise.
6498 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6500 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
6501 attribute from vecsimple to vecperm.
6502 (altivec_vbpermq2): Likewise.
6504 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6507 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
6509 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6510 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
6511 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
6512 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
6513 case where N arg is SIZE_MAX.
6514 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
6515 (cmpstrsi): Add pattern.
6517 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
6519 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6520 __builtin_vec_revb builtins.
6521 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
6522 built-in functions to support generation of the ISA 3.0 XXBR<x>
6523 vector byte reverse instructions.
6524 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
6525 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
6526 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
6527 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
6528 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
6529 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
6530 (P9V_BUILTIN_VEC_REVB): Likewise.
6531 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
6532 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
6533 (p9_xxbrq_v16qi): Likewise.
6534 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
6535 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
6536 (p9_xxbrh_v8hi): Likewise.
6537 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
6538 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
6539 vec_revb built-in functions.
6541 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
6543 PR rtl-optimization/78952
6544 * config/i386/i386.md (any_extract): New code iterator.
6545 (*insvqi_2): Use any_extract for source operand.
6546 (*insvqi_3): Use any_shiftrt for source operand.
6548 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
6550 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
6552 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
6554 2017-01-18 Matthias Klose <doko@ubuntu.com>
6556 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
6558 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6560 * config/rs6000/altivec.h (vec_bperm): Change #define.
6561 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
6562 (altivec_vbpermq2): New define_insn.
6563 (altivec_vbpermd): Likewise.
6564 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
6566 (VBPERMD): Likewise.
6567 (VBPERM): New polymorphic function interface.
6568 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
6569 Add entries for P9V_BUILTIN_VEC_VBPERM.
6570 * doc/extend.texi: Add interfaces for vec_bperm.
6572 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6574 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
6575 first letter of error messages.
6576 (s390_resolve_overloaded_builtin): Likewise.
6577 * config/s390/s390.c (s390_expand_builtin): Likewise.
6578 (s390_invalid_arg_for_unprototyped_fn): Likewise.
6579 (s390_valid_target_attribute_inner_p): Likewise.
6580 * config/s390/s390.md ("tabort"): Likewise.
6582 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
6584 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
6585 (ISA_AVOID_DIV_HILO): New macro.
6586 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
6587 (ISA_HAS_DDIV): Likewise.
6589 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6591 * doc/invoke.texi (fabi-version): Correct number of occurrences.
6593 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6595 * doc/invoke.texi (fabi-version): Spelling fix.
6597 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6600 * doc/invoke.texi (fabi-version): Mention mangling fix for
6603 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
6606 * doc/invoke.texi (fabi-version): Document discriminator mangling.
6608 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
6611 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
6612 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
6614 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
6616 (stack_protect_test): Ditto.
6617 * config/rs6000/rs6000.opt (mstack-protector-guard=,
6618 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
6620 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
6621 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
6622 -mstack-protector-guard-offset=.
6623 (RS/6000 and PowerPC Options): Ditto.
6625 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
6627 * config/i386/i386.h (MASK_CLASS_P): New define.
6628 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
6629 there are no registers from different register sets also when
6630 mask registers are used. Update function comment.
6631 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
6632 to (*k/*r) and (*k/*km) alternatives.
6634 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
6636 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
6637 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
6638 (EH_RETURN_HANDLER_RTX): New define.
6639 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
6640 Force frame pointer in EH return functions.
6641 (aarch64_expand_epilogue): Add barrier for eh_return.
6642 (aarch64_final_eh_return_addr): Remove.
6643 (aarch64_eh_return_handler_rtx): New function.
6644 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
6646 (aarch64_eh_return_handler_rtx): New prototype.
6648 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6650 * config/rs6000/altivec.h (vec_rlmi): New #define.
6651 (vec_vrlnm): Likewise.
6652 (vec_rlnm): Likewise.
6653 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
6654 (UNSPEC_VRLNM): Likewise.
6655 (VIlong): New mode iterator.
6656 (altivec_vrl<VI_char>mi): New define_insn.
6657 (altivec_vrl<VI_char>nm): Likewise.
6658 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
6661 (RLNM): New polymorphic function entry.
6662 (VRLWMI): New monomorphic function entry.
6664 (RLMI): New polymorphic function entry.
6665 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
6666 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
6667 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
6670 2017-01-17 Jakub Jelinek <jakub@redhat.com>
6673 * dwarf2out.c (field_byte_offset): Restore the
6674 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
6675 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
6676 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
6679 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
6682 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
6684 2017-01-17 Jakub Jelinek <jakub@redhat.com>
6687 * dwarf2out.c (add_data_member_location_attribute): For constant
6688 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
6689 instead of DW_AT_data_member_location, DW_AT_bit_offset and
6690 DW_AT_byte_size attributes.
6692 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
6694 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
6695 after forcing to constant memory when the code model is medium.
6697 2017-01-17 Julia Koval <julia.koval@intel.com>
6700 * config/i386/avx512fintrin.h
6701 (_mm512_i32gather_ps): Change __addr type to void const*.
6702 (_mm512_mask_i32gather_ps): Ditto.
6703 (_mm512_i32gather_pd): Ditto.
6704 (_mm512_mask_i32gather_pd): Ditto.
6705 (_mm512_i64gather_ps): Ditto.
6706 (_mm512_mask_i64gather_ps): Ditto.
6707 (_mm512_i64gather_pd): Ditto.
6708 (_mm512_mask_i64gather_pd): Ditto.
6709 (_mm512_i32gather_epi32): Ditto.
6710 (_mm512_mask_i32gather_epi32): Ditto.
6711 (_mm512_i32gather_epi64): Ditto.
6712 (_mm512_mask_i32gather_epi64): Ditto.
6713 (_mm512_i64gather_epi32): Ditto.
6714 (_mm512_mask_i64gather_epi32): Ditto.
6715 (_mm512_i64gather_epi64): Ditto.
6716 (_mm512_mask_i64gather_epi64): Ditto.
6717 (_mm512_i32scatter_ps): Change __addr type to void*.
6718 (_mm512_mask_i32scatter_ps): Ditto.
6719 (_mm512_i32scatter_pd): Ditto.
6720 (_mm512_mask_i32scatter_pd): Ditto.
6721 (_mm512_i64scatter_ps): Ditto.
6722 (_mm512_mask_i64scatter_ps): Ditto.
6723 (_mm512_i64scatter_pd): Ditto.
6724 (_mm512_mask_i64scatter_pd): Ditto.
6725 (_mm512_i32scatter_epi32): Ditto.
6726 (_mm512_mask_i32scatter_epi32): Ditto.
6727 (_mm512_i32scatter_epi64): Ditto.
6728 (_mm512_mask_i32scatter_epi64): Ditto.
6729 (_mm512_i64scatter_epi32): Ditto.
6730 (_mm512_mask_i64scatter_epi32): Ditto.
6731 (_mm512_i64scatter_epi64): Ditto.
6732 (_mm512_mask_i64scatter_epi64): Ditto.
6733 * config/i386/avx512pfintrin.h
6734 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
6735 (_mm512_mask_prefetch_i32gather_ps): Ditto.
6736 (_mm512_mask_prefetch_i64gather_pd): Ditto.
6737 (_mm512_mask_prefetch_i64gather_ps): Ditto.
6738 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
6739 (_mm512_prefetch_i32scatter_ps): Ditto.
6740 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
6741 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
6742 (_mm512_prefetch_i64scatter_pd): Ditto.
6743 (_mm512_prefetch_i64scatter_ps): Ditto.
6744 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
6745 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
6746 * config/i386/avx512vlintrin.h
6747 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
6748 (_mm_mmask_i32gather_ps): Ditto.
6749 (_mm256_mmask_i32gather_pd): Ditto.
6750 (_mm_mmask_i32gather_pd): Ditto.
6751 (_mm256_mmask_i64gather_ps): Ditto.
6752 (_mm_mmask_i64gather_ps): Ditto.
6753 (_mm256_mmask_i64gather_pd): Ditto.
6754 (_mm_mmask_i64gather_pd): Ditto.
6755 (_mm256_mmask_i32gather_epi32): Ditto.
6756 (_mm_mmask_i32gather_epi32): Ditto.
6757 (_mm256_mmask_i32gather_epi64): Ditto.
6758 (_mm_mmask_i32gather_epi64): Ditto.
6759 (_mm256_mmask_i64gather_epi32): Ditto.
6760 (_mm_mmask_i64gather_epi32): Ditto.
6761 (_mm256_mmask_i64gather_epi64): Ditto.
6762 (_mm_mmask_i64gather_epi64): Ditto.
6763 (_mm256_i32scatter_ps): Change __addr type to void*.
6764 (_mm256_mask_i32scatter_ps): Ditto.
6765 (_mm_i32scatter_ps): Ditto.
6766 (_mm_mask_i32scatter_ps): Ditto.
6767 (_mm256_i32scatter_pd): Ditto.
6768 (_mm256_mask_i32scatter_pd): Ditto.
6769 (_mm_i32scatter_pd): Ditto.
6770 (_mm_mask_i32scatter_pd): Ditto.
6771 (_mm256_i64scatter_ps): Ditto.
6772 (_mm256_mask_i64scatter_ps): Ditto.
6773 (_mm_i64scatter_ps): Ditto.
6774 (_mm_mask_i64scatter_ps): Ditto.
6775 (_mm256_i64scatter_pd): Ditto.
6776 (_mm256_mask_i64scatter_pd): Ditto.
6777 (_mm_i64scatter_pd): Ditto.
6778 (_mm_mask_i64scatter_pd): Ditto.
6779 (_mm256_i32scatter_epi32): Ditto.
6780 (_mm256_mask_i32scatter_epi32): Ditto.
6781 (_mm_i32scatter_epi32): Ditto.
6782 (_mm_mask_i32scatter_epi32): Ditto.
6783 (_mm256_i32scatter_epi64): Ditto.
6784 (_mm256_mask_i32scatter_epi64): Ditto.
6785 (_mm_i32scatter_epi64): Ditto.
6786 (_mm_mask_i32scatter_epi64): Ditto.
6787 (_mm256_i64scatter_epi32): Ditto.
6788 (_mm256_mask_i64scatter_epi32): Ditto.
6789 (_mm_i64scatter_epi32): Ditto.
6790 (_mm_mask_i64scatter_epi32): Ditto.
6791 (_mm256_i64scatter_epi64): Ditto.
6792 (_mm256_mask_i64scatter_epi64): Ditto.
6793 (_mm_i64scatter_epi64): Ditto.
6794 (_mm_mask_i64scatter_epi64): Ditto.
6795 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
6796 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
6797 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
6798 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
6799 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
6800 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
6801 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
6802 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
6803 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
6804 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
6805 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
6806 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
6807 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
6808 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
6809 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
6810 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
6811 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
6812 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
6813 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
6814 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
6815 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
6816 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
6817 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
6818 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
6819 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
6820 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
6821 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
6822 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
6823 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
6824 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
6825 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
6826 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
6827 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
6828 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
6829 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
6830 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
6831 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
6832 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
6833 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
6834 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
6835 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
6836 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
6837 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
6838 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
6839 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
6840 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
6841 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
6842 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
6843 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
6844 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
6845 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
6846 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
6847 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
6848 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
6849 definitions accordingly.
6851 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
6852 Kuan-Lin Chen <kuanlinchentw@gmail.com>
6855 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
6858 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
6861 * ira-conflicts.c (ira_build_conflicts): Update total conflict
6862 hard regs for inner regno.
6864 2017-01-17 Martin Liska <mliska@suse.cz>
6867 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
6868 assumption and add comment.
6870 2017-01-17 Nathan Sidwell <nathan@acm.org>
6872 * ipa-visibility.c (localize_node): New function, broken out of ...
6873 (function_and_variable_visibility): ... here. Call it.
6875 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
6878 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
6879 correctly set frequency of oudgoing edge.
6880 (duplicate_thread_path): Fix profile updating.
6882 2017-01-17 Jakub Jelinek <jakub@redhat.com>
6885 * configure.ac: Add GCC_BASE_VER.
6886 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
6887 version from BASE-VER file.
6888 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
6889 (gcc.o): Depend on $(BASEVER).
6890 * common.opt (dumpfullversion): New option.
6891 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
6892 * doc/invoke.texi: Document -dumpfullversion.
6893 * doc/install.texi: Document --with-gcc-major-version-only.
6894 * configure: Regenerated.
6896 2017-01-17 Richard Biener <rguenther@suse.de>
6898 PR tree-optimization/71433
6899 * tree-vrp.c (register_new_assert_for): Merge same asserts
6900 on all incoming edges.
6901 (process_assert_insertions_for): Handle insertions at the
6904 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
6906 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
6907 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
6909 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
6912 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
6915 2017-01-17 Alan Modra <amodra@gmail.com>
6918 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
6919 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
6920 symbolic stack limit when pic.
6922 2017-01-16 Martin Sebor <msebor@redhat.com>
6924 PR tree-optimization/78608
6925 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
6927 2017-01-16 Jeff Law <law@redhat.com>
6930 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
6931 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
6932 for several include directories that may be relative to sysroot.
6933 * config/i386/x-mingw32 (gplus_includedir): Define.
6934 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
6935 (native_system_includedir): Likewise.
6936 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
6937 override if TARGET_SYSTEM_ROOT is defined.
6938 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
6940 PR tree-optimization/79090
6941 PR tree-optimization/33562
6942 PR tree-optimization/61912
6943 PR tree-optimization/77485
6944 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
6945 and computed trims into the dump file.
6947 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
6949 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
6951 2017-01-16 Jakub Jelinek <jakub@redhat.com>
6954 * gimplify.c (gimplify_init_constructor): If want_value and
6955 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
6959 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
6960 sequence. Formatting fixes.
6961 (doloop_optimize): Formatting fixes.
6964 * gcc.c (debug_level_greater_than_spec_func): New function.
6965 (static_spec_functions): Add debug-level-gt spec function.
6966 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
6968 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
6969 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
6970 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
6971 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
6972 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
6973 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
6975 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
6977 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
6978 QImode fixups to general and mask registers only.
6980 2017-01-16 Carl Love <cel@us.ibm.com>
6982 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
6983 for built-in functions
6984 vector signed char vec_nabs (vector signed char)
6985 vector signed short vec_nabs (vector signed short)
6986 vector signed int vec_nabs (vector signed int)
6987 vector signed long long vec_nabs (vector signed long long)
6988 vector float vec_nabs (vector float)
6989 vector double vec_nabs (vector double)
6990 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
6992 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
6993 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
6994 * doc/extend.texi: Update the documentation file for the new built-in
6997 2017-01-16 Martin Sebor <msebor@redhat.com>
6999 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
7002 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7004 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
7005 UNSPEC_VSX__XXSPLTD to require special splat handling.
7007 2017-01-16 David Malcolm <dmalcolm@redhat.com>
7010 * system.h: Poison strndup.
7012 2017-01-16 Alan Modra <amodra@gmail.com>
7015 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
7018 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
7020 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
7022 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
7024 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
7025 call recog here. Assert that INSN_CODE (insn) is non-negative.
7027 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
7030 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
7032 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
7033 in the currently scheduled RTL fragment.
7035 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
7037 PR rtl-optimization/78751
7038 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
7041 2017-01-14 Jeff Law <law@redhat.com>
7043 PR tree-optimization/79090
7044 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
7045 variable length stores.
7046 (compute_trims): Delete dead assignment to *trim_tail.
7047 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
7050 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
7052 PR rtl-optimization/78626
7053 PR rtl-optimization/78727
7054 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
7055 of a block, and split such blocks after everything else is finished.
7057 2017-01-14 Alan Modra <amodra@gmail.com>
7060 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
7061 target legitimate_combined_insn.
7062 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
7063 (rs6000_legitimate_combined_insn): New function.
7064 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
7066 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
7067 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
7068 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
7070 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
7072 * doc/frontends.texi (G++ and GCC): Remove references to Java.
7074 2017-01-13 Jeff Law <law@redhat.com>
7076 PR tree-optimization/33562
7077 PR tree-optimization/61912
7078 PR tree-optimization/77485
7079 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
7081 (delete_dead_assignment): Likewise.
7082 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
7083 statement to delete_dead_call and delete_dead_assignment.
7085 2017-01-13 David Malcolm <dmalcolm@redhat.com>
7088 * substring-locations.c (format_warning_va): Strengthen case 1 so
7089 that both endpoints of the substring must be within the format
7090 range for just the substring to be printed.
7092 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
7094 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
7095 * config/i386/i386.c (ix86_target_string): Add missing options
7096 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
7097 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
7098 flags_other and ix86_target_other to flags2_other. Display unknown
7100 (ix86_valid_target_attribute_inner_p): Add missing options and
7101 reorder options by implied ISAs, as in ix86_target_string.
7103 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
7105 * hash-table.h (hash_table::too_empty_p): New function.
7106 (hash_table::expand): Use it.
7107 (hash_table::traverse): Likewise.
7108 (hash_table::empty_slot): Use sizeof (value_type) instead of
7109 sizeof (PTR) to convert bytes to elements. Shrink the table
7110 if the current size is excessive for the current number of
7113 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
7115 * ira-costs.c (record_reg_classes): Break from the inner loop
7116 early once alt_fail is known to be true. Update outer loop
7117 handling accordingly.
7119 2017-01-13 Jeff Law <law@redhat.com>
7121 * tree-ssa-dse.c (decrement_count): New function.
7122 (increment_start_addr, maybe_trim_memstar_call): Likewise.
7123 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
7124 when we know the partially dead statement is a mem* function.
7126 PR tree-optimization/61912
7127 PR tree-optimization/77485
7128 * tree-ssa-dse.c: Include expr.h.
7129 (maybe_trim_constructor_store): New function.
7130 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
7132 PR tree-optimization/33562
7133 PR tree-optimization/61912
7134 PR tree-optimization/77485
7135 * doc/invoke.texi: Document new dse-max-object-size param.
7136 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
7137 * tree-ssa-dse.c: Include params.h.
7138 (dse_store_status): New enum.
7139 (initialize_ao_ref_for_dse): New, partially extracted from
7141 (valid_ao_ref_for_dse, normalize_ref): New.
7142 (setup_live_bytes_from_ref, compute_trims): Likewise.
7143 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
7144 (maybe_trim_partially_dead_store): Likewise.
7145 (maybe_trim_complex_store): Likewise.
7146 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
7147 Track what bytes live from the original store. Return tri-state
7148 for dead, partially dead or live.
7149 (dse_dom_walker): Add constructor, destructor and new private members.
7150 (delete_dead_call, delete_dead_assignment): New extracted from
7152 (dse_optimize_stmt): Make a member of dse_dom_walker.
7153 Use initialize_ao_ref_for_dse.
7155 PR tree-optimization/33562
7156 PR tree-optimization/61912
7157 PR tree-optimization/77485
7158 * sbitmap.h (bitmap_count_bits): Prototype.
7159 (bitmap_clear_range, bitmap_set_range): Likewise.
7160 * sbitmap.c (bitmap_clear_range): New function.
7161 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
7163 2017-01-13 Martin Liska <mliska@suse.cz>
7166 * function.c (set_cfun): Add new argument force.
7167 * function.h (set_cfun): Likewise.
7168 * ipa-inline-transform.c (inline_call): Use the function when
7169 strict alising from is dropped for function we inline to.
7171 2017-01-13 Richard Biener <rguenther@suse.de>
7173 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
7174 for dumping GIMPLE INTEGER_CSTs.
7176 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7178 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
7179 to 201112L since C++17.
7181 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7184 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
7185 if -fsanitize=kernel-address is present.
7187 2017-01-13 Richard Biener <rguenther@suse.de>
7189 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
7190 as _Literal ( type ) number in case usual suffixes do not
7191 preserve all information.
7193 2017-01-13 Richard Biener <rguenther@suse.de>
7195 PR tree-optimization/77283
7196 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
7197 and ssa-iterators.h.
7198 (is_feasible_trace): Implement a cost model based on joiner
7201 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7204 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
7205 char or short to __float128/_Float128 directly.
7207 2017-01-12 Martin Sebor <msebor@redhat.com>
7209 to -Wformat-overflow.
7210 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
7211 (min_bytes_remaining): Same.
7212 (get_string_length): Same.
7213 (format_string): Same.
7214 (format_directive): Same.
7216 (pass_sprintf_length::handle_gimple_call): Same.
7218 2017-01-12 Jakub Jelinek <jakub@redhat.com>
7220 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
7221 info.nowrite calls with no lhs that can't throw. Return bool
7222 whether gsi_remove has been called or not.
7223 (pass_sprintf_length::handle_gimple_call): Return bool whether
7224 try_substitute_return_value called gsi_remove. Formatting fix.
7225 (pass_sprintf_length::execute): Don't use gsi_remove if
7226 handle_gimple_call returned true.
7229 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
7230 be removed due to side-effects, don't remove following barrier nor
7231 turn the successor edge into fallthru edge.
7233 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7236 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
7237 element-reversing loads and stores as not swappable.
7239 2017-01-12 Nathan Sidwell <nathan@acm.org>
7240 Nicolai Stange <nicstange@gmail.com>
7242 * combine.c (try_combine): Don't ignore result of overlap checking
7243 loop. Combine overlap & asm check into single loop.
7245 2017-01-12 Richard Biener <rguenther@suse.de>
7247 * tree-pretty-print.c (dump_generic_node): Provide -gimple
7248 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
7250 2017-01-12 Richard Biener <rguenther@suse.de>
7252 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
7253 and TS_TARGET_OPTION directly derive from TS_BASE.
7254 * tree-core.h (tree_optimization_option): Derive from tree_base.
7255 (tree_target_option): Likewise.
7257 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
7259 * config/i386/i386.c (memory_address_length): Increase len
7260 only when rip_relative_addr_p returns false.
7262 2017-01-11 Julia Koval <julia.koval@intel.com>
7264 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
7265 (OPTION_MASK_ISA_SGX_SET): New.
7266 (ix86_handle_option): Handle OPT_msgx.
7267 * config.gcc: Added sgxintrin.h.
7268 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
7269 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
7270 * config/i386/i386.c (ix86_target_string): Add -msgx.
7272 (ix86_option_override_internal): Handle new options.
7273 (ix86_valid_target_attribute_inner_p): Add sgx.
7274 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
7275 * config/i386/i386.opt: Add msgx.
7276 * config/i386/sgxintrin.h: New file.
7277 * config/i386/x86intrin.h: Add sgxintrin.h.
7279 2017-01-11 Jakub Jelinek <jakub@redhat.com>
7282 * fold-const.c (maybe_nonzero_address): Return 1 for function
7284 (tree_single_nonzero_warnv_p): Don't handle function local objects
7288 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
7291 2017-01-11 David Malcolm <dmalcolm@redhat.com>
7294 * opts.c: Include "spellcheck.h"
7295 (struct string_fragment): New struct.
7296 (struct edit_distance_traits<const string_fragment &>): New
7298 (get_closest_sanitizer_option): New function.
7299 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
7301 2017-01-11 Jakub Jelinek <jakub@redhat.com>
7303 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
7305 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
7306 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
7307 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
7308 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
7309 for initial die_offset if dwarf_split_debug_info.
7310 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
7311 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
7313 (output_skeleton_debug_sections): Formatting fix. Use
7314 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
7315 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
7317 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
7319 * config/arm/cortex-a53.md: Add bypasses for
7321 (cortex_a53_r2f): Only use for transfers.
7322 (cortex_a53_f2r): Likewise.
7323 (cortex_a53_r2f_cvt): Add reservation for conversions.
7324 (cortex_a53_f2r_cvt): Likewise.
7326 2017-01-11 Tamar Christina <tamar.christina@arm.com>
7328 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
7329 to all inlined functions, change static to extern.
7331 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
7334 * config/arm/arm.c (legitimize_pic_address): Handle reference to
7336 (arm_assemble_integer): Likewise.
7338 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
7340 * config.gcc: Use new awk script to check CPU, FPU and architecture
7341 parameters for --with-... options.
7342 * config/arm/parsecpu.awk: New file
7343 * config/arm/arm-cpus.in: New file.
7344 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
7346 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
7348 * config/arm/t-arm: Update dependency rules.
7349 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
7350 of processing .def files.
7351 * config/arm/genopt.sh: Deleted.
7352 * config/arm/gentune.sh: Deleted.
7353 * config/arm/arm-cores.def: Deleted.
7354 * config/arm/arm-arches.def: Deleted.
7355 * config/arm/arm-fpus.def: Deleted.
7356 * config/arm/arm-tune.md: Regenerated.
7357 * config/arm/arm-tables.opt: Regenerated.
7358 * config/arm/arm-cpu.h: New generated file.
7359 * config/arm/arm-cpu-data.h: New generated file.
7360 * config/arm/arm-cpu-cdata.h: New generated file.
7362 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
7365 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
7367 (input_varpool_node): Unpack dynamically_initialized bit.
7369 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
7371 PR rtl-optimization/79032
7372 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
7373 the alignment of the adjusted memory reference against that of MODE,
7374 instead of the alignment of the original memory reference.
7376 2017-01-11 Martin Jambor <mjambor@suse.cz>
7378 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
7380 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
7381 decorated functions.
7383 2017-01-11 Richard Biener <rguenther@suse.de>
7385 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
7386 set range/nonnull info for PHI results. Do not set it on
7387 stmts marked for removal.
7389 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
7391 * expr.c (store_field): In the bitfield case, fetch the return value
7392 from the registers before applying a single big-endian adjustment.
7393 Always do a final load for a BLKmode value not larger than a word.
7395 2017-01-10 David Malcolm <dmalcolm@redhat.com>
7398 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
7399 that we correctly handle column numbers greater than
7400 LINE_MAP_MAX_COLUMN_NUMBER.
7402 2017-01-10 Martin Sebor <msebor@redhat.com>
7405 * gimple-ssa-sprintf.c (get_destination_size): Call
7406 {init,fini}object_sizes.
7407 * tree-object-size.c (addr_object_size): Adjust.
7408 (pass_through_call): Adjust.
7409 (pass_object_sizes::execute): Adjust.
7410 * tree-object-size.h (fini_object_sizes): Declare.
7412 2017-01-10 Martin Sebor <msebor@redhat.com>
7414 PR tree-optimization/78775
7415 * builtins.c (get_size_range): Move...
7416 * calls.c: ...to here.
7417 (alloc_max_size): Accept zero argument.
7418 (operand_signed_p): Remove.
7419 (maybe_warn_alloc_args_overflow): Call get_size_range.
7420 * calls.h (get_size_range): Declare.
7422 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
7424 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
7425 from TI's devices.csv file as of September 2016.
7426 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
7428 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
7430 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
7431 * doc/invoke.texi: Likewise.
7432 * doc/md.texi: Likewise.
7433 * doc/objc.texi: Likewise.
7435 2017-01-10 Joshua Conner <joshconner@google.com>
7437 * config/arm/fuchsia-elf.h: New file.
7438 * config/fuchsia.h: New file.
7439 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
7440 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
7442 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
7444 2016-01-10 Richard Biener <rguenther@suse.de>
7446 PR tree-optimization/79034
7447 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
7448 Propagate out degenerate PHIs in the joiner.
7450 2017-01-10 Martin Liska <mliska@suse.cz>
7452 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
7453 (sort_congruence_classes_by_decl_uid): Likewise.
7454 (sort_congruence_class_groups_by_decl_uid): Likewise.
7455 (sem_item_optimizer::merge_classes): Sort class, groups in these
7456 classes and members in the groups by DECL_UID of declarations.
7457 This would make merge operations stable.
7459 2017-01-10 Martin Liska <mliska@suse.cz>
7461 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
7462 usage of m_classes_vec.
7463 (sem_item_optimizer::~sem_item_optimizer): Likewise.
7464 (sem_item_optimizer::get_group_by_hash): Likewise.
7465 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
7466 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
7467 (sem_item_optimizer::verify_classes): Likewise.
7468 (sem_item_optimizer::process_cong_reduction): Likewise.
7469 (sem_item_optimizer::dump_cong_classes): Likewise.
7470 (sem_item_optimizer::merge_classes): Likewise.
7471 * ipa-icf.h (congruence_class_hash): Rename from
7472 congruence_class_group_hash. Remove declaration of m_classes_vec.
7474 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
7476 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
7477 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
7478 * config.gcc: Add avx512vpopcntdqintrin.h.
7479 * config/i386/avx512vpopcntdqintrin.h: New.
7480 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
7481 * config/i386/i386-builtin-types.def: Add new types.
7482 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
7483 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
7484 __builtin_ia32_vpopcountq_v8di_mask): New.
7485 * config/i386/i386-c.c (ix86_target_macros_internal): Define
7486 __AVX512VPOPCNTDQ__.
7487 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
7488 (PTA_AVX512VPOPCNTDQ): Define.
7489 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
7490 TARGET_AVX512VPOPCNTDQ_P): Define.
7491 * config/i386/i386.opt: Add mavx512vpopcntdq.
7492 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
7493 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
7495 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7498 * predict.def (PRED_CALL): Set to 67.
7500 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
7502 * expr.c (store_field): In the bitfield case, if the value comes from
7503 a function call and is of an aggregate type returned in registers, do
7504 not modify the field mode; extract the value in all cases if the mode
7505 is BLKmode and the size is not larger than a word.
7507 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
7510 * config/i386/cpuid.h: Fix undefined behavior.
7512 2017-01-04 Jeff Law <law@redhat.com>
7514 PR tree-optimization/79007
7515 PR tree-optimization/67955
7516 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
7517 conservative for pt.null when flag_non_call_exceptions is on.
7519 2017-01-09 Jakub Jelinek <jakub@redhat.com>
7521 PR translation/79019
7522 PR translation/79020
7523 * params.def (PARAM_INLINE_MIN_SPEEDUP,
7524 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
7525 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
7527 * config/avr/avr.opt (maccumulate-args): Likewise.
7528 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
7529 * common.opt (freport-bug): Likewise.
7530 * cif-code.def (CIF_FINAL_ERROR): Likewise.
7531 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
7532 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
7533 translatable string.
7534 * config/i386/i386.c (function_value_32): Likewise.
7535 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
7536 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
7538 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
7539 * common/config/msp430/msp430-common.c (msp430_handle_option):
7541 * symtab.c (symtab_node::verify_base): Likewise.
7542 * opts.c (set_debug_level): Likewise.
7543 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
7544 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
7545 missing whitespace to translatable strings.
7546 * config/avr/avr.md (bswapsi2): Fix typo in comment.
7547 * config/sh/superh.h: Likewise.
7548 * config/i386/xopintrin.h: Likewise.
7549 * config/i386/znver1.md: Likewise.
7550 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
7551 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
7552 * double-int.h (struct double_int): Likewise.
7553 * double-int.c (div_and_round_double): Likewise.
7554 * wide-int.cc: Likewise.
7555 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
7556 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
7557 * cfgcleanup.c (crossjumps_occured): Renamed to ...
7558 (crossjumps_occurred): ... this.
7559 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
7562 PR tree-optimization/78899
7563 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
7564 returning bool return struct loop *, NULL for failure and the new
7566 (versionable_outer_loop_p): Don't version outer loop if it has
7567 dont_vectorized bit set.
7568 (tree_if_conversion): When versioning outer loop, ensure
7569 tree_if_conversion is performed also on the inner loop of the
7570 non-vectorizable outer loop copy.
7571 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
7572 LOOP_VECTORIZED in inner loop of the scalar outer loop and
7573 prevent vectorization of it.
7574 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
7575 the outer loop vectorization of the non-scalar version is attempted
7576 before vectorization of the inner loop in scalar version. If
7577 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
7578 vectorization of its inner loop.
7579 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
7580 has 2 inner loops, rename also on edges from bb whose single pred
7581 is outer_loop->header. Fix typo in function comment.
7583 2017-01-09 Martin Sebor <msebor@redhat.com>
7586 * asan.c (asan_emit_stack_protection): Increase local buffer size
7587 to avoid snprintf truncation warning.
7589 2017-01-09 Andrew Pinski <apinski@cavium.com>
7591 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
7592 to reference thunderx2t99 for the tuning structure
7593 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
7595 (thunderx2t99_extra_costs): This.
7596 * config/aarch64/aarch64-tune.md: Regenerate.
7597 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
7598 (vulcan_addrcost_table): This.
7599 (vulcan_regmove_cost): Rename to ...
7600 (thunderx2t99_regmove_cost): This.
7601 (vulcan_vector_cost): Rename to ...
7602 (thunderx2t99_vector_cost): this.
7603 (vulcan_branch_cost): Rename to ...
7604 (thunderx2t99_branch_cost): This.
7605 (vulcan_tunings): Rename to ...
7606 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
7607 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
7609 2017-01-09 Martin Jambor <mjambor@suse.cz>
7613 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
7614 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
7615 (propagate_vr_accross_jump_function): Use the above function for all
7616 value range computations for pass-through jump functions and type
7617 converasion from explicit value range values.
7618 (ipcp_propagate_stage): Do not attempt to deduce types of formal
7619 parameters from TYPE_ARG_TYPES.
7620 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
7621 (ipa_write_node_info): Stream type of the actual argument.
7622 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
7624 2017-01-09 Martin Liska <mliska@suse.cz>
7627 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
7628 (lookup_compiler): Do not show error message with have_E.
7630 2017-01-09 Jakub Jelinek <jakub@redhat.com>
7632 PR tree-optimization/78938
7633 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
7634 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
7635 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
7636 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
7639 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7641 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
7644 2017-01-09 Richard Biener <rguenther@suse.de>
7646 PR tree-optimization/78997
7647 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
7648 name condition properly.
7650 2017-01-09 Richard Biener <rguenther@suse.de>
7653 * dwarf2out.c (is_cxx): New overload with context.
7654 (is_naming_typedef_decl): Use it.
7656 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7658 * invoke.texi (Option Summary): Correct spacing in option lists
7659 and add line breaks to fix over-long lines.
7661 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7665 * extend.texi (Common Variable Attributes): Add xref to GCC
7666 Internals manual to explain mode attribute keywords.
7668 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
7671 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
7672 and Preprocessor Options.
7673 (Options for Linking): Document -pthread here....
7674 (RS/6000 and PowerPC Options): ...not here.
7675 (Solaris 2 Options): ...or here.
7676 * doc/cppopts.texi: Document -pthread.
7678 2017-01-08 Martin Sebor <msebor@redhat.com>
7681 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
7682 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
7683 New member functions.
7684 (format_directive): Used them.
7686 (pass_sprintf_length::handle_gimple_call): Same.
7687 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
7688 to avoid truncation for any argument.
7689 (extract_affine_mul): Same.
7690 * tree.c (get_file_function_name): Same.
7692 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7695 * predict.def (PRED_INDIR_CALL): Set to 86.
7697 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
7699 PR preprocessor/54124
7700 * doc/cppopts.texi: Reformat -d subtable to list the full name
7701 of the options. Add cross-reference to the docs for the general
7702 compiler -d options.
7703 * doc/invoke.texi (Developer Options): Add cross-reference to the
7704 preprocessor-specific -d option documentation.
7706 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
7708 PR preprocessor/13498
7709 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
7710 redudant material, and reflect new command-line options.
7711 (System Headers): Likewise.
7713 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
7715 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
7716 -isystem, and -idirafter. Copy-edit.
7717 * doc/cppopts.texi: Copy-edit. Remove contradiction about
7718 default for -ftrack-macro-expansion. Delete obsolete and
7719 badly-formatted implementation details about -fdebug-cpp output.
7720 * doc/cppwarnopts.texi: Copy-edit.
7722 2017-01-07 David Malcolm <dmalcolm@redhat.com>
7725 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
7726 that the transition from a max line width >= 1<<10 to narrower
7727 lines works correctly.
7729 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
7731 * doc/options.texi (PerFunction): New.
7732 * opt-functions.awk (switch_flags): Map both Optimization and
7733 PerFunction to CL_OPTIMIZATION.
7734 * opth-gen.awk: Test for PerFunction flag along with
7736 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
7737 it only when the latter is present. Skip those that don't in
7738 the hash function generator.
7739 * common.opt (fvar-tracking): Mark as PerFunction instead of
7741 (fvar-tracking-assignments): Likewise.
7742 (fvar-tracking-assignments-toggle): Likewise.
7743 (fvar-tracking-uninit): Likewise.
7745 2017-01-07 Jakub Jelinek <jakub@redhat.com>
7747 PR translation/79018
7748 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
7751 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
7754 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
7755 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
7756 TARGET_LONG_JUMP_TABLE_OFFSETS.
7757 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
7758 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
7759 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
7760 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
7761 * config/m68k/m68k.md (tablejump expander): Likewise.
7762 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
7763 TARGET_LONG_JUMP_TABLE_OFFSETS.
7764 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
7765 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
7767 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
7768 David Holsgrove <david.holsgrove@xilinx.com>
7770 * common/config/microblaze/microblaze-common.c
7771 (TARGET_EXCEPT_UNWIND_INFO): Remove.
7772 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
7774 * config/microblaze/microblaze.c (microblaze_must_save_register)
7775 (microblaze_expand_epilogue, microblaze_return_addr): Handle
7777 (microblaze_eh_return): New function.
7778 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
7779 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
7780 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
7781 * config/microblaze/microblaze.md (eh_return): New pattern.
7783 2017-01-06 Jakub Jelinek <jakub@redhat.com>
7785 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
7786 GCC_DIAGNOSTIC_STRINGIFY): Define.
7788 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
7790 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7792 * config/arm/arm.md (<mcrr>): New.
7794 * config/arm/arm.c (arm_arch5te): New.
7795 (arm_option_override): Set arm_arch5te.
7796 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
7798 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
7799 (arm_mcrr_qualifiers): ... this. New.
7800 (MRRC_QUALIFIERS): Define to...
7801 (arm_mrrc_qualifiers): ... this. New.
7802 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
7804 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
7805 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
7806 (MRRCI, mrrc, MRRC): New.
7807 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
7808 VUNSPEC_MRRC2): New.
7810 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7812 * config/arm/arm.md (<mcr>): New.
7814 * config/arm/arm.c (arm_coproc_builtin_available): Add
7815 support for mcr, mrc, mcr2 and mrc2.
7816 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
7817 (arm_mcr_qualifiers): ... this. New.
7818 (MRC_QUALIFIERS): Define to ...
7819 (arm_mrc_qualifiers): ... this. New.
7820 (MCR_QUALIFIERS): Define to ...
7821 (arm_mcr_qualifiers): ... this. New.
7822 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
7824 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
7825 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
7826 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
7829 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7831 * config/arm/arm.md (*ldc): New.
7835 * config/arm/arm.c (arm_coproc_builtin_available): Add
7836 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
7837 (arm_coproc_ldc_stc_legitimate_address): New.
7838 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
7839 'qualifier_const_pointer'.
7840 (LDC_QUALIFIERS): Define to...
7841 (arm_ldc_qualifiers): ... this. New.
7842 (STC_QUALIFIERS): Define to...
7843 (arm_stc_qualifiers): ... this. New.
7844 * config/arm/arm-protos.h
7845 (arm_coproc_ldc_stc_legitimate_address): New.
7846 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
7847 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
7848 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
7849 stc2, stcl, stc2l): New.
7850 * config/arm/constraints.md (Uz): New.
7851 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
7852 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
7853 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
7854 VUNSPEC_STC2L): New.
7856 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7858 * config/arm/arm.md (<cdp>): New.
7859 * config/arm/arm.c (neon_const_bounds): Rename this ...
7860 (arm_const_bounds): ... this.
7861 (arm_coproc_builtin_available): New.
7862 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
7863 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
7864 (CDP_QUALIFIERS): Define to...
7865 (arm_cdp_qualifiers): ... this. New.
7867 (arm_expand_builtin_args): Add case for 6 arguments.
7868 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
7869 (arm_const_bounds): ... this.
7870 (arm_coproc_builtin_available): New.
7871 * config/arm/arm_acle.h (__arm_cdp): New.
7873 * config/arm/arm_acle_builtins.def (cdp): New.
7875 * config/arm/iterators.md (CDPI,CDP,cdp): New.
7876 * config/arm/neon.md: Rename all 'neon_const_bounds' to
7878 * config/arm/types.md (coproc): New.
7879 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
7880 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
7881 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
7882 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
7884 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7886 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
7887 (UBINOP_QUALIFIERS): New.
7889 (acle_builtin_data): New. Change comment.
7890 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
7891 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
7892 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
7893 arm_acle_builtins.def.
7894 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
7895 (arm_init_acle_builtins): New.
7896 (CRC32_BUILTIN): Remove.
7897 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
7898 crc32cb, crc32ch and crc32cw.
7899 (arm_init_crc32_builtins): Remove.
7900 (arm_init_builtins): Use arm_init_acle_builtins rather
7901 than arm_init_crc32_builtins.
7902 (arm_expand_acle_builtin): New.
7903 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
7904 * config/arm/arm_acle_builtins.def: New.
7906 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
7908 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
7909 (arm_builtin_datum): ... this.
7910 (arm_init_neon_builtin): Rename to ...
7911 (arm_init_builtin): ... this. Add a new parameters PREFIX
7912 and USE_SIG_IN_NAME.
7913 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
7914 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
7915 'arm_builtin_datum'.
7916 (arm_init_vfp_builtins): Likewise.
7917 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
7918 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
7919 (arm_expand_neon_args): Rename to ...
7920 (arm_expand_builtin_args): ... this. Rename builtin_arg
7921 enum values and differentiate between ARG_BUILTIN_MEMORY
7922 and ARG_BUILTIN_NEON_MEMORY.
7923 (arm_expand_neon_builtin_1): Rename to ...
7924 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
7925 values, arm_expand_builtin_args and add bool parameter NEON.
7926 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
7927 (arm_expand_vfp_builtin): Likewise.
7928 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
7930 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
7933 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
7934 * predict.c (tree_estimate_probability_bb): Reverse direction of
7935 polymorphic call predictor.
7937 2017-01-06 David Malcolm <dmalcolm@redhat.com>
7939 * passes.c (execute_one_pass): Split out pass-skipping logic into...
7940 (determine_pass_name_match): ...this new function and...
7941 (should_skip_pass_p): ...this new function.
7943 2017-01-06 Nathan Sidwell <nathan@acm.org>
7945 * ipa-visibility.c (function_and_variable_visibility): Reformat
7946 comments and long lines. Remove extrneous if.
7947 * symtab.c (symtab_node::make_decl_local): Fix code format.
7948 (symtab_node::set_section_for_node): Fix comment typo.
7950 2017-01-06 Martin Liska <mliska@suse.cz>
7953 * lra-constraints.c: Rename invariant to lra_invariant.
7954 * predict.c (set_even_probabilities): Initialize e to NULL.
7956 2017-01-05 Martin Sebor <msebor@redhat.com>
7958 PR tree-optimization/78910
7959 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
7960 (format_integer): Correct off-by-one error in the handling
7961 of precision with negative numbers in signed conversions..
7963 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
7965 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
7967 2017-01-05 Jakub Jelinek <jakub@redhat.com>
7969 PR tree-optimization/71016
7970 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
7971 factor_out_conditional_conversion. Formatting fix.
7972 (factor_out_conditional_conversion): Add cond_stmt argument.
7973 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
7974 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
7977 2017-01-05 David Malcolm <dmalcolm@redhat.com>
7979 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
7980 read-rtl-function.o, and selftest-rtl.o.
7981 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
7982 (selftest::aarch64_test_loading_full_dump): New function.
7983 (selftest::aarch64_run_selftests): New function.
7984 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
7985 selftest::aarch64_run_selftests.
7986 * config/i386/i386.c
7987 (selftest::ix86_test_loading_dump_fragment_1): New function.
7988 (selftest::ix86_test_loading_call_insn): New function.
7989 (selftest::ix86_test_loading_full_dump): New function.
7990 (selftest::ix86_test_loading_unspec): New function.
7991 (selftest::ix86_run_selftests): Call the new functions.
7992 * emit-rtl.c (maybe_set_max_label_num): New function.
7993 * emit-rtl.h (maybe_set_max_label_num): New decl.
7994 * function.c (instantiate_decls): Guard call to
7995 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
7996 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
7998 * gensupport.c (gen_reader::gen_reader): Pass "false"
7999 for new "compact" param of rtx_reader.
8000 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
8001 rather than an empty string for NULL strings.
8002 * read-md.c: Potentially include config.h rather than bconfig.h.
8003 Wrap include of errors.h with #ifdef GENERATOR_FILE.
8004 (have_error): New global, copied from errors.c.
8005 (md_reader::read_name): Rename to...
8006 (md_reader::read_name_1): ...this, adding "out_loc" param,
8007 and converting "missing name or number" to returning false, rather
8009 (md_reader::read_name): Reimplement in terms of read_name_1.
8010 (md_reader::read_name_or_nil): New function.
8011 (md_reader::read_string): Handle "(nil)" by returning NULL.
8012 (md_reader::md_reader): Add new param "compact".
8013 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
8014 (md_reader::read_file): New method.
8015 * read-md.h (md_reader::md_reader): Add new param "compact".
8016 (md_reader::read_file): New method.
8017 (md_reader::is_compact): New accessor.
8018 (md_reader::read_name): Convert return type from void to file_location.
8019 (md_reader::read_name_or_nil): New decl.
8020 (md_reader::read_name_1): New decl.
8021 (md_reader::m_compact): New field.
8022 (noop_reader::noop_reader): Pass "false" for new "compact" param
8024 (rtx_reader::rtx_reader): Add new "compact" param.
8025 (rtx_reader::read_rtx_operand): Make virtual and convert return
8026 type from void to rtx.
8027 (rtx_reader::read_until): New decl.
8028 (rtx_reader::handle_any_trailing_information): New virtual function.
8029 (rtx_reader::postprocess): New virtual function.
8030 (rtx_reader::finalize_string): New virtual function.
8031 (rtx_reader::m_in_call_function_usage): New field.
8032 (rtx_reader::m_reuse_rtx_by_id): New field.
8033 * read-rtl-function.c: New file.
8034 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
8035 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
8036 (selftest::verify_three_block_rtl_cfg): New decl.
8037 * read-rtl-function.h: New file.
8038 * read-rtl.c: Potentially include config.h rather than bconfig.h.
8039 For host, include function.h, memmodel.h, and emit-rtl.h.
8040 (one_time_initialization): New function.
8041 (struct compact_insn_name): New struct.
8042 (compact_insn_names): New array.
8043 (find_code): Handle insn codes in compact dumps.
8044 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
8045 (bind_subst_iter_and_attr): Likewise.
8046 (add_condition_to_string): Likewise.
8047 (add_condition_to_rtx): Likewise.
8048 (apply_attribute_uses): Likewise.
8049 (add_current_iterators): Likewise.
8050 (apply_iterators): Likewise.
8051 (initialize_iterators): Guard usage of apply_subst_iterator with
8052 #ifdef GENERATOR_FILE.
8053 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
8054 (md_reader::read_mapping): Likewise.
8055 (add_define_attr_for_define_subst): Likewise.
8056 (add_define_subst_attr): Likewise.
8057 (read_subst_mapping): Likewise.
8058 (check_code_iterator): Likewise.
8059 (rtx_reader::read_rtx): Likewise. Move one-time initialization
8061 (one_time_initialization): New function.
8062 (rtx_reader::read_until): New method.
8063 (read_flags): New function.
8064 (parse_reg_note_name): New function.
8065 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
8066 Handle reuse_rtx ids.
8067 Wrap iterator lookup within #ifdef GENERATOR_FILE.
8068 Add parsing support for RTL dumps, mirroring the special-cases in
8069 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
8070 values, and calling handle_any_trailing_information.
8071 (rtx_reader::read_rtx_operand): Convert return type from void
8072 to rtx, returning return_rtx. Handle case 'e'. Call
8073 finalize_string on XSTR and XTMPL fields.
8074 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
8075 "(nil)" values were omitted. Call the postprocess vfunc on the
8077 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
8078 class ctor. Initialize m_in_call_function_usage. Call
8079 one_time_initialization.
8080 * rtl-tests.c (selftest::test_uncond_jump): Call
8081 set_new_first_and_last_insn.
8082 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
8083 * selftest-rtl.c: New file.
8084 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
8085 (selftest::get_insn_by_uid): New decl.
8086 * selftest-run-tests.c (selftest::run_tests): Call
8087 read_rtl_function_c_tests.
8088 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
8089 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
8092 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
8094 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
8095 operands in a special way. Assert that pos+len <= mode precision.
8097 2017-01-05 Jakub Jelinek <jakub@redhat.com>
8099 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
8100 3 argument Alias with unlimited for the negative form.
8101 (fno-vect-cost-model): Removed.
8103 2017-01-05 Martin Liska <mliska@suse.cz>
8105 * hsa-gen.c (gen_hsa_divmod): New function.
8106 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
8108 2017-01-05 Martin Liska <mliska@suse.cz>
8111 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
8114 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8116 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
8117 small constant length operands.
8119 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8121 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
8122 between loop iterations.
8124 2017-01-05 Martin Liska <mliska@suse.cz>
8127 * gimplify.c (gimplify_decl_expr): Compare to
8128 asan_poisoned_variables instread of checking flags.
8129 (gimplify_target_expr): Likewise.
8130 (gimplify_expr): Likewise.
8131 (gimplify_function_tree): Conditionally initialize
8132 asan_poisoned_variables.
8134 2017-01-04 Jeff Law <law@redhat.com>
8136 PR tree-optimizatin/78812
8137 * rtl.h (contains_mem_rtx_p): Prototype.
8138 * ifcvt.c (containts_mem_rtx_p): Move from here to...
8139 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
8140 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
8141 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
8142 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
8144 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8146 * input.c (assert_char_at_range): Default-initialize actual_range.
8148 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8150 * df-scan.c (df_ref_create_structure): Make regno unsigned,
8151 to match the caller.
8153 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8155 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
8156 insns after final jump in test to emit dummy move.
8158 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8160 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
8161 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
8163 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
8165 * multiple_target.c (create_dispatcher_calls): Init e_next.
8166 * tree-ssa-loop-split.c (split_loop): Init border.
8167 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
8170 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
8175 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
8176 (altivec_register_operand): Do not return true if the operand
8177 contains a SUBREG mixing SImode and SFmode.
8178 (vsx_register_operand): Likewise.
8179 (vsx_reg_sfsubreg_ok): New predicate.
8180 (vfloat_operand): Do not return true if the operand contains a
8181 SUBREG mixing SImode and SFmode.
8182 (vint_operand): Likewise.
8183 (vlogical_operand): Likewise.
8184 (gpc_reg_operand): Likewise.
8185 (int_reg_operand): Likewise.
8186 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
8187 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
8188 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
8190 (rs6000_emit_move_si_sf_subreg): New helper function.
8191 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
8192 fixup SUBREGs involving SImode and SFmode.
8193 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
8194 numbers for the new peephole2 optimization.
8195 (peephole2 for SFmode unions): New peephole2 to optimize cases in
8196 the GLIBC math library that do AND/IOR/XOR operations on single
8197 precision floating point.
8198 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
8199 target macros to say whether we need to avoid SUBREGs mixing
8201 (TARGET_ALLOW_SF_SUBREG): Likewise.
8202 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
8203 (UNSPEC_SI_FROM_SF): Likewise.
8204 (iorxor): Change spacing.
8205 (and_ior_xor): New iterator for AND, IOR, and XOR.
8206 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
8207 (movdi_from_sf_zero_ext): Likewise.
8208 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
8209 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
8210 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
8211 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
8212 (fms<mode>4): Likewise.
8213 (fnma<mode>4): Likewise.
8214 (fnms<mode>4): Likewise.
8215 (nfma<mode>4): Likewise.
8216 (nfms<mode>4): Likewise.
8218 2017-01-04 Marek Polacek <polacek@redhat.com>
8221 * doc/invoke.texi: Document -Wpointer-compare.
8223 2017-01-04 Jakub Jelinek <jakub@redhat.com>
8225 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
8228 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
8229 descriptions for -gdwarf-5 and emit them as uleb128 instead of
8232 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
8235 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
8236 documentation of the powerpc_popcntb_ok attribute.
8237 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
8238 code to issue warning messages if a requested CPU configuration is
8239 not supported by the binary (assembler and loader) toolchain.
8240 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
8241 made to define a built-in function that has been disabled.
8242 (paired_init_builtins): Add assertion to prevent ICE if attempt is
8243 made to define a built-in function that has been disabled.
8244 (altivec_init_builtins): Add comment explaining why definition
8245 of the DST built-in functions is not preceded by an assertion
8246 check. Add assertions to prevent ICE if attempts are made to
8247 define an altivec predicate or an abs* built-in function that has
8249 (htm_init_builtins): Add comment explaining why definition of the
8250 htm built-in functions is not preceded by an assertion check.
8252 2017-01-04 Jeff Law <law@redhat.com>
8254 PR tree-optimizatin/67955
8255 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
8256 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
8257 the points-to solution does not include pt_null. Use DECL_PT_UID
8260 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
8262 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
8263 Use gen_int_mode instead of gen_lopwart for const_int operands.
8265 2017-01-04 Jakub Jelinek <jakub@redhat.com>
8267 PR tree-optimization/71563
8268 * match.pd: Simplify X << Y into X if Y is known to be 0 or
8269 out of range value - has low bits known to be zero.
8271 2017-01-04 Alan Modra <amodra@gmail.com>
8273 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
8274 * configure: Regenerate.
8275 * config.in: Regenerate.
8277 2017-01-04 Jakub Jelinek <jakub@redhat.com>
8280 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
8281 a substring of the message, but strcmp with the whole message. Ifdef
8282 ENABLE_NLS, translate the message first using dgettext.
8284 2017-01-03 Jeff Law <law@redhat.com>
8286 PR tree-optimizatin/78856
8287 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
8288 (mark_threaded_blocks): Remove code to truncate thread paths that
8289 cross multiple loop headers. Instead invalidate the cached loop
8290 iteration information and handle case of a thread path walking
8291 into an irreducible region.
8293 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
8296 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
8297 assertions. Add support for doing the signbit if the IEEE 128-bit
8298 floating point value is in a GPR.
8299 * config/rs6000/rs6000.md (Fsignbit): Delete.
8300 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
8301 Update the length attribute if the value is in a GPR.
8302 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
8303 the sign or zero extension instruction, since the value is always 0/1.
8304 (signbit<mode>2_dm2): Delete using <Fsignbit>.
8307 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
8308 extracting SImode to a GPR register so that we can generate a
8309 store, limit the vector to be in a traditional Altivec register
8310 for the vextuwrx instruction.
8312 2017-01-03 Ian Lance Taylor <iant@google.com>
8314 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
8316 2017-01-03 Martin Sebor <msebor@redhat.com>
8318 PR tree-optimization/78696
8319 * gimple-ssa-sprintf.c (format_floating): Correct handling of
8320 precision. Use MPFR for %f for greater fidelity. Correct handling
8322 (pass_sprintf_length::compute_format_length): Set width and precision
8323 specified by asrerisk to void_node for vararg functions.
8324 (try_substitute_return_value): Adjust dump output.
8326 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
8328 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
8330 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
8332 * doc/invoke.texi (SPARC options): Document -mlra as the default.
8333 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
8334 -mlra/-mno-lra was passed to the compiler.
8336 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
8338 PR rtl-optimization/65618
8339 * emit-rtl.c (try_split): Move initialization of "before" and
8340 "after" to just before the call to emit_insn_after_setloc.
8342 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
8344 * doc/md.texi (Standard Names): Remove reference to Java frontend.
8346 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
8348 * dwarf2out.c (gen_enumeration_type_die): When
8349 -gno-strict-dwarf, add a DW_AT_encoding attribute.
8351 2017-01-03 Jakub Jelinek <jakub@redhat.com>
8353 PR tree-optimization/78965
8354 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
8355 Change first argument from const call_info & to call_info &. For %n
8356 set info.nowrite to false.
8359 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
8360 possibly throwing calls.
8362 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
8363 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
8364 and fns handling, rather than in a separate case SSA_NAME.
8366 2017-01-02 Jeff Law <law@redhat.com>
8368 * config/darwin-driver.c (darwin_driver_init): Const-correctness
8369 fixes for first_period and second_period variables.
8371 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
8374 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
8375 (*insvqi_1): New insn pattern.
8376 (*insvqi_1_mem_rex64): Ditto.
8378 (*insvqi_3): Rename from *insvqi.
8380 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
8382 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
8384 * doc/cfg.texi (Edges): Remove reference to Java.
8385 (Maintaining the CFG): Ditto.
8387 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8390 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
8391 transparent aliases.
8393 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
8396 * predict.def (PRED_CALL): Update hitrate.
8397 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
8398 * predict.c (tree_estimate_probability_bb): Split CALL predictor
8399 into direct/indirect/polymorphic variants.
8401 2017-01-01 Jakub Jelinek <jakub@redhat.com>
8403 Update copyright years.
8405 * gcc.c (process_command): Update copyright notice dates.
8406 * gcov-dump.c (print_version): Ditto.
8407 * gcov.c (print_version): Ditto.
8408 * gcov-tool.c (print_version): Ditto.
8409 * gengtype.c (create_file): Ditto.
8410 * doc/cpp.texi: Bump @copying's copyright year.
8411 * doc/cppinternals.texi: Ditto.
8412 * doc/gcc.texi: Ditto.
8413 * doc/gccint.texi: Ditto.
8414 * doc/gcov.texi: Ditto.
8415 * doc/install.texi: Ditto.
8416 * doc/invoke.texi: Ditto.
8418 Copyright (C) 2017 Free Software Foundation, Inc.
8420 Copying and distribution of this file, with or without modification,
8421 are permitted in any medium without royalty provided the copyright
8422 notice and this notice are preserved.