1 2018-11-05 Richard Biener <rguenther@suse.de>
3 PR rtl-optimization/87852
4 * fwprop.c (use_killed_between): Only consider single-defs of the
5 use whose definition statement dominates the use.
7 2018-11-05 Martin Liska <mliska@suse.cz>
10 * doc/invoke.texi: Remove options that are
11 not disabled with -Os.
13 2018-11-05 Martin Liska <mliska@suse.cz>
16 * doc/extend.texi: Update constrain about the last argument
17 of __builtin_expect_with_probability.
19 2018-11-05 Martin Liska <mliska@suse.cz>
22 * predict.c (expr_expected_value_1): Verify
23 that last argument is a real constants and emit
26 2018-11-05 Martin Liska <mliska@suse.cz>
29 * ipa-profile.c (ipa_profile): Adjust hotness threshold
32 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
34 PR tree-optimization/86572
35 * builtins.c (c_strlen): Handle negative offsets in a safe way.
37 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
39 PR tree-optimization/87672
40 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
41 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
43 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
46 * cfgexpand.c (pass_expand::execute): Move the call to
47 finish_eh_generation in front of the call to expand_stack_alignment.
49 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
51 * common/config/i386/i386-common.c (processor_alias_table): Add
53 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
54 (case ${target}): Add znver2.
55 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
56 -march=native recognize znver2 processors.
57 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
58 * config/i386/i386.c (m_znver2): New definition.
59 (m_ZNVER): New definition.
60 (m_AMD_MULTIPLE): Includes m_znver2.
61 (processor_cost_table): Add znver2 entry.
62 (processor_target_table): Add znver2 entry.
63 (get_builtin_code_for_version): Set priority for
65 (processor_model): Add M_AMDFAM17H_ZNVER2.
66 (arch_names_table): Ditto.
67 (ix86_reassociation_width): Include znver2.
68 * config/i386/i386.h (TARGET_znver2): New definition.
69 (struct ix86_size_cost): Add TARGET_ZNVER2.
70 (enum processor_type): Add PROCESSOR_ZNVER2.
71 * config/i386/i386.md (define_attr "cpu"): Add znver2.
72 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
73 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
74 (ix86_adjust_cost): Add znver2.
75 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
76 * gcc/doc/extend.texi: Add details about znver2.
77 * gcc/doc/invoke.texi: Add details about znver2.
79 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
83 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
86 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
88 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
89 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
91 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
93 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
94 shrink-wrapping when -mtrack-speculation.
96 2018-11-02 Richard Biener <rguenther@suse.de>
98 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
99 (pop_cost_one_pair): Do not free pair.
100 (pop_best_coalesce): Likewise.
101 (create_coalesce_list): Initialize obstack.
102 (delete_coalesce_list): Free obstack.
103 (find_coalesce_pair): Obstack-allocate coalesce pairs.
104 (add_cost_one_coalesce): Likewise.
105 (struct live_track): Remove bitmap pointer indirections.
106 (new_live_track): Adjust.
107 (delete_live_track): Likewise.
108 (live_track_remove_partition): Likewise.
109 (live_track_add_partition): Likewise.
110 (live_track_live_p): Likewise.
111 (live_track_process_def): Likewise.
112 (live_track_clear_base_vars): Likewise.
114 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
116 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
117 * configure: Regenerate.
118 * config.in: Regenerate.
119 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
120 HAVE_AS_REGISTER_PSEUDO_OP guard.
121 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
122 !HAVE_AS_REGISTER_PSEUDO_OP support.
123 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
126 2018-11-02 Richard Biener <rguenther@suse.de>
128 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
131 2018-11-02 Richard Biener <rguenther@suse.de>
133 PR tree-optimization/87776
134 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
135 executable when iterating but running into rpo-vn-max-loop-depth
136 and not eliding the iteration.
138 2018-11-30 Jan Hubicka <jh@suse.cz>
140 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
143 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
145 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
147 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
148 Combine with rs6000_address_for_fpconvert.
149 (rs6000_address_for_fpconvert): Combine with
150 rs6000_force_indexed_or_indirect_mem.
151 (rs6000_expand_vector_init): Change function call from
152 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
153 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
154 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
155 (floatsi<mode>2_lfiwax_mem): Ditto.
156 (floatunssi<mode>2_lfiwzx): Ditto.
157 (floatunssi<mode>2_lfiwzx_mem): Ditto.
158 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
159 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
160 (fix_trunc<mode>si2_stfiwx): Ditto.
161 (fixuns_trunc<mode>si2_stfiwx): Ditto.
162 (float_<mode>si2_hw): Ditto.
163 (floatuns_<mode>si2_hw): Ditto.
164 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
165 (vsx_splat_<mode>): Ditto.
167 2018-11-01 Joseph Myers <joseph@codesourcery.com>
169 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
170 at end of assembler input text.
171 * configure: Regenerate.
173 2018-11-01 Jakub Jelinek <jakub@redhat.com>
175 PR tree-optimization/87826
176 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
177 negative or larger or equal to type's precision.
179 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
181 * opts.c (default_options_table): Do not enable
182 OPT_fdelayed_branch at -Og.
183 * doc/invoke.texi (-fdelayed-branch): Document it.
185 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
187 * optabs-libfuncs.c (build_libfunc_function_visibility):
188 New, split out from...
189 (build_libfunc_function): ... here.
190 (init_one_libfunc_visibility): New, split out from ...
191 (init_one_libfunc): ... here.
193 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
194 scratch register need not be early-clobber. Document the reason
195 why we cannot use ST<OP>.
197 2018-10-31 Joseph Myers <joseph@codesourcery.com>
200 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
201 line for second argument of AC_DEFINE_UNQUOTED.
202 * doc/install.texi (Tools/packages necessary for modifying GCC):
203 Update to autoconf 2.69 and automake 1.15.1.
204 * aclocal.m4, config.in, configure: Regenerate.
206 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
208 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
211 2018-10-31 Martin Liska <mliska@suse.cz>
214 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
215 Add new argument invalid_extension.
216 (aarch64_get_all_extension_candidates): New function.
217 (aarch64_rewrite_selected_cpu): Add NULL to function call.
218 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
220 (aarch64_get_all_extension_candidates): New function.
221 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
222 argument invalid_extension.
223 (aarch64_parse_cpu): Likewise.
224 (aarch64_print_hint_for_extensions): New function.
225 (aarch64_validate_mcpu): Provide hint about invalid extension.
226 (aarch64_validate_march): Likewise.
227 (aarch64_handle_attr_arch): Pass new argument.
228 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
229 (aarch64_handle_attr_isa_flags): Likewise.
231 2018-10-31 Richard Biener <rguenther@suse.de>
235 * tree-outof-ssa.c (insert_backedge_copies): Restrict
236 copy generation to useful cases. Place the copy before
237 the definition of the backedge value when possible.
239 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
241 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
242 * config/arc/arc.c (arc_active_insn): New function.
243 (check_store_cacheline_hazard): Likewise.
244 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
245 (arc_override_options): Disable delay slot scheduler for older
247 (arc_store_addr_hazard_p): New implementation, old one renamed to
249 (arc_store_addr_hazard_internal_p): Renamed.
250 (arc_reorg): Don't combine into brcc instructions which are part
251 of hardware hazard solution.
252 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
253 (tune_arc700): Likewise.
254 * config/arc/arc.opt (arc7xx): New tune value.
255 * config/arc/arc700.md: Improve A7 scheduler.
257 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
259 * config/arc/arc.c (arc_override_options): Remove
260 TARGET_COMPACT_CASESI.
261 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
262 (CASE_VECTOR_MODE): Likewise.
263 (CASE_VECTOR_PC_RELATIVE): Likewise.
264 (CASE_VECTOR_SHORTEN_MODE): Likewise.
265 (CASE_VECTOR_SHORTEN_MODE1): Delete.
266 (ADDR_VEC_ALIGN): Update.
267 (ASM_OUTPUT_CASE_LABEL): Undefine.
268 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
269 (TARGET_BI_BIH): Define.
270 (DEFAULT_BRANCH_INDEX): Likewise.
271 * config/arc/arc.md (casesi): Rework to accept BI/BIH
272 instructions, remove compact_casesi use case.
273 (casesi_compact_jump): Remove.
274 (casesi_dispatch): New pattern.
275 * config/arc/arc.opt: Add mbranch-index option. Deprecate
276 compact_casesi option.
277 * doc/invoke.texi: Document mbranch-index option.
279 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
281 * config/arc/arc.c (arc_get_tp): Remove function.
282 (arc_emit_call_tls_get_addr): Likewise.
283 (arc_call_tls_get_addr): New function.
284 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
285 * config/arc/arc.md (tls_load_tp_soft): Remove.
286 (tls_gd_get_addr): Likewise.
288 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
290 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
291 (mulsi3_600_lib): Remove pattern.
292 (umulsi3_highpart_600_lib_le): Likewise.
293 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
294 (umulsidi3): Remove call to umulsidi3_600_lib.
295 (umulsidi3_600_lib): Remove pattern.
296 (peephole2): Remove peephole using the above deprecated patterns.
298 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
301 * config/arm/arm.c (arm_option_check_internal): Disable the combined
302 use of -mslow-flash-data and -mword-relocations.
303 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
304 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
306 * doc/invoke.texi (-mword-relocations): Mention conflict with
308 (-mslow-flash-data): Reciprocally.
310 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
312 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
313 16-byte modes held in GP registers to use an even regno.
315 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
316 (aarch64_atomic_ldop_supported_p): Remove.
317 (aarch64_gen_atomic_ldop): Remove.
318 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
319 Fully expand LSE operations here.
320 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
321 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
322 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
323 and use ATOMIC_LDOP instead; use register_operand for the input;
324 drop the split and emit insns directly.
325 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
326 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
327 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
329 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
330 (aarch64_gen_atomic_ldop): Don't call it.
331 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
332 Use aarch64_reg_or_zero.
333 (aarch64_atomic_exchange<ALLI>): Likewise.
334 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
335 operand 0; use aarch64_reg_or_zero for input; merge ...
336 (@aarch64_atomic_swp<ALLI>): ... this and remove.
338 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
339 (aarch64_split_compare_and_swap): Use it.
340 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
341 test oldval against the proper predicate.
342 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
343 Use nonmemory_operand for expected.
344 (cas_short_expected_pred): New.
345 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
346 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
347 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
348 (aarch64_plushi_operand): New.
350 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
351 Force oldval into the rval register for TARGET_LSE; emit the compare
352 during initial expansion so that it may be deleted if unused.
353 (aarch64_gen_atomic_cas): Remove.
354 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
355 Change =&r to +r for operand 0; use match_dup for operand 2;
356 remove is_weak and mod_f operands as unused. Drop the split
358 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
359 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
360 (@aarch64_atomic_cas<GPI>): Similarly.
362 2018-10-31 Richard Biener <rguenther@suse.de>
364 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
367 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
369 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
370 * config/aarch64/aarch64.md: Include saphira.md
371 * config/aarch64/saphira.md: New file for pipeline description.
373 2018-10-30 Martin Sebor <msebor@redhat.com>
376 * gimple-ssa-sprintf.c (format_directive): Use %G to include
378 (sprintf_dom_walker::compute_format_length):
379 Avoid setting POSUNDER4K here.
380 (get_destination_size): Handle null argument values.
381 (get_user_idx_format): New function.
382 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
383 functions, including user-defined with attribute format printf.
384 Use %G to include inlining context.
387 2018-10-30 Jan Hubicka <jh@suse.cz>
389 * params.def (lto-partitions): Bump from 32 to 128.
391 2018-10-30 Jan Hubicka <jh@suse.cz>
394 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
396 (free_lang_data_in_type): Forward declare.
397 (fld_type_variant_equal_p): New function.
398 (fld_type_variant): New function
399 (fld_incomplete_types): New hash.
400 (fld_incomplete_type_of): New function
401 (fld_simplfied-type): New function.
402 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
403 (free_lang_data): Allocate and free fld_incomplete_type; update call
404 of free_lang_data_in_decl.
406 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
408 * gcov.c (output_lines): Remove duplicate line.
410 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
412 * config/rs6000/rs6000.md (bswapdi2): Force address into register
413 if not in indexed or indirect form.
414 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
415 (bswapdi2_store): Ditto.
416 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
418 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
419 Prototype for helper function.
421 2018-10-30 Martin Sebor <msebor@redhat.com>
423 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
424 (target, pragma GCC optimize, pragma GCC target): Ditto.
426 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
428 * doc/extend.texi: Fix prototype and description of
429 __builtin_expect_with_probability.
431 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
433 * cgraph.h (clone_function_name_1): Replaced by new
434 clone_function_name_numbered that takes name as string; for
435 privatize_symbol_name_1 use only.
436 (clone_function_name): Renamed to
437 clone_function_name_numbered to be explicit about numbering.
438 (clone_function_name): New two-argument function that does
439 not number its output.
440 (clone_function_name): New three-argument function that
441 takes a number to append to its output.
442 * cgraphclones.c (duplicate_thunk_for_node):
443 (clone_function_name_1): Renamed.
444 (clone_function_name_numbered): Two new functions.
445 (clone_function_name): Improved documentation.
446 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
447 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
448 * final.c (final_scan_insn_1): Use the new clone_function_name
450 * multiple_target.c (create_dispatcher_calls): Ditto.
451 (create_target_clone): Ditto.
452 * omp-expand.c (grid_expand_target_grid_body): Ditto.
453 * omp-low.c (create_omp_child_function_name): Ditto.
454 * omp-simd-clone.c (simd_clone_create): Ditto.
455 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
456 new clone_function_name without numbering.
458 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
460 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
461 Assert that the allocation size is not zero.
463 2018-10-30 Richard Biener <rguenther@suse.de>
465 PR tree-optimization/87800
466 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
467 non-induction or reduction PHIs.
469 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
471 * config/aarch64/falkor-tag-collision-avoidance.c
472 (execute_tag_collision_avoidance): Call df_note_add_problem.
474 2018-10-30 Martin Liska <mliska@suse.cz>
476 * doc/extend.texi: Fix typo in documentation
477 of __builtin_expect_with_probability.
479 2018-10-29 David Malcolm <dmalcolm@redhat.com>
482 * input.c (get_substring_ranges_for_loc): Detect if
483 linemap_resolve_location gives us a NULL map, and reject
486 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
488 * config.gcc (xstormy16-*-elf): Set tm_d_file.
490 2018-10-29 David Malcolm <dmalcolm@redhat.com>
491 Martin Sebor <msebor@redhat.com>
492 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
494 * doc/ux.texi (Quoting): New subsection, adapted from material at
495 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
496 MartinSebor and ManuelLopezIbanez.
497 (Fix-it hints): Note that fix-it hints shouldn't be marked for
500 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
503 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
506 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
508 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
510 2018-10-29 Paul Koning <ni1d@arrl.net>
512 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
514 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
516 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
517 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
518 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
519 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
520 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
521 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
522 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
523 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
524 Change 'vector' to '__vector'.
525 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
526 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
527 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
528 _mm_avg_pu16): Likewise. And, whitespace corrections.
530 2018-10-29 Richard Biener <rguenther@suse.de>
532 PR tree-optimization/87785
533 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
536 2018-10-29 Olivier Hainque <hainque@adacore.com>
538 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
540 2018-10-29 Olivier Hainque <hainque@adacore.com>
542 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
543 ports configuration macro, defaults to "gnu".
544 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
545 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
547 2018-10-29 Olivier Hainque <hainque@adacore.com>
549 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
551 2018-10-29 Richard Biener <rguenther@suse.de>
553 PR tree-optimization/87790
554 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
555 (vect_make_slp_decision): Adjust.
556 (vect_slp_analyze_bb_1): Likewise.
557 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
560 2018-10-29 Richard Biener <rguenther@suse.de>
562 PR tree-optimization/87785
563 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
565 (vect_build_slp_tree): Likewise.
566 (vect_gather_slp_loads): New function.
567 (vect_analyze_slp_instance): Gather loads separately from the
570 2018-10-29 Martin Liska <mliska@suse.cz>
572 * Makefile.in: Make dependency to json.o.
573 * doc/gcov.texi: Document new JSON format, remove
574 old intermediate format documentation.
575 * gcov.c (struct function_info): Come up with m_name and
577 (function_info::function_info): Initialize it.
578 (function_info::~function_info): Release it.
579 (main): Rename flag_intermediate_format to flag_json_format.
580 (print_usage): Describe --json-format.
581 (process_args): Set flag_json_format.
582 (output_intermediate_line): Remove.
583 (output_intermediate_json_line): Likewise.
584 (get_gcov_intermediate_filename): Return new extension
586 (output_intermediate_file): Implement JSON emission.
587 (output_json_intermediate_file): Implement JSON emission.
588 (generate_results): Use ::get_name for function name.
589 Handle JSON output file.
590 (read_graph_file): Use ::get_name instead of cplus_demangle.
591 (read_count_file): Likewise.
592 (solve_flow_graph): Likewise.
593 (add_line_counts): Likewise.
594 (accumulate_line_counts): Use new flag_json_format.
595 (output_function_details): Use ::get_name instead of cplus_demangle.
596 (output_lines): Likewise.
597 * json.cc (test_writing_literals): Add new tests.
598 * json.h (class literal): Add new boolean constructor.
600 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
602 PR rtl-optimization/87701
603 PR rtl-optimization/87780
604 * combine.c (make_more_copies): Rewrite.
606 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
608 * doc/generic.texi (ABSU_EXPR): Document.
609 * match.pd (absu(x)*absu(x) -> x*x): Handle.
610 (absu(absu(X)) -> absu(X)): Likewise.
611 (absu(-X) -> absu(X)): Likewise.
612 (absu(X) where X is nonnegative -> X): Likewise.
614 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
616 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
617 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
618 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
619 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
620 (s-tm-texi): Also check timestamp on d-target.def.
621 (generated_files): Add TM_D_H and d-target-hooks-def.h.
622 (build/genhooks.o): Also depend on D_TARGET_DEF.
623 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
625 * config/aarch64/aarch64-d.c: New file.
626 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
628 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
630 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
631 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
632 * config/arm/arm-d.c: New file.
633 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
634 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
635 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
636 * config/arm/t-arm (arm-d.o): New rule.
637 * config/default-d.c: New file.
638 * config/glibc-d.c: New file.
639 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
640 * config/i386/i386-d.c: New file.
641 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
642 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
643 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
644 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
645 * config/i386/t-i386 (i386-d.o): New rule.
646 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
647 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
648 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
649 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
650 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
651 * config/mips/mips-d.c: New file.
652 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
653 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
654 * config/mips/t-mips (mips-d.o): New rule.
655 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
656 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
657 * config/powerpcspe/powerpcspe-d.c: New file.
658 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
660 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
661 Support GNU D by using 0 as the language type.
662 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
663 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
664 * config/riscv/riscv-d.c: New file.
665 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
667 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
668 * config/riscv/t-riscv (riscv-d.o): New rule.
669 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
670 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
671 * config/rs6000/rs6000-d.c: New file.
672 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
674 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
675 Support GNU D by using 0 as the language type.
676 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
677 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
678 * config/s390/s390-d.c: New file.
679 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
680 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
681 * config/s390/t-s390 (s390-d.o): New rule.
682 * config/sparc/sparc-d.c: New file.
683 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
685 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
686 * config/sparc/t-sparc (sparc-d.o): New rule.
687 * config/t-glibc (glibc-d.o): New rule.
688 * configure: Regenerated.
689 * configure.ac (tm_d_file): New variable.
690 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
691 * doc/contrib.texi (Contributors): Add self for the D frontend.
692 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
693 * doc/install.texi (Configuration): Mention libphobos as an option for
694 --enable-shared. Mention d as an option for --enable-languages.
695 (Testing): Mention check-d as a target.
696 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
697 name suffixes. Mention d as a -x option.
698 * doc/sourcebuild.texi (Top Level): Mention libphobos.
699 * doc/standards.texi (Standards): Add section on D language.
700 * doc/tm.texi: Regenerated.
701 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
702 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
703 * dwarf2out.c (is_dlang): New function.
704 (gen_compile_unit_die): Use DW_LANG_D for D.
705 (declare_in_namespace): Return module die for D, instead of adding
706 extra declarations into the namespace.
707 (gen_namespace_die): Generate DW_TAG_module for D.
708 (gen_decl_die): Handle CONST_DECLSs for D.
709 (dwarf2out_decl): Likewise.
710 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
711 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
713 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
714 * genhooks.c: Include d/d-target.def.
716 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
719 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
720 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
721 (STACK_DYNAMIC_OFFSET): Likewise.
723 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
726 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
729 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
731 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
732 return type and other typos.
734 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
736 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
737 a shorter sequence with fewer branches.
738 (emit_final_str_compare_gpr): Ditto.
740 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
742 * config/rs6000/tmmintrin.h: New file.
743 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
745 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
747 * config/rs6000/mmintrin.h: Enable 32bit compilation.
748 * config/rs6000/xmmintrin.h: Likewise.
750 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
752 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
754 2018-10-26 Richard Biener <rguenther@suse.de>
756 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
758 (vect_mark_slp_stmts_relevant): Likewise.
759 (vect_detect_hybrid_slp_stmts): Likewise.
760 (vect_bb_slp_scalar_cost): Likewise.
761 (vect_remove_slp_scalar_calls): Likewise.
763 2018-10-26 Jan Hubicka <jh@suse.cz>
765 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
766 (warn_types_mismatch): Fix walk of DECL_NAME.
767 (odr_types_equivalent_p): Fix overactive assert.
769 2018-10-26 Richard Biener <rguenther@suse.de>
771 PR tree-optimization/87105
772 * tree-vectorizer.h (_slp_tree::refcnt): New member.
773 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
775 (vect_create_new_slp_node): Initialize refcnt to one.
777 (scalar_stmts_set_t, bst_fail): Remove.
778 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
779 (vect_build_slp_tree): Add bst_map argument and lookup
780 already created SLP nodes.
781 (vect_print_slp_tree): Handle a SLP graph, print SLP node
783 (vect_slp_rearrange_stmts): Handle a SLP graph.
784 (vect_analyze_slp_instance): Adjust and free SLP nodes from
785 the CSE map. Fix indenting.
786 (vect_schedule_slp_instance): Add short-cut.
788 2018-10-26 Martin Liska <mliska@suse.cz>
791 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
792 addr_expr and not with pointers.
794 2018-10-26 Jan Hubicka <jh@suse.cz>
796 * tree.c (free_lang_data_in_type): Only check main variants.
797 * ipa-devirt.c (warn_odr): Make static.
798 (types_same_for_odr): Drop strict variant.
799 (types_odr_comparable): Likewise.
800 (odr_or_derived_type_p): Look for main variants.
801 (odr_name_hasher::equal): Cleanup comment.
802 (odr_subtypes_equivalent): Add warn and warned arguments; check main
804 (type_variants_equivalent_p): break out from ...
805 (odr_types_equivalent): ... here; go for main variants where needed.
806 (warn_odr): ... here; turn static.
807 (warn_types_mismatch): Compare mangled names of main variants.
808 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
809 (type_with_linkage_p): Sanity check that we look at main variant.
810 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
811 * tree.h (types_same_for_odr): Drop strict argument.
813 2018-10-26 Richard Biener <rguenther@suse.de>
815 PR tree-optimization/87746
816 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
817 Simplify and fix WRT strided store groups with size not
818 equal to step in element count.
819 (vect_analyze_group_access_1): Dump the whole group.
821 2018-10-25 Carl Love <cel@us.ibm.com>
823 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
824 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
825 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
826 precicion entry for each overloaded builtin.
827 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
828 VSCEDPUO): Rename overloaded name.
829 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
830 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
831 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
832 define_expand for xscmpexqp instruction.
833 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
835 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
836 Jinsong Ji <jji@us.ibm.com>
838 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
839 function with vec_sl.
840 (_mm_slli_epi32): Likewise.
841 (_mm_slli_epi64): Likewise.
842 (_mm_srai_epi16): Replace deprecated function with vec_sra.
843 (_mm_srai_epi32): Likewise.
844 (_mm_srli_epi16): Replace deprecated function with vec_sr.
845 (_mm_srli_epi32): Likewise.
846 (_mm_srli_epi64): Likewise.
847 (_mm_sll_epi16): Replace deprecated function with vec_sl.
848 (_mm_sll_epi32): Likewise.
849 (_mm_sll_epi64): Likewise.
850 (_mm_sra_epi16): Replace deprecated function with vec_sra.
851 (_mm_sra_epi32): Likewise.
852 (_mm_srl_epi16): Replace deprecated function with vec_sr.
853 (_mm_srl_epi32): Likewise.
854 (_mm_srl_epi64): Likewise.
856 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
857 Jinsong Ji <jji@us.ibm.com>
859 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
860 comparison operators with vec_cmp* for compatibility due to
861 unfortunate history; clean up formatting and use types more
863 (_mm_sll_epi32): Likewise.
864 (_mm_sll_epi64): Likewise.
865 (_mm_srl_epi16): Likewise.
866 (_mm_srl_epi32): Likewise.
867 (_mm_srl_epi64): Likewise.
869 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
870 Jinsong Ji <jji@us.ibm.com>
872 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
873 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
874 __vector __bool int. Use vec_cmpgt in preference to deprecated
875 function vec_vcmpgtfp.
876 (_mm_max_ps): Likewise.
878 2018-10-25 Jeff Law <law@redhat.com>
880 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
883 2018-10-25 Martin Sebor <msebor@redhat.com>
885 * doc/extend.texi (aligned): Expand attribute description.
886 (Alignment): Rename section. Discuss function arguments.
888 2018-10-25 Jan Hubicka <jh@suse.cz>
890 * ipa-devirt.c (main_odr_variant): Remove.
891 (hash_odr_name, types_same_for_odr, types_odr_comparable,
892 odr_name_hasher::equal, odr_subtypes_equivalent_p):
893 Drop use of main_odr_variant.
894 (add_type_duplicate): Silence confused warnings on integer types.
895 (get_odr_type): Always look for main variant.
896 (register_odr_type): Simplify.
898 2018-10-25 Richard Biener <rguenther@suse.de>
900 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
901 Initialize ng to silence error with release checking bootstrap.
903 2018-10-25 Richard Biener <rguenther@suse.de>
905 * tree-if-conv.c: Include tree-ssa-sccvn.h.
906 (tree_if_conversion): Run CSE on the if-converted loop body.
908 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
910 * config/s390/constraints.md (ZL): New constraint.
911 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
913 * config/s390/s390.md (movdi_larl): Remove.
914 (movdi_64): Add the LARL alternative.
916 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
919 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
920 (RTX_CODE_HWINT_P): New macro.
921 (rtx_code_size): Use RTX_CODE_HWINT_P ().
923 2018-10-25 Jan Hubicka <jh@suse.cz>
925 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
928 2018-10-25 Richard Biener <rguenther@suse.de>
930 PR tree-optimization/87665
931 PR tree-optimization/87745
932 * tree-vectorizer.h (get_earlier_stmt): Remove.
933 (get_later_stmt): Pick up UID from the original non-pattern stmt.
935 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
937 * options.texi (Deprecated): Move list to Var section.
939 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
940 Jinsong Ji <jji@us.ibm.com>
942 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
943 __vector long to __vector long long.
944 (_mm_cvtpd_ps): Likewise.
945 (_mm_cvttpd_epi32): Likewise.
946 (_mm_cvtpi32_pd): Likewise.
947 (_mm_unpackhi_epi64): Likewise.
948 (_mm_unpacklo_epi64): Likewise.
950 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
952 PR rtl-optimization/87720
953 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
955 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
957 * gimple-ssa-isolate-paths.c
958 (find_implicit_erroneous_behavior): Do not change code if the
959 pass is running for warnings only.
960 (find_explicit_erroneous_behavior): Likewise.
962 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
964 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
965 Define as rs6000_mangle_decl_assembler_name.
966 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
967 long double to IEEE long double, switch the names of the long
968 double built-in functions to be <func>f128 instead of <func>l.
970 2018-10-24 Martin Sebor <msebor@redhat.com>
972 * doc/extend.texi (nonnull): List no-argument form. Reference
973 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
975 2018-10-24 Richard Biener <rguenther@suse.de>
977 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
979 2018-10-24 Martin Liska <mliska@suse.cz>
981 PR tree-optimization/84436
982 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
984 (switch_conversion::contains_linear_function_p): New.
985 (switch_conversion::build_one_array): Support linear
986 transformation on input.
987 * tree-switch-conversion.h (struct switch_conversion): Add
988 contains_linear_function_p declaration.
990 2018-10-24 Richard Biener <rguenther@suse.de>
992 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
993 if its argument is CONSTANT_CLASS_P.
995 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
997 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
998 it is wrong for forward declarations.
1000 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
1002 * config/s390/s390.c (s390_check_qrst_address): Add the missing
1003 SYMBOL_REF_P () check.
1005 2018-10-24 Richard Biener <rguenther@suse.de>
1007 PR tree-optimization/87105
1008 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
1009 dump classification.
1010 (vect_analyze_data_ref_accesses): Handle duplicate loads and
1011 stores by splitting the affected group after the fact.
1012 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
1013 fail the SLP build because of size constraints.
1015 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1017 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
1018 * configure: Regenerate.
1019 * config.in: Regenerate.
1020 * varasm.c (mergeable_string_section): Use readonly_data_section
1021 if linker doesn't support SHF_MERGE with alignment > 8.
1022 (mergeable_constant_section): Likewise.
1024 2018-10-24 Richard Biener <rguenther@suse.de>
1026 PR tree-optimization/84013
1027 * tree-ssa-structalias.c (struct msdi_data): New struct for
1028 marshalling data to walk_stmt_load_store_ops.
1029 (maybe_set_dependence_info): Refactor as callback for
1030 walk_stmt_load_store_ops.
1031 (compute_dependence_clique): Set restrict info on all stmt kinds.
1033 2018-10-24 Martin Liska <mliska@suse.cz>
1035 * cgraph.c (cgraph_node::dump):
1036 Remove reduntant dumps and make tp_first_run dump more compact.
1038 2018-10-24 Richard Biener <rguenther@suse.de>
1040 PR tree-optimization/87665
1041 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
1044 2018-10-12 Jeff Law <law@redhat.com>
1046 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
1049 2018-10-23 Richard Biener <rguenther@suse.de>
1051 * tree-vrp.c (add_assert_info): Guard dump_printf with
1053 * gimple-ssa-evrp-analyze.c
1054 (evrp_range_analyzer::record_ranges_from_incoming_edge):
1055 Use value_range::ignore_equivs_equal_p.
1057 2018-10-23 Richard Biener <rguenther@suse.de>
1059 PR tree-optimization/87105
1060 PR tree-optimization/87608
1061 * passes.def (pass_all_early_optimizations): Add early phi-opt
1063 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
1064 addition to debug stmts.
1065 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
1066 and abs replacement early.
1067 * tree-cfg.c (gimple_empty_block_p): Likewise.
1069 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
1072 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
1073 specified to configure.
1074 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
1076 2018-10-23 Richard Biener <rguenther@suse.de>
1078 PR tree-optimization/87700
1079 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
1081 2018-10-23 Jakub Jelinek <jakub@redhat.com>
1084 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
1085 second argument from __mmask16 to __mmask8.
1086 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
1087 _mm_mask_packs_epi32): Likewise.
1088 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
1090 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
1092 2018-10-23 Richard Biener <rguenther@suse.de>
1094 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
1096 2018-10-23 Richard Biener <rguenther@suse.de>
1098 PR tree-optimization/86144
1099 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
1100 over simd attribute.
1102 2018-10-23 Richard Biener <rguenther@suse.de>
1104 PR tree-optimization/87693
1105 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
1106 the case we do not find the taken edge.
1108 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
1109 Jinsong Ji <jji@us.ibm.com>
1111 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
1112 (_mm_store_pd): Use unaligned vector type for pointer cast.
1113 (_mm_maskmoveu_si128): Likewise.
1114 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
1115 (_mm_store_ps): Use unaligned vector type for pointer cast.
1117 2018-10-22 Paul Koning <ni1d@arrl.net>
1119 * symtab.c (symtab_node::increase_alignment): Correct max
1122 2018-10-22 Yury Gribov <tetra2005@gmail.com>
1124 PR tree-optimization/87633
1125 * match.pd: Do not generate unordered integer comparisons.
1127 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
1129 PR rtl-optimization/87600
1130 * combine.c: Add include of expr.h.
1131 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
1132 register to a pseudo.
1133 (make_more_copies): New function, add a copy to a new pseudo after
1134 the moves from hard registers into pseudos.
1135 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
1136 later. Call make_more_copies.
1138 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
1140 * lra-constraints.c (process_alt_operands): New local array,
1141 matching_early_clobber. Check matching_early_clobber before
1142 decrementing reject, and set matching_early_clobber after.
1144 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
1147 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
1148 call output_addr_const and hope for the best.
1150 2018-10-22 Richard Biener <rguenther@suse.de>
1152 * gimple-ssa-evrp-analyze.c
1153 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
1154 smarter about what ranges to use.
1155 * tree-vrp.c (add_assert_info): Dump here.
1156 (register_edge_assert_for_2): Instead of here at multiple but
1159 * gcc.dg/tree-ssa/evrp12.c: New testcase.
1160 * gcc.dg/predict-6.c: Adjust.
1161 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
1162 * gcc.dg/tree-ssa/vrp02.c: Likewise.
1163 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
1165 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
1166 Richard Biener <rguenther@suse.de>
1168 * bitmap.h: Update data structure documentation, including a
1169 description of bitmap views as either linked-lists or splay trees.
1170 (struct bitmap_element_def): Update comments for splay tree bitmaps.
1171 (struct bitmap_head_def): Likewise.
1172 (bitmap_list_view, bitmap_tree_view): New prototypes.
1173 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
1175 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
1176 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
1177 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
1178 released bitmap element here.
1179 (bitmap_element_free): Remove.
1180 (bitmap_elt_clear_from): Work on splay tree bitmaps.
1181 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
1182 this function similar ones such that linked-list bitmap implementation
1183 functions are grouped.
1184 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
1185 and moved for grouping.
1186 (bitmap_list_insert_element_after): Renamed from
1187 bitmap_elt_insert_after, and moved for grouping.
1188 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
1189 (bitmap_tree_link_left, bitmap_tree_link_right,
1190 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
1191 bitmap_tree_link_element, bitmap_tree_unlink_element,
1192 bitmap_tree_find_element): New functions for splay-tree bitmap
1194 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
1195 Renamed and moved, see above entries.
1196 (bitmap_tree_listify_from): New function to convert part of a splay
1197 tree bitmap to a linked-list bitmap.
1198 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
1199 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
1200 (bitmap_find_bit): Remove.
1201 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
1202 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
1203 Handle splay tree bitmaps.
1204 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
1205 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
1206 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
1207 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
1208 bitmap_intersect_compl_p, bitmap_ior_and_compl,
1209 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
1210 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
1211 corresponding changes to use linked-list specific bitmap_element
1212 manipulation functions as applicable for efficiency.
1213 (bitmap_tree_to_vec): New function.
1214 (debug_bitmap_elt_file): New function split out from ...
1215 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
1216 (bitmap_print): Likewise.
1218 PR tree-optimization/63155
1219 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
1221 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
1224 2018-10-22 Martin Liska <mliska@suse.cz>
1226 PR tree-optimization/87686
1228 2018-08-29 Martin Liska <mliska@suse.cz>
1230 * tree-switch-conversion.c (switch_conversion::expand):
1231 Strenghten assumption about gswitch statements.
1233 2018-10-22 Martin Liska <mliska@suse.cz>
1235 * ipa-icf.c (sem_item::compare_attributes): Remove.
1236 (sem_item::compare_referenced_symbol_properties): Use
1237 attribute_list_equal instead.
1238 (sem_function::equals_wpa): Likewise.
1239 * ipa-icf.h: Remove compare_attributes.
1241 2018-10-22 Richard Biener <rguenther@suse.de>
1244 * mem-stats.h (mem_usage::operator==): Fix pasto.
1246 2018-10-22 Richard Biener <rguenther@suse.de>
1248 PR tree-optimization/87640
1249 * tree-vrp.c (set_value_range_with_overflow): Decompose
1251 (extract_range_from_binary_expr_1): Adjust.
1253 2018-10-22 Martin Jambor <mjambor@suse.cz>
1255 * tree-eh.h (stmt_could_throw_p): Add function parameter.
1256 (stmt_can_throw_external): Likewise.
1257 (stmt_can_throw_internal): Likewise.
1258 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
1259 (lower_eh_constructs_2): Likewise.
1260 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
1261 (stmt_can_throw_external): Likewise.
1262 (stmt_can_throw_internal): Likewise.
1263 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
1264 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
1265 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
1266 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
1267 (pass_lower_eh_dispatch::execute): Pass cfun to
1268 stmt_can_throw_external.
1269 (cleanup_empty_eh): Likewise.
1270 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
1271 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
1272 stmt_can_throw_external instead of pushing it to cfun.
1273 (symbol_table::create_edge): Likewise.
1274 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
1275 stmt_can_throw_internal.
1276 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
1277 to stmt_could_throw_p.
1278 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
1279 stmt_can_throw_internal.
1280 (pass_store_merging::execute): Likewise.
1281 * gimple-ssa-strength-reduction.c
1282 (find_candidates_dom_walker::before_dom_children): Pass cfun to
1284 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
1285 stmt_can_throw_internal.
1286 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
1287 to stmt_can_throw_external.
1288 (check_stmt): Pass cfun to stmt_could_throw_p.
1289 (check_stmt): Pass cfun to stmt_can_throw_external.
1290 (pass_nothrow::execute): Likewise.
1291 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
1292 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
1293 stmt_can_throw_internal.
1294 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
1295 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
1296 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
1297 * tree-complex.c (expand_complex_libcall): Pass cfun to
1298 stmt_could_throw_p and to stmt_can_throw_internal.
1299 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
1300 * tree-inline.c (copy_edges_for_bb): Likewise.
1301 (maybe_move_debug_stmts_to_successors): Likewise.
1302 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
1304 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
1305 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
1306 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
1307 stmt_can_throw_internal.
1308 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
1309 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
1311 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
1312 stmt_can_throw_internal.
1313 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
1314 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
1316 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
1317 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
1318 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1319 (convert_mult_to_fma_1): Likewise.
1320 (convert_to_divmod): Likewise.
1321 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
1322 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
1323 * tree-ssa-propagate.c
1324 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
1325 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
1326 (maybe_optimize_range_tests): Likewise.
1327 (linearize_expr_tree): Likewise.
1328 (reassociate_bb): Likewise.
1329 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
1330 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
1331 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
1332 (handle_char_store): Likewise.
1333 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
1334 stmt_can_throw_internal.
1335 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
1337 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
1338 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
1339 (vectorizable_simd_clone_call): Likewise.
1340 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
1341 (gimple_stringop_fixed_value): Likewise.
1343 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
1345 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
1346 literal pool references.
1347 (s390_check_qrst_address): Adapt to the new behavior of
1348 s390_loadrelative_operand_p ().
1350 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1353 * config/i386/sse.md (*andnot<mode>3_bcst): New.
1355 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1358 * config/i386/sse.md (*<code><mode>3_bcst): New.
1360 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
1363 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
1364 V4DI, V16SI and V8DI.
1365 (*sub<mode>3<mask_name>_bcst): New.
1366 (*add<mode>3<mask_name>_bcst): Likewise.
1368 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
1369 Jinsong Ji <jji@us.ibm.com>
1371 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
1372 __m64 with __vector unsigned long long for compatibility.
1373 (_mm_movemask_epi8): Likewise.
1374 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
1375 (_mm_cvttps_pi32): Likewise.
1376 (_mm_cvtpi32_ps): Likewise.
1377 (_mm_cvtps_pi16): Likewise.
1378 (_mm_loadh_pi): Likewise.
1379 (_mm_storeh_pi): Likewise.
1380 (_mm_movehl_ps): Likewise.
1381 (_mm_movelh_ps): Likewise.
1382 (_mm_loadl_pi): Likewise.
1383 (_mm_storel_pi): Likewise.
1384 (_mm_movemask_ps): Likewise.
1385 (_mm_shuffle_pi16): Likewise.
1387 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1390 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
1391 __builtin_ia32_vfnmsubpd512_mask.
1392 (_mm512_mask_fnmsub_round_pd): Likewise.
1393 (_mm512_fnmsub_pd): Likewise.
1394 (_mm512_mask_fnmsub_pd): Likewise.
1395 (_mm512_maskz_fnmsub_round_pd): Use
1396 __builtin_ia32_vfnmsubpd512_maskz.
1397 (_mm512_maskz_fnmsub_pd): Likewise.
1398 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
1399 (_mm512_mask_fnmsub_round_ps): Likewise.
1400 (_mm512_fnmsub_ps): Likewise.
1401 (_mm512_mask_fnmsub_ps): Likewise.
1402 (_mm512_maskz_fnmsub_round_ps): Use
1403 __builtin_ia32_vfnmsubps512_maskz.
1404 (_mm512_maskz_fnmsub_ps): Likewise.
1405 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
1406 __builtin_ia32_vfnmsubpd256_mask.
1407 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
1408 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1409 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
1410 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
1411 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
1412 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
1413 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
1414 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
1415 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
1416 __builtin_ia32_vfnmsubpd.
1417 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
1418 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
1419 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
1420 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
1421 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
1422 * config/i386/i386-builtin.def: Add
1423 __builtin_ia32_vfnmsubpd256_mask,
1424 __builtin_ia32_vfnmsubpd256_maskz,
1425 __builtin_ia32_vfnmsubpd128_mask,
1426 __builtin_ia32_vfnmsubpd128_maskz,
1427 __builtin_ia32_vfnmsubps256_mask,
1428 __builtin_ia32_vfnmsubps256_maskz,
1429 __builtin_ia32_vfnmsubps128_mask,
1430 __builtin_ia32_vfnmsubps128_maskz,
1431 __builtin_ia32_vfnmsubpd512_mask,
1432 __builtin_ia32_vfnmsubpd512_maskz,
1433 __builtin_ia32_vfnmsubps512_mask,
1434 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
1435 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
1436 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
1437 __builtin_ia32_vfnmsubpd256.
1438 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
1439 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
1440 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
1442 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
1444 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
1446 (fmai_vmfnmsub_<mode><round_name>): Likewise.
1448 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1451 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
1452 __builtin_ia32_vfnmaddpd512_mask.
1453 (_mm512_mask_fnmadd_round_pd): Likewise.
1454 (_mm512_fnmadd_pd): Likewise.
1455 (_mm512_mask_fnmadd_pd): Likewise.
1456 (_mm512_maskz_fnmadd_round_pd): Use
1457 __builtin_ia32_vfnmaddpd512_maskz.
1458 (_mm512_maskz_fnmadd_pd): Likewise.
1459 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
1460 (_mm512_mask_fnmadd_round_ps): Likewise.
1461 (_mm512_fnmadd_ps): Likewise.
1462 (_mm512_mask_fnmadd_ps): Likewise.
1463 (_mm512_maskz_fnmadd_round_ps): Use
1464 __builtin_ia32_vfnmaddps512_maskz.
1465 (_mm512_maskz_fnmadd_ps): Likewise.
1466 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
1467 __builtin_ia32_vfnmaddpd256_mask.
1468 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
1469 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
1470 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
1471 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1472 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
1473 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
1474 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
1475 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
1476 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
1477 __builtin_ia32_vfnmaddpd.
1478 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
1479 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
1480 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
1481 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
1482 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
1483 * config/i386/i386-builtin.def: Add
1484 __builtin_ia32_vfnmaddpd256_mask,
1485 __builtin_ia32_vfnmaddpd256_maskz,
1486 __builtin_ia32_vfnmaddpd128_mask,
1487 __builtin_ia32_vfnmaddpd128_maskz,
1488 __builtin_ia32_vfnmaddps256_mask,
1489 __builtin_ia32_vfnmaddps256_maskz,
1490 __builtin_ia32_vfnmaddps128_mask,
1491 __builtin_ia32_vfnmaddps128_maskz,
1492 __builtin_ia32_vfnmaddpd512_mask,
1493 __builtin_ia32_vfnmaddpd512_maskz,
1494 __builtin_ia32_vfnmaddps512_mask,
1495 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
1496 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
1497 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
1498 __builtin_ia32_vfnmaddpd256.
1499 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
1500 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
1501 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
1503 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
1505 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
1507 (fmai_vmfnmadd_<mode><round_name>): Likewise.
1509 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1512 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
1513 __builtin_ia32_vfmsubpd512_mask.
1514 (_mm512_mask_fmsub_round_pd): Likewise.
1515 (_mm512_fmsub_pd): Likewise.
1516 (_mm512_mask_fmsub_pd): Likewise.
1517 (_mm512_maskz_fmsub_round_pd): Use
1518 __builtin_ia32_vfmsubpd512_maskz.
1519 (_mm512_maskz_fmsub_pd): Likewise.
1520 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
1521 (_mm512_mask_fmsub_round_ps): Likewise.
1522 (_mm512_fmsub_ps): Likewise.
1523 (_mm512_mask_fmsub_ps): Likewise.
1524 (_mm512_maskz_fmsub_round_ps): Use
1525 __builtin_ia32_vfmsubps512_maskz.
1526 (_mm512_maskz_fmsub_ps): Likewise.
1527 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
1528 __builtin_ia32_vfmsubpd256_mask.
1529 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
1530 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
1531 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
1532 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1533 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
1534 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
1535 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
1536 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
1537 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
1538 __builtin_ia32_vfmsubpd.
1539 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
1540 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
1541 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
1542 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
1543 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
1544 * config/i386/i386-builtin.def: Add
1545 __builtin_ia32_vfmsubpd256_mask,
1546 __builtin_ia32_vfmsubpd256_maskz,
1547 __builtin_ia32_vfmsubpd128_mask,
1548 __builtin_ia32_vfmsubpd128_maskz,
1549 __builtin_ia32_vfmsubps256_mask,
1550 __builtin_ia32_vfmsubps256_maskz,
1551 __builtin_ia32_vfmsubps128_mask,
1552 __builtin_ia32_vfmsubps128_maskz,
1553 __builtin_ia32_vfmsubpd512_mask,
1554 __builtin_ia32_vfmsubpd512_maskz,
1555 __builtin_ia32_vfmsubps512_mask,
1556 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
1557 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
1558 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
1559 __builtin_ia32_vfmsubpd256.
1560 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
1561 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
1562 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
1564 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
1566 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
1568 (fmai_vmfmsub_<mode><round_name>): Likewise.
1570 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1572 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
1573 Remove plus. Renamed to ...
1574 (*sub<mode>3<mask_name>_bcst): This.
1575 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
1576 (*add<mode>3<mask_name>_bcst): This.
1578 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1581 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
1583 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
1586 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
1587 (_mm_or_epi32): Likewise.
1588 (_mm256_xor_epi32): Likewise.
1589 (_mm_xor_epi32): Likewise.
1590 (_mm256_or_epi64): Likewise.
1591 (_mm_or_epi64): Likewise.
1592 (_mm256_xor_epi64): Likewise.
1593 (_mm_xor_epi64): Likewise.
1595 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
1598 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
1600 2018-10-20 Jakub Jelinek <jakub@redhat.com>
1603 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
1605 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
1607 * doc/ux.texi: Move @section directly after @node.
1609 2018-10-19 Jakub Jelinek <jakub@redhat.com>
1613 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
1614 depend closely nested inside of loop with ordered clause with
1617 2018-10-19 David Malcolm <dmalcolm@redhat.com>
1619 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
1620 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
1621 * doc/ux.texi: New file.
1623 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
1625 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
1626 be the first CR field allocated.
1628 2018-10-19 Richard Biener <rguenther@suse.de>
1631 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
1632 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
1634 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1637 * config/i386/sse.md
1638 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
1639 (*add<mode>3<mask_name>_bcst_2): Likewise.
1641 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
1643 * config/i386/sse.md
1644 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1645 Replace nonimmediate_operand with register_operand.
1646 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1648 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1651 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
1653 PR rtl-optimization/87596
1654 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
1655 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
1656 for instructions in FROM..TO range.
1658 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
1660 * cfgexpand.c (expand_one_var): Use specific wording in error message
1661 for non-local frame variables.
1662 * stor-layout.c (layout_decl): Do not issue a warning for them.
1664 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
1666 * haifa-sched.c (priority): Add force_recompute parameter.
1667 (apply_replacement): Call priority () with force_recompute = true.
1668 (restore_pattern): Likewise.
1670 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1672 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
1673 HOST_BITS_PER_WIDE_INT.
1674 (test_vector_ops_duplicate): Likewise.
1676 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1679 * config/i386/sse.md (VF_AVX512): New.
1680 (avx512bcst): Likewise.
1681 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
1683 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
1685 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
1688 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
1690 * doc/invoke.texi (-dumpversion): Improve grammar.
1691 (-dumpfullversion): Make more consistent with -dumpversion.
1693 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
1695 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
1696 Set JUMP_LABEL to the jump insn.
1697 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
1698 Predict emitted jump and add label to jump insn.
1700 2018-10-18 David Malcolm <dmalcolm@redhat.com>
1702 PR tree-optimization/87562
1703 * input.c (get_substring_ranges_for_loc): Use
1704 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
1705 getting the linemap for the endpoint. Verify that it's either
1706 in the same linemap as the start point's spelling location, or
1707 at least in the same file.
1709 2018-10-18 Richard Biener <rguenther@suse.de>
1711 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
1712 feed width-specific load/store costs through ix86_vec_cost.
1713 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
1714 (k8_cost): Likewise.
1715 (bdver_cost): Likewise.
1716 (znver1_cost): Likewise.
1717 (btver1_cost): Likewise.
1718 (btver2_cost): Likewise.
1720 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1722 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
1723 to simplify subreg of vec_merge.
1725 2018-10-18 Richard Biener <rguenther@suse.de>
1727 * config/i386/i386.c: Fix costing of vector FMA.
1729 2018-10-18 Richard Biener <rguenther@suse.de>
1731 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
1733 (ix86_builtin_vectorization_cost): For vec_construct properly
1734 cost insertion into SSE regs.
1735 (...): Adjust calls to ix86_vec_cost.
1737 2018-10-18 Richard Biener <rguenther@suse.de>
1741 2018-02-07 Richard Biener <rguenther@suse.de>
1743 PR tree-optimization/84204
1744 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
1747 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
1750 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
1752 (test_vector_ops_duplicate): Add test for a scalar subreg of a
1753 VEC_MERGE of a VEC_DUPLICATE.
1755 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1757 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
1758 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
1759 * doc/standards.texi (C Language): Document C2X.
1760 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
1761 (rl78_option_override): Handle "GNU C2X" language name.
1763 2018-10-17 Joseph Myers <joseph@codesourcery.com>
1765 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
1766 Document C17 as published in 2018.
1768 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
1771 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
1772 bail out if both sides do not have the same storage order.
1774 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1776 * bitmap.c (bitmap_head::dump): New.
1777 * bitmap.h (bitmap_head): Add dump().
1778 * gimple-ssa-evrp-analyze.c
1779 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
1780 (evrp_range_analyzer::set_ssa_range_info): Same.
1781 (evrp_range_analyzer::record_ranges_from_phis): Same.
1782 (evrp_range_analyzer::record_ranges_from_stmt): Same.
1783 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
1784 * gimple-ssa-sprintf.c (get_int_range): Same.
1785 (format_integer): Same.
1786 (sprintf_dom_walker::handle_gimple_call): Same.
1787 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
1788 (ipcp_vr_lattice::top_p): Same.
1789 (ipcp_vr_lattice::bottom_p): Same.
1790 (ipcp_vr_lattice::set_to_bottom): Same.
1791 (ipa_vr_operation_and_type_effects): Same.
1792 (propagate_vr_across_jump_function): Same.
1793 (ipcp_store_vr_results): Same.
1794 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
1795 (ipa_print_node_jump_functions_for_edge): Same.
1796 (ipa_get_value_range): Same.
1797 (ipa_compute_jump_functions_for_edge): Same.
1798 (ipa_write_jump_function): Same.
1799 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
1800 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
1802 * vr-values.c (set_value_range_to_nonnegative): Same.
1803 (set_value_range_to_truthvalue): Same.
1804 (vr_values::get_value_range): Same.
1805 (vr_values::set_defs_to_varying): Same.
1806 (vr_values::update_value_range): Same.
1807 (symbolic_range_based_on_p): Same.
1808 (vr_values::op_with_boolean_value_range_p): Same.
1809 (vr_values::extract_range_for_var_from_comparison_expr): Same.
1810 (vr_values::extract_range_from_ssa_name): Same.
1811 (vr_values::extract_range_from_binary_expr): Same.
1812 (vr_values::extract_range_from_unary_expr): Same.
1813 (vr_values::extract_range_from_cond_expr): Same.
1814 (vr_values::extract_range_from_comparison): Same.
1815 (vr_values::check_for_binary_op_overflow): Same.
1816 (vr_values::extract_range_basic): Same.
1817 (vr_values::extract_range_from_assignment): Same.
1818 (compare_ranges): Same.
1819 (compare_range_with_value): Same.
1820 (vr_values::adjust_range_with_scev): Same.
1821 (vrp_valueize): Same.
1822 (vrp_valueize_1): Same.
1823 (vr_values::get_vr_for_comparison): Same.
1824 (vr_values::compare_name_with_value): Same.
1825 (vr_values::compare_names): Same.
1826 (vr_values::vrp_evaluate_conditional): Same.
1827 (find_case_label_ranges): Same.
1828 (vr_values::vrp_visit_switch_stmt): Same.
1829 (vr_values::extract_range_from_phi_node): Same.
1830 (vr_values::simplify_div_or_mod_using_ranges): Same.
1831 (vr_values::simplify_bit_ops_using_ranges): Same.
1832 (test_for_singularity): Same.
1833 (range_fits_type_p): Same.
1834 (vr_values::simplify_cond_using_ranges_1): Same.
1835 (vr_values::simplify_switch_using_ranges): Same.
1836 (vr_values::simplify_float_conversion_using_ranges): Same.
1837 (vr_values::two_valued_val_range_p): Same.
1838 (vr_values::add_equivalence): Move to value_range::equiv_add.
1839 * vr-values.h (vr_values::add_equivalence): Remove.
1840 (VR_INITIALIZER): Remove.
1841 * tree-vrp.c (value_range::set): New.
1842 (value_range::equiv_add): New.
1843 (value_range::value_range): New.
1844 (value_range::deep_copy): New.
1845 (value_range::check): New.
1846 (value_range::equal_p): New.
1847 (value_range::ignore_equivs_equal_p): New.
1848 (value_range::operator==): New.
1849 (value_range::operator!=): New.
1850 (value_range::symbolic_p): New.
1851 (value_range::numeric_p): New.
1852 (value_range::set_undefined): New.
1853 (value_range::set_varying): New.
1854 (value_range::may_contain_p): New.
1855 (value_range::equiv_clear): New.
1856 (value_range::singleton_p): New.
1857 (value_range::intersect): New.
1858 (value_range::dump): New.
1859 (value_range::set_and_canonicalize): New.
1860 (set_value_range): Adjust for value_range API.
1861 (set_value_range_to_undefined): Same.
1862 (set_value_range_to_varying): Same.
1863 (set_and_canonicalize_value_range): Same.
1864 (set_value_range_to_nonnull): Same.
1865 (set_value_range_to_null): Same.
1866 (range_is_null): Same.
1867 (range_is_nonnull): Same.
1868 (range_int_cst_p): Same.
1869 (range_int_cst_singleton_p): Same.
1870 (symbolic_range_p): Same.
1871 (range_includes_zero_p): Same.
1872 (value_range_constant_singleton): Same.
1873 (vrp_set_zero_nonzero_bits): Same.
1874 (ranges_from_anti_range): Same.
1875 (extract_range_into_wide_ints): Same.
1876 (extract_range_from_multiplicative_op): Same.
1877 (set_value_range_with_overflow): Same.
1878 (extract_range_from_binary_expr_1): Same.
1879 (extract_range_from_unary_expr): Same.
1880 (dump_value_range): Same.
1881 (debug_value_range): Same.
1882 (vrp_prop::check_array_ref): Same.
1883 (vrp_prop::check_mem_ref): Same.
1884 (vrp_prop::vrp_initialize): Same.
1885 (vrp_prop::visit_stmt): Same.
1886 (intersect_ranges): Same.
1887 (vrp_prop::visit_phi): Same.
1888 (vrp_prop::vrp_finalize): Same.
1889 (determine_value_range_1): Same.
1890 (determine_value_range): Same.
1891 (vrp_intersect_ranges_1): Rename to...
1892 (vrp_intersect_1): this.
1893 (vrp_intersect_ranges): Rename to...
1894 (value_range::intersect_helper): ...this.
1895 (vrp_meet_1): Rename to...
1896 (value_range::union_helper): ...this.
1897 (vrp_meet): Rename to...
1898 (value_range::union_): ...this.
1899 (copy_value_range): Remove.
1900 * tree-vrp.h (struct value_range): Rewrite into a proper class.
1901 (value_range::vrtype): New.
1902 (value_range::type): New.
1903 (value_range::equiv): New.
1904 (value_range::min): New.
1905 (value_range::max): New.
1906 (value_range::varying_p): New.
1907 (value_range::undefined_p): New.
1908 (value_range::null_p): New.
1909 (value_range::equiv_add): New.
1910 (copy_value_range): Remove.
1912 2018-10-17 David Malcolm <dmalcolm@redhat.com>
1914 * Makefile.in (SELFTEST_TARGETS): New.
1915 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
1916 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
1917 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
1919 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
1920 (selftest-c++-gdb, selftest-c++-valgrind): Move to
1922 * configure: Regenerate.
1923 * configure.ac (selftest_languages): New.
1925 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1927 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
1928 overflow wraps argument.
1929 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
1930 wide_int_range_multiplicative_op.
1931 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
1932 overflow wraps argument.
1933 (wide_int_range_multiplicative_op): Same.
1934 (wide_int_range_lshift): Same.
1935 (wide_int_range_div): Same.
1936 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
1937 (wide_int_range_lshift): Same.
1938 (wide_int_range_div): Same.
1940 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
1942 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
1943 use sign as argument.
1944 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
1945 wide_int_range_shift_undefined_p.
1947 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1949 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
1951 (@despeculate_copy<ALLI_TI:mode>): ... This.
1952 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
1955 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1957 * config.gcc: Obsolete *-*-solaris2.10*.
1958 * doc/install.texi (Specific, *-*-solaris2*): Document it.
1960 2018-10-12 Jeff Law <law@redhat.com>
1962 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
1963 reg + sym +- const_int addressing modes.
1965 2018-10-15 David Malcolm <dmalcolm@redhat.com>
1967 * common.opt (fdiagnostics-minimum-margin-width=): New option.
1968 * diagnostic-show-locus.c (layout::layout): Apply the minimum
1970 (layout::start_annotation_line): Only print up to 3 of the
1971 margin character, to avoid touching the left-hand side.
1972 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
1973 minimum margin width, as set by test_diagnostic_context's ctor.
1974 (selftest::test_fixit_insert_containing_newline): Likewise.
1975 (selftest::test_fixit_insert_containing_newline_2): Likewise.
1976 (selftest::test_line_numbers_multiline_range): Clear
1977 dc.min_margin_width.
1978 * diagnostic.c (diagnostic_initialize): Initialize
1980 * diagnostic.h (struct diagnostic_context): Add field
1982 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
1983 * opts.c (common_handle_option): Handle
1984 OPT_fdiagnostics_minimum_margin_width_.
1985 * selftest-diagnostic.c
1986 (selftest::test_diagnostic_context::test_diagnostic_context):
1987 Initialize min_margin_width to 6.
1988 * toplev.c (general_init): Initialize global_dc->min_margin_width.
1990 2018-10-15 David Malcolm <dmalcolm@redhat.com>
1992 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
1993 Fix usage of "error_at_rich_loc" in the comment.
1995 2018-10-15 Renlin Li <renlin.li@arm.com>
1998 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
1999 if-conversioned loop when it contains ifn with types not
2000 supported by backend.
2001 * internal-fn.c (expand_direct_optab_fn): Add an assert.
2002 (direct_internal_fn_supported_p): New helper function.
2003 * internal-fn.h (direct_internal_fn_supported_p): Declare.
2005 2018-10-15 Jakub Jelinek <jakub@redhat.com>
2008 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
2009 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
2010 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
2012 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
2014 PR tree-optimization/87022
2015 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
2016 bits in dist vector rather than the first one.
2018 2018-10-15 Richard Biener <rguenther@suse.de>
2021 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
2022 (visit_loadstore): When a used restrict tag escaped verify that
2023 the points-to solution of "other" pointers do not include
2025 (compute_dependence_clique): If a used restrict tag escaped
2026 communicated that down to visit_loadstore.
2028 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
2030 * config/s390/s390.c (s390_expand_vec_init): Force vector element
2031 into reg if it isn't a general operand.
2033 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
2036 * config/i386/sse.md (*vec_dupv2di): Add register source to
2039 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
2042 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
2043 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
2044 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
2045 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
2047 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
2049 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
2050 (notice_args_size): Set it in the current trace if no insn that can
2051 throw internally has been seen yet.
2052 (connect_traces): When connecting args_size between traces, allow the
2053 incoming values not to match if there is an insn setting it before the
2054 first insn that can throw internally; in that case, force the creation
2055 of a CFI note on this latter insn.
2057 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
2059 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
2060 the base class, because using the injected-class-name was not clearly
2061 specified until DR 176.
2063 2018-10-12 Paul Koning <ni1d@arrl.net>
2065 * config/pdp11/pdp11.md (doloop_end): New expander.
2066 (doloop_end_insn): renamed from "doloop_end".
2067 (addqi3): New pattern.
2068 (subqi3): New pattern.
2069 * config/pdp11/predicates.md (incdec_operand): New predicate.
2071 2018-10-12 Yury Gribov <tetra2005@gmail.com>
2074 * real.c (format_helper::can_represent_integral_type_p): New function
2075 * real.h (format_helper::can_represent_integral_type_p): Ditto.
2076 * match.pd: New pattern.
2078 2018-10-12 Alexandre Oliva <oliva@adacore.com>
2080 * configure.ac: Introduce --enable-large-address-aware
2081 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
2082 * doc/install.texi: Document it.
2083 * configure, config.in: Rebuilt.
2084 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
2085 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
2086 (LINK_SPEC): Insert it.
2087 * config/i386/mingw-w64.h: Likewise.
2089 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
2091 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
2093 PR rtl-optimization/87600
2094 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
2096 2018-10-12 Paul Koning <ni1d@arrl.net>
2098 * doc/md.texi (doloop_end): Document that the pattern code may
2099 need to check operand mode.
2101 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
2103 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
2104 to zero-extend between int and floating-point registers.
2105 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
2106 ldp into floating-point registers. Add type and arch attributes.
2107 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
2108 Use f_loads for type attribute.
2110 2018-10-11 Martin Sebor <msebor@redhat.com>
2112 * doc/extend.texi (attribute packed): Correct typos.
2114 2018-10-11 Martin Sebor <msebor@redhat.com>
2116 * doc/extend.texi (attribute flatten): Mention interaction with
2119 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
2122 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
2123 Set new_decl virtual flag to zero.
2125 2018-10-11 Martin Sebor <msebor@redhat.com>
2128 * doc/extend.texi (attribute format_arg): Discuss using multiple
2129 attributes on a single function.
2131 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
2133 PR tree-optimization/86829
2134 * match.pd (sin (atan (x))): New simplification rules.
2135 (cos (atan (x))): Likewise.
2136 * real.c (build_sinatan_real): New function.
2137 * real.h (build_sinatan_real): Prototype.
2139 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
2141 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
2143 (fold_mergeeo_helper): New helper function.
2144 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
2145 intrinsics. Correct some whitespace indentation issues.
2147 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
2150 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
2151 Use HOST_WIDE_INT_1U for shift.
2153 2018-10-11 Doug Rupp <rupp@adacore.com>
2154 Olivier Hainque <hainque@adacore.com>
2156 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
2157 Pass --relax to the linker for RTPs.
2158 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
2160 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
2161 Jan Hubicka <jh@suse.cz>
2162 Martin Jambor <mjambor@suse.cz>
2164 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
2165 the same elements are repeated rather than printing all of them.
2166 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
2167 "repeated" elements.
2168 * read-rtl-function.c (test_loading_repeat): New function.
2169 (read_rtl_function_c_tests): Call test_loading_repeat.
2170 * rtl-tests.c (test_dumping_repeat): New function.
2171 (rtl_tests_c_tests): Call test_dumping_repeat.
2173 2018-10-11 Richard Biener <rguenther@suse.de>
2175 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
2176 bdver?_cost): Unify to ...
2177 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
2178 * config/i386/i386.c (processor_cost_table): Adjust.
2180 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
2183 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
2184 the thunk when expanding to GIMPLE.
2186 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
2188 * varasm.c (mergeable_string_section): Don't try to move zero-length
2189 strings to the merge section.
2191 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
2194 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
2196 2018-10-10 Jakub Jelinek <jakub@redhat.com>
2199 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
2200 to special_args set.
2202 2018-10-10 Richard Biener <rguenther@suse.de>
2204 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
2205 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
2206 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
2207 and recursing and pattern terminating the recursion on SSE
2208 vector width using ix86_expand_reduc.
2209 (reduc_sminmax_scal_<mode>): Split into part reducing to half
2210 width and recursing and SSE2 vector variant doing the final
2211 reduction with ix86_expand_reduc.
2212 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
2213 with terminating the recursion at AVX level, splitting that
2216 2018-10-09 David Malcolm <dmalcolm@redhat.com>
2218 * genmatch.c (error_cb): Rename to...
2219 (diagnostic_cb): ...this, converting int params to enums.
2220 (fatal_at): Update for renaming.
2221 (warning_at): Likewise.
2223 * input.c (selftest::ebcdic_execution_charset::apply):
2224 Update for renaming of...
2225 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
2227 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
2228 converting level and reason to enums.
2229 (class selftest::lexer_error_sink): Rename to...
2230 (class selftest::lexer_test_options): ...this, renaming field
2231 "m_errors" to "m_diagnostics".
2232 (selftest::lexer_test_options::apply): Update for renaming of...
2233 (selftest::lexer_test_options::on_error): ...this, renaming to...
2234 (selftest::lexer_test_options::on_diagnostic): ...this
2235 converting level and reason to enums.
2236 (selftest::test_lexer_string_locations_raw_string_unterminated):
2237 Update for renamings.
2238 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
2241 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
2243 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
2244 * config/rs6000/pmmintrin.h: New file.
2246 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
2248 PR tree-optimization/86659
2249 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
2251 2018-10-09 Richard Biener <rguenther@suse.de>
2253 PR tree-optimization/63155
2254 * tree-ssa-structalias.c: Include tree-ssa.h.
2255 (get_constraint_for_ssa_var): For undefs return nothing_id.
2256 (find_func_aliases): Cleanup PHI handling.
2258 2018-10-09 Richard Biener <rguenther@suse.de>
2260 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
2263 2018-10-09 Martin Liska <mliska@suse.cz>
2265 * asan.c (asan_emit_stack_protection): If a stack variable
2266 is located in a same file as current function, then emit
2267 line info into variable definition string.
2269 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
2271 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
2274 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
2276 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
2279 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
2281 PR tree-optimization/86659
2282 * gimple-match.h (struct gimple_match_op): Add reverse field.
2283 (gimple_match_op::set_op): New overloaded method.
2284 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
2285 the REF_REVERSE_STORAGE_ORDER flag on the value.
2286 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
2287 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
2289 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
2292 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
2293 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
2295 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
2298 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
2299 Defined with __builtin_ia32_vfmaddsubpd512_mask.
2301 2018-10-08 Richard Biener <rguenther@suse.de>
2303 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
2304 cost the same as AVX128 ones.
2306 2018-10-08 Paul Koning <ni1d@arrl.net>
2308 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
2309 (expand_block_move): New function.
2310 * config/pdp11/pdp11.c (output_block_move): Remove.
2311 (expand_block_move): New function.
2312 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
2313 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
2314 (*movmemhi1): Remove.
2316 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
2318 * config/s390/2827.md: Increase latencies for some FP instructions.
2320 2018-10-08 Richard Biener <rguenther@suse.de>
2322 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
2324 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
2325 * tree-vectorizer.h (dump_stmt_cost): Adjust.
2326 (add_stmt_cost): Dump return value of the hook.
2328 2018-10-08 Richard Biener <rguenther@suse.de>
2330 PR tree-optimization/63155
2331 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
2332 (ssa_propagation_engine::ssa_propagate): Remove redundant
2333 bitmap bit clearing.
2335 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
2337 PR rtl-optimization/86939
2338 PR rtl-optimization/87479
2339 * ira.h (non_conflicting_reg_copy_p): New prototype.
2340 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
2341 (make_hard_regno_dead): Don't add conflicts for register
2342 ignore_reg_for_conflicts.
2343 (make_object_dead): Likewise.
2344 (non_conflicting_reg_copy_p): New function.
2345 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
2346 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
2347 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
2348 (make_hard_regno_dead): Don't add conflicts for register
2349 ignore_reg_for_conflicts. Remove special conflict handling of
2350 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
2351 check_pic_pseudo_p and update callers.
2352 (mark_pseudo_dead): Don't add conflicts for register
2353 ignore_reg_for_conflicts.
2354 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
2356 2018-10-05 Andrew Waterman <andrew@sifive.com>
2357 Jim Wilson <jimw@sifive.com>
2359 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
2360 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
2361 new pattern using HONOR_SNANS that emits one extra instruction.
2363 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2365 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
2366 patterns): Merge SI and DI patterns to a GPR pattern.
2367 (unnamed define_insn and define_split for record form of that): Merge
2368 to a single define_insn_and_split pattern.
2370 2018-10-05 David Malcolm <dmalcolm@redhat.com>
2373 * input.c (expand_location_to_spelling_point): Add param "aspect"
2374 and use rather than hardcoding LOCATION_ASPECT_CARET.
2375 (get_substring_ranges_for_loc): Handle the case of a single token
2376 within a macro expansion.
2377 * input.h (expand_location_to_spelling_point): Add "aspect" param,
2378 defaulting to LOCATION_ASPECT_CARET.
2380 2018-10-05 Paul Koning <ni1d@arrl.net>
2382 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
2383 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
2384 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
2385 (pdp11_guard_type): New function.
2387 2018-10-05 Paul Koning <ni1d@arrl.net>
2389 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
2390 * config/pdp11/pdp11.opt (mfloat32): Remove.
2392 * doc/invoke.texi (pdp11 -mfloat32): Remove:
2393 (pdp11 -mfloat64): Remove.
2395 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2397 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
2398 (*cmp<mode>_cc_i387): Ditto.
2399 (*cmpu<mode>_cc_i387): Ditto.
2400 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2401 * config/i386/i386.c (ix86_expand_fp_compare): Remove
2403 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
2404 Emit x86_sahf_1 pattern.
2405 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
2406 (ix86_expand_carry_flag_compare): Ditto.
2408 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2410 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
2411 to reg_or_0_operand. Add "C" constraint.
2412 (*cmpxf_cc_i387): Ditto.
2413 (*cmp<mode>_i387): Change operand 2 predicate
2414 to nonimm_or_0_operand. Add "C" constraint.
2415 (*cmp<mode>_cc_i387): Ditto.
2416 (*cmp<mode>_0_i387): Remove insn pattern.
2417 (*cmp<mode>_0_cc_i387): Ditto.
2419 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
2421 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
2422 * config/i386/predicates.md (nonimm_or_0_operand): Rename
2423 from vector_move_operand. Update all uses.
2425 2018-10-05 Martin Sebor <msebor@redhat.com>
2427 PR tree-optimization/87490
2428 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
2431 2018-10-05 Richard Biener <rguenther@suse.de>
2433 PR tree-optimization/63155
2434 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
2435 vertical space in dumpfiles.
2436 * tree-ssa-propagate.h
2437 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
2438 * tree-ssa-propagate.c (cfg_blocks_back): New global.
2439 (ssa_edge_worklist_back): Likewise.
2440 (curr_order): Likewise.
2441 (cfg_blocks_get): Remove abstraction.
2442 (cfg_blocks_add): Likewise.
2443 (cfg_blocks_empty_p): Likewise.
2444 (add_ssa_edge): Add to current or next worklist based on
2446 (add_control_edge): Likewise.
2447 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
2449 (ssa_propagation_engine::ssa_propagate): ... here. Unify
2450 iteration from CFG and SSA edge worklist so we process
2451 everything in RPO order, prioritizing forward progress
2453 (ssa_prop_init): Allocate new worklists, do not dump
2455 (ssa_prop_fini): Free new worklists.
2457 2018-10-05 Richard Biener <rguenther@suse.de>
2459 * tree-core.h (tree_block::abstract_flag): Remove.
2460 (tree_block::block_num): Make full 32bits.
2461 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
2462 * tree.h (BLOCK_ABSTRACT): Remove.
2463 * dwarf2out.c (gen_lexical_block_die): Remove dead code
2464 resulting from BLOCK_ABSTRACT being always false.
2465 (gen_inlined_subroutine_die): Likewise.
2466 (gen_block_die): Likewise.
2467 * tree.c (block_ultimate_origin): Likewise.
2468 * tree-pretty-print.c (dump_block_node): Remove code dealing
2469 with BLOCK_ABSTRACT.
2470 * tree-ssa-live.c (dump_scope_block): Likewise.
2471 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
2472 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
2474 2018-10-05 Richard Biener <rguenther@suse.de>
2476 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
2477 is asked for initialize mode to the component mode of the
2480 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
2483 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
2484 assembler for -mavx.
2485 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
2487 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
2490 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
2492 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
2493 to be DImode. When using mffscrn, force the operand to a register.
2495 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
2497 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
2498 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
2499 X87MODEF mode iterator.
2500 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
2501 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
2502 X87MODEF mode iterator.
2504 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
2506 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
2507 -Wno-prio-ctor-dtor.
2509 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2511 * Makefile.in (OBJS): Add opt-problem.o.
2512 * dump-context.h: Include "selftest.h.
2513 (selftest::temp_dump_context): New forward decl.
2514 (class dump_context): Make friend of class
2515 selftest::temp_dump_context.
2516 (dump_context::dump_loc_immediate): New decl.
2517 (class dump_pretty_printer): Move here from dumpfile.c.
2518 (class temp_dump_context): Move to namespace selftest.
2519 (temp_dump_context::temp_dump_context): Add param
2520 "forcibly_enable_dumping".
2521 (selftest::verify_dumped_text):
2522 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
2523 (selftest::verify_item):
2524 (ASSERT_IS_TEXT): Move here from dumpfile.c.
2525 (ASSERT_IS_TREE): Likewise.
2526 (ASSERT_IS_GIMPLE): Likewise.
2527 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
2529 (dump_context::dump_loc_immediate): ...this new function.
2530 (class dump_pretty_printer): Move to dump-context.h.
2531 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
2532 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
2533 (temp_dump_context::temp_dump_context): Move to "selftest"
2534 namespace. Add param "forcibly_enable_dumping", and use it to
2535 conditionalize the use of m_pp;
2536 (selftest::verify_dumped_text): Make non-static.
2537 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
2538 (selftest::verify_item): Make non-static.
2539 (ASSERT_IS_TEXT): Move to dump-context.h.
2540 (ASSERT_IS_TREE): Likewise.
2541 (ASSERT_IS_GIMPLE): Likewise.
2542 (selftest::test_capture_of_dump_calls): Pass "true" for new
2543 param of temp_dump_context.
2544 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
2545 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
2547 * opt-problem.cc: New file.
2548 * opt-problem.h: New file.
2549 * optinfo-emit-json.cc
2550 (selftest::test_building_json_from_dump_calls): Pass "true" for
2551 new param of temp_dump_context.
2552 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
2553 (optinfo::emit_for_opt_problem): New function.
2554 (optinfo::emit): Clarity which emit_item is used.
2555 * optinfo.h (optinfo::get_dump_location): New accessor.
2556 (optinfo::emit_for_opt_problem): New decl.
2557 (optinfo::emit): Make const.
2558 * selftest-run-tests.c (selftest::run_tests): Call
2559 selftest::opt_problem_cc_tests.
2560 * selftest.h (selftest::opt_problem_cc_tests): New decl.
2561 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
2562 bool to opt_result, converting fprintf messages to
2563 opt_result::failure_at calls. Add "stmt" param for use by the
2565 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
2566 (runtime_alias_check_p): Convert return type from bool to
2567 opt_result, converting dump_printf calls to
2568 opt_result::failure_at, using the statement DDR_A for their
2570 (find_data_references_in_stmt): Convert return type from bool to
2571 opt_result, converting "return false" to opt_result::failure_at
2573 * tree-data-ref.h: Include "opt-problem.h".
2574 (dr_analyze_innermost): Convert return type from bool to opt_result,
2575 and add a const gimple * param.
2576 (find_data_references_in_stmt): Convert return type from bool to
2578 (runtime_alias_check_p): Likewise.
2579 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
2580 dr_analyze_innermost.
2581 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
2582 Convert return type from bool to opt_result, adding a message for
2583 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
2584 (vect_analyze_data_ref_dependence): Convert return type from bool
2585 to opt_result. Change sense of return type from "false"
2586 effectively meaning "no problems" to "false" meaning a problem,
2587 so that "return false" becomes "return opt_result::success".
2588 Convert "return true" calls to opt_result::failure_at, using
2589 the location of statement A rather than vect_location.
2590 (vect_analyze_data_ref_dependences): Convert return type from bool
2592 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
2593 calls to opt_result::failure_at, using the stmt location rather
2595 (vect_verify_datarefs_alignment): Convert return type from bool
2597 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
2598 into multiple more-tightly-scoped copies.
2599 (vect_analyze_data_refs_alignment): Convert return type from bool
2601 (vect_analyze_data_ref_accesses): Likewise, converting a
2602 "return false" to a "return opt_result::failure_at", adding a
2604 (vect_prune_runtime_alias_test_list): Convert return type from
2605 bool to opt_result, converting dump_printf_loc to
2606 opt_result::failure_at. Add a %G to show the pertinent statement,
2607 and use the stmt's location rather than vect_location.
2608 (vect_find_stmt_data_reference): Convert return type from
2609 bool to opt_result, converting dump_printf_loc to
2610 opt_result::failure_at, using stmt's location.
2611 (vect_analyze_data_refs): Convert return type from bool to
2612 opt_result. Convert "return false" to "return
2613 opt_result::failure_at", adding messages as needed.
2614 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
2615 type from bool to opt_result.
2616 (vect_determine_vf_for_stmt): Likewise.
2617 (vect_determine_vectorization_factor): Likewise, converting
2618 dump_printf_loc to opt_result::failure_at, using location of phi
2619 rather than vect_location.
2620 (vect_analyze_loop_form_1): Convert return type from bool to
2621 opt_result, converting dump_printf_loc calls, retaining the use of
2623 (vect_analyze_loop_form): Convert return type from loop_vec_info
2624 to opt_loop_vec_info.
2625 (vect_analyze_loop_operations): Convert return type from bool to
2626 opt_result, converting dump_printf_loc calls, using the location
2627 of phi/stmt rather than vect_location where available. Convert
2628 various "return false" to "return opt_result::failure_at" with
2629 "unsupported phi" messages.
2630 (vect_get_datarefs_in_loop): Convert return type from bool to
2631 opt_result. Add a message for the
2632 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
2633 (vect_analyze_loop_2): Convert return type from bool to
2634 opt_result. Ensure "ok" is set to a opt_result::failure_at before
2635 each "goto again;", adding new messages where needed.
2636 Add "unsupported grouped {store|load}" messages.
2637 (vect_analyze_loop): Convert return type from loop_vec_info to
2639 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
2641 * tree-vect-stmts.c (process_use): Likewise, converting
2642 dump_printf_loc call and using stmt location, rather than
2644 (vect_mark_stmts_to_be_vectorized): Likeise.
2645 (vect_analyze_stmt): Likewise, adding a %G.
2646 (vect_get_vector_types_for_stmt): Convert return type from bool to
2647 opt_result, converting dump_printf_loc calls and using stmt
2648 location, rather than vect_location.
2649 (vect_get_mask_type_for_stmt): Convert return type from tree to
2650 opt_tree, converting dump_printf_loc calls and using stmt location.
2651 * tree-vectorizer.c: Include "opt-problem.h.
2652 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
2653 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
2654 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
2655 enabled, use it to report at the top level "couldn't vectorize
2656 loop" followed by the problem.
2657 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
2658 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
2660 (vect_analyze_stmt): Likewise.
2661 (vect_get_vector_types_for_stmt): Likewise.
2662 (tree vect_get_mask_type_for_stmt): Likewise.
2663 (vect_analyze_data_ref_dependences): Likewise.
2664 (vect_enhance_data_refs_alignment): Likewise.
2665 (vect_analyze_data_refs_alignment): Likewise.
2666 (vect_verify_datarefs_alignment): Likewise.
2667 (vect_analyze_data_ref_accesses): Likewise.
2668 (vect_prune_runtime_alias_test_list): Likewise.
2669 (vect_find_stmt_data_reference): Likewise.
2670 (vect_analyze_data_refs): Likewise.
2671 (vect_analyze_loop): Convert return type from loop_vec_info to
2673 (vect_analyze_loop_form): Likewise.
2674 (vect_analyze_slp): Convert return type from bool to opt_result.
2676 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2678 * doc/invoke.texi (-fopt-info): Document new "internals"
2680 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
2681 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
2683 (optinfo_verbosity_options): Add "internals".
2684 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
2685 (dump_context::apply_dump_filter_p): New member function.
2686 (dump_context::dump_loc): Use apply_dump_filter_p rather than
2687 explicitly masking the dump_kind.
2688 (dump_context::begin_scope): Increment the scope depth first. Use
2689 apply_dump_filter_p rather than explicitly masking the dump_kind.
2690 (dump_context::emit_item): Use apply_dump_filter_p rather than
2691 explicitly masking the dump_kind.
2692 (dump_dec): Likewise.
2693 (dump_hex): Likewise.
2694 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
2695 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
2696 (opt_info_switch_p): Update handling of default
2697 MSG_OPTIMIZED_LOCATIONS to cope with default of
2698 MSG_PRIORITY_USER_FACING.
2699 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
2700 masking the dump_kind.
2701 (selftest::test_capture_of_dump_calls): Update test_dump_context
2702 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
2703 than MSG_ALL. Generalize scope test to be run at all four
2704 combinations of with/without MSG_PRIORITY_USER_FACING and
2705 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
2706 for each of the two values.
2707 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
2708 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
2709 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
2710 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
2711 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
2712 with MSG_PRIORITY_*.
2713 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
2714 dump messages as MSG_PRIORITY_USER_FACING.
2715 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
2716 about the interaction with MSG_PRIORITY_*.
2718 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
2720 * varasm.c (output_constant): Add new parameter merge_strings.
2721 Make strings properly zero terminated in merge string sections.
2722 (mergeable_string_section): Don't fail if the last char is non-zero.
2723 (assemble_variable_contents): Handle merge string sections.
2724 (assemble_variable): Likewise.
2725 (assemble_constant_contents): Likewise.
2726 (output_constant_def_contents): Likewise.
2727 (output_constructor_array_range,
2728 output_constructor_regular_field): Adjust call to output_constant.
2729 (output_object_block): Adjust call to assemble_constant_contents
2730 and assemble_variable_contents.
2732 2018-10-04 Martin Liska <mliska@suse.cz>
2735 * cgraphunit.c (process_function_and_variable_attributes):
2736 Warn about a function with alias attribute and a body.
2738 2018-10-04 Martin Liska <mliska@suse.cz>
2741 * multiple_target.c (redirect_to_specific_clone): New function.
2742 (ipa_target_clone): Use it.
2743 * tree-inline.c: Fix comment.
2745 2018-10-04 David Malcolm <dmalcolm@redhat.com>
2747 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
2749 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
2750 (gcc::dump_manager::register_pass): New member function, adapted
2751 from loop body in gcc::pass_manager::register_pass, adding a
2752 call to update_dfi_for_opt_info.
2753 (gcc::dump_manager::opt_info_enable_passes): Store the
2754 -fopt-info options into the new fields. Move the loop
2756 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
2758 * dumpfile.h (struct opt_pass): New forward decl.
2759 (gcc::dump_manager::register_pass): New decl.
2760 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
2761 (class gcc::dump_manager): Add fields "m_optgroup_flags",
2762 "m_optinfo_flags", and "m_optinfo_filename".
2763 * passes.c (gcc::pass_manager::register_pass): Move all of the
2764 dump-handling code to gcc::dump_manager::register_pass.
2766 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
2768 PR rtl-optimization/87466
2769 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
2770 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
2771 * doc/tm.texi: Regenerate.
2772 * ira-lives.c (process_bb_node_lives): Use the new target hook.
2773 * lra-lives.c (process_bb_lives): Likewise.
2774 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
2777 2018-10-04 Tamar Christina <tamar.christina@arm.com>
2779 * params.c (add_params): Fix initialization.
2781 2018-10-04 Martin Liska <mliska@suse.cz>
2783 PR gcov-profile/84107
2784 * tree-profile.c (init_ic_make_global_vars):
2785 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
2786 Come up with new ic_tuple* variables. Emit
2787 __gcov_indirect_call{,_topn} variables.
2788 (gimple_gen_ic_profiler): Access the variable
2790 (gimple_gen_ic_func_profiler): Access
2791 __gcov_indirect_call.callee field.
2792 (gimple_init_gcov_profiler): Use ptr_type_node.
2793 * value-prof.c (gimple_ic): Use ptr_type_node.
2795 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2797 PR tree-optimization/85787
2798 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
2799 into this function and add support for detecting multiple phis.
2800 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
2802 2018-10-04 Martin Liska <mliska@suse.cz>
2805 * ipa-inline.c (inline_to_all_callers_1):
2806 Call ultimate_alias_target for node being inlined.
2808 2018-10-03 Jeff Law <law@redhat.com>
2810 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
2812 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
2813 * tree.h (get_typenode_from_name): Prototype.
2815 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2817 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
2818 Change operand 2 predicate to nonimmediate_operand.
2819 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
2821 2018-10-03 Martin Sebor <msebor@redhat.com>
2822 Jeff Law <law@redhat.com>
2824 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
2826 (get_string_length): Detect unterminated arrays.
2827 (format_string): Same.
2828 (format_directive): Warn about unterminated arrays.
2829 (handle_gimple_call): Mark statements with no_warning as needed.
2831 2018-10-03 Jim Wilson <jimw@sifive.com>
2833 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
2834 also define __riscv_abi_rve. Delete trailing white space.
2836 2018-10-03 Paul Koning <ni1d@arrl.net>
2838 Enable LRA register allocator for PDP11.
2839 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
2842 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
2843 * config/pdp11/pdp11.opt (-mlra): New option.
2844 * doc/invoke.texi (PDP-11 Options): Document -mlra.
2846 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
2848 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
2849 (*<absneg:code>extend<mode>xf2): Ditto.
2851 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2853 PR tree-optimization/87415
2854 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
2857 2018-10-02 Jeff Law <law@redhat.com>
2859 * gimple-fold.c (get_range_strlen): Only set *nonstr when
2860 an unterminated string is discovered. Bubble up range
2861 even for unterminated strings.
2862 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
2863 indicates the string was not terminated via NONSTR.
2865 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
2867 * tree-vrp.c (extract_range_from_unary_expr): Special case all
2868 pointer conversions.
2869 Do not do anything special for anti-ranges.
2871 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
2873 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
2874 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
2876 2018-10-03 Martin Liska <mliska@suse.cz>
2878 PR gcov-profile/86109
2879 * coverage.c (coverage_begin_function): Do not
2880 mark lambdas as artificial.
2881 * tree-core.h (struct GTY): Remove tm_clone_flag
2882 and introduce new lambda_function.
2883 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
2885 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
2888 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
2889 P8_VECTOR and VSX are enabled.
2891 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
2893 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
2894 0x3907 as CPU model number.
2896 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
2898 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
2899 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
2900 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
2901 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
2902 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
2903 * config/s390/s390.md: Likewise. Rename also the cpu attribute
2904 value from arch12 to z14.
2906 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
2908 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
2910 (isinf<mode>2): Ditto.
2912 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
2914 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
2915 before emitting fxam. Perform calculations in XFmode.
2917 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
2919 * match.pd (((X /[ex] A) +- B) * A): New transformation.
2921 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
2924 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
2925 * tree.c (signed_or_unsigned_type_for): Handle complex.
2927 2018-10-02 Jeff Law <law@redhat.com>
2929 * gimple-fold.c (get_range_strlen): Remove dead code.
2931 2018-10-02 Martin Sebor <msebor@redhat.com>
2932 Jeff Law <law@redhat.com>
2934 * builtins.c (unterminated_array): Add new arguments.
2935 If argument is not terminated, bubble up size and exact
2937 (expand_builtin_strnlen): Detect, avoid expanding
2938 and diagnose unterminated arrays.
2939 (c_strlen): Fill in offset of start of unterminated strings.
2940 * builtins.h (unterminated_array): Update prototype.
2942 2018-10-02 Richard Biener <rguenther@suse.de>
2944 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
2945 of haddv4df, first reduce to SSE width and exploit the fact
2946 that we only need element zero with the reduction result.
2947 (reduc_plus_scal_v2df): Likewise.
2949 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
2951 * dojump.h (do_jump): Delete.
2952 (do_jump_1): Likewise.
2953 (split_comparison): Move around.
2954 * dojump.c (do_jump): Make static.
2955 (do_jump_1): Likewise.
2956 (jumpifnot): Move around.
2957 (jumpifnot_1): Likewise.
2959 (jumpif_1): Likewise.
2960 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
2962 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
2964 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
2965 insns in the delay slot and add_insn_after for the jump insn.
2967 2018-10-02 Richard Biener <rguenther@suse.de>
2969 * tree-inline.c (expand_call_inline): Use the location of
2970 the callee declaration for the inline-entry marker.
2971 * final.c (notice_source_line): Remove special-casing of
2972 NOTE_INSN_INLINE_ENTRY.
2974 2018-10-01 Carl Love <cel@us.ibm.com>
2977 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
2978 (__builtin_mtfsb0): New.
2979 (__builtin_mtfsb1): New.
2980 ( __builtin_set_fpscr_rn): New.
2981 (__builtin_set_fpscr_drn): New.
2982 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
2983 (rs6000_expand_set_fpscr_rn_builtin): Add.
2984 (rs6000_expand_set_fpscr_drn_builtin): Add.
2985 (rs6000_expand_builtin): Add case statement entries for
2986 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
2987 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
2988 RS6000_BUILTIN_MFFSL.
2989 (rs6000_init_builtins): Add ftype initialization and def_builtin
2990 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
2991 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
2992 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
2993 rs6000_mffscdrn): Add define_insn.
2994 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
2995 * doc/extend.texi: Add documentation for the builtins.
2997 2018-10-01 Richard Biener <rguenther@suse.de>
2999 PR tree-optimization/87465
3000 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
3001 causing branch miscounts.
3003 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3005 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
3006 aarch64_option_default_param): New.
3007 (params.h): Include.
3008 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
3009 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
3010 stack-clash protection validation code.
3012 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3014 * params.c (validate_param): New.
3015 (add_params): Use it.
3016 (set_param_value): Refactor param validation into validate_param.
3017 (diagnostic.h): Include.
3018 * diagnostic.h (diagnostic_ready_p): New.
3020 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3022 * params.c (set_param_value):
3023 Add index of parameter being validated.
3024 * common/common-target.def (option_validate_param): New.
3025 * common/common-targhooks.h (default_option_validate_param): New.
3026 * common/common-targhooks.c (default_option_validate_param): New.
3027 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
3028 * doc/tm.texi: Regenerate.
3030 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3033 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3034 Add validation for stack-clash parameters and set defaults.
3036 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3039 * configure.ac: Add stack-clash-protection-guard-size.
3040 * doc/install.texi: Document it.
3041 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
3042 * params.def: Update comment for guard-size.
3043 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
3044 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
3045 * configure: Regenerate.
3047 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3050 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
3051 STACK_DYNAMIC_OFFSET): New.
3052 * config/aarch64/aarch64.c (aarch64_layout_frame):
3053 Update outgoing args size.
3054 (aarch64_stack_clash_protection_alloca_probe_range,
3055 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
3057 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3060 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
3062 * target.def (stack_clash_protection_alloca_probe_range): New.
3063 (stack_clash_protection_final_dynamic_probe): Remove.
3064 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
3065 (default_stack_clash_protection_final_dynamic_probe): Remove.
3066 * targhooks.c: Likewise.
3067 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
3068 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
3069 * doc/tm.texi: Regenerate.
3071 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3074 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
3075 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
3076 aarch64_clamp_to_uimm12_shift): New.
3077 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
3078 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
3080 2018-10-01 Tamar Christina <tamar.christina@arm.com>
3083 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
3085 2018-10-01 Jeff Law <law@redhat.com>
3086 Richard Sandiford <richard.sandiford@linaro.org>
3087 Tamar Christina <tamar.christina@arm.com>
3090 * config/aarch64/aarch64.md
3091 (probe_stack_range): Add k (SP) constraint.
3092 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
3093 STACK_CLASH_MAX_UNROLL_PAGES): New.
3094 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
3095 stack probes for stack clash.
3096 (aarch64_allocate_and_probe_stack_space): New.
3097 (aarch64_expand_prologue): Use it.
3098 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
3099 (aarch64_sub_sp): Add emit_move_imm optional param.
3101 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
3103 PR tree-optimization/87261
3104 * match.pd: Remove trailing whitespace.
3105 Add (x & y) | ~(x | y) -> ~(x ^ y),
3106 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
3108 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
3110 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
3112 (ashlsi3_insn): Update instruction constraints.
3113 (ashrsi3_insn): Likewise.
3114 (rotrsi3): Likewise.
3115 (add_shift): Likewise.
3116 * config/arc/constraints.md (Csz): New 32 bit constraint. It
3117 avoids placing in the limm field small constants which, otherwise,
3118 could end into a small instruction.
3120 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
3122 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
3123 destination register is not odd-even.
3124 (umaddsidi4_split): Likewise.
3126 2018-10-01 Richard Biener <rguenther@suse.de>
3128 * tree-inline.c (expand_call_inline): Store origin of fn
3129 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
3130 * tree.c (block_ultimate_origin): Simplify and do some
3133 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
3135 * config/i386/mmx.md (EMMS): New int iterator.
3136 (emms): New int attribute.
3137 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
3138 EMMS int iterator. Explicitly declare clobbers.
3139 (mmx_emms): Remove expander.
3141 * config/i386/predicates.md (emms_operation): Remove predicate.
3142 (vzeroall_pattern): New predicate.
3143 (vzeroupper_pattern): Rename from vzeroupper_operation.
3144 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
3145 vzeroupper_pattern and vzeroall_pattern predicates.
3147 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
3149 PR rtl-optimization/86939
3150 * ira-lives.c (make_hard_regno_born): Rename from this...
3151 (make_hard_regno_live): ... to this. Remove update to conflict
3152 information. Update function comment.
3153 (make_hard_regno_dead): Add conflict information update. Update
3155 (make_object_born): Rename from this...
3156 (make_object_live): ... to this. Remove update to conflict information.
3157 Update function comment.
3158 (make_object_dead): Add conflict information update. Update function
3160 (mark_pseudo_regno_live): Call make_object_live.
3161 (mark_pseudo_regno_subword_live): Likewise.
3162 (mark_hard_reg_dead): Update function comment.
3163 (mark_hard_reg_live): Call make_hard_regno_live.
3164 (process_bb_node_lives): Likewise.
3165 * lra-lives.c (make_hard_regno_born): Rename from this...
3166 (make_hard_regno_live): ... to this. Remove update to conflict
3167 information. Remove now uneeded check_pic_pseudo_p argument.
3168 Update function comment.
3169 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
3170 to conflict information. Update function comment.
3171 (mark_pseudo_live): Remove update to conflict information. Update
3173 (mark_pseudo_dead): Add conflict information update.
3174 (mark_regno_live): Call make_hard_regno_live.
3175 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
3176 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
3178 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
3181 * config/i386/i386.c (construct_container): Use TImode for
3182 BLKmode values in 2 integer registers.
3184 2018-09-29 Jeff Law <law@redhat.com>
3186 * builtins.c (unterminated_array): Pass in c_strlen_data * to
3187 c_strlen rather than just a tree *.
3188 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
3189 Update recursive calls appropriately. If caller did not provide a
3190 suitable data pointer, create a local one. When a non-terminated
3191 string is discovered, bubble up information about the string via the
3192 c_strlen_data object.
3193 * builtins.h (c_strlen): Update prototype.
3194 (c_strlen_data): New structure.
3195 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
3196 For a type 2 call, if c_strlen indicates a non-terminated string
3197 use the length of the non-terminated string.
3198 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
3200 2018-09-29 Jakub Jelinek <jakub@redhat.com>
3203 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
3204 __m512d type for __A argument rather than __m512.
3206 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
3208 * match.pd (simple_comparison): Don't optimize if either operand is
3209 a function pointer when target needs function pointer canonicalization.
3211 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
3213 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
3214 power5 .. power9 to remove indirection.
3215 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
3216 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
3217 ASM_CPU_476_SPEC): Delete.
3218 (ASM_CPU_SPEC): Adjust.
3219 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
3220 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
3222 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
3224 * config.in: Delete HAVE_AS_DCI.
3225 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
3226 * config/rs6000/rs6000.h: Ditto.
3227 * configure.ac: Delete HAVE_AS_DCI.
3228 * configure: Regenerate.
3230 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
3232 * config.in (HAVE_AS_LWSYNC): Delete.
3233 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
3234 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
3236 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
3237 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
3239 * configure.ac: Delete HAVE_AS_LWSYNC.
3240 * configure: Regenerate.
3242 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
3243 Pierre-Marie de Rodat <derodat@adacore.com>
3245 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
3246 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
3247 (cgraph_node::create_thunk): Add indirect_offset parameter.
3248 (thunk_adjust): Likewise.
3249 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
3250 and initialize the corresponding field with it.
3251 (cgraph_node::dump): Dump indirect_offset field.
3252 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
3253 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
3254 (thunk_adjust): Add indirect_offset parameter and deal with it.
3255 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
3256 pass it to thunk_adjust. Do not call the target hook if it's non-zero
3257 or if the thunk is external or local. Fix formatting. Do not chain
3258 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
3259 if any, in the GIMPLE representation.
3260 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
3261 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
3262 (input_node): Read indirect_offset field.
3263 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
3264 call to thunk_adjust.
3265 * tree-nested.c (struct nesting_info): Add thunk_p field.
3266 (create_nesting_tree): Set it.
3267 (convert_all_function_calls): Copy static chain from targets to thunks.
3268 (finalize_nesting_tree_1): Return early for thunks.
3269 (unnest_nesting_tree_1): Do not finalize thunks.
3270 (gimplify_all_functions): Do not gimplify thunks.
3272 2018-09-28 David Malcolm <dmalcolm@redhat.com>
3274 * opt-suggestions.c (option_proposer::build_option_suggestions):
3275 Release "option_values".
3277 2018-09-28 David Malcolm <dmalcolm@redhat.com>
3279 * coverage.c (get_coverage_counts): Convert problem-reporting dump
3280 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
3281 * dumpfile.c (kind_as_string): New function.
3282 (dump_loc): Rather than a hardcoded prefix of "note: ", use
3283 kind_as_string to vary the prefix based on dump_kind.
3284 (selftest::test_capture_of_dump_calls): Update for above.
3286 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
3288 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
3289 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
3291 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
3293 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
3294 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
3295 INVALID_REGNUM instead of FPSR_REG.
3296 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
3297 * config/i386/i386.md: Update comment of FP compares.
3298 (fldenv): Do not clobber FPSR_REG.
3300 2018-09-28 Richard Biener <rguenther@suse.de>
3302 * tree.h (BLOCK_ORIGIN): New.
3303 * omp-expand.c (grid_expand_target_grid_body): Assign
3304 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
3305 * tree-inline.c (remap_block): Likewise.
3306 * auto-profile.c (get_function_decl_from_block): Simplify
3307 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
3308 * langhooks.c (lhd_print_error_function): Likewise.
3309 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
3311 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
3312 * tree.c (block_nonartificial_location): Likewise.
3313 (block_ultimate_origin): Likewise.
3314 * tree-pretty-print.c (percent_K_format): Likewise. Remove
3315 no longer needed LTO case.
3317 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
3318 Jan Hubicka <jh@suse.cz>
3319 Martin Jambor <mjambor@suse.cz>
3321 * simplify-rtx.c (simplify_merge_mask): New function.
3322 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
3323 same masks are used in op1 or op2.
3324 (test_vec_merge): New function.
3325 (test_vector_ops): Call test_vec_merge.
3327 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
3329 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
3330 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
3331 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
3332 (cypress_costs): Set it.
3333 (supersparc_costs): Likewise.
3334 (hypersparc_costs): Likewise.
3335 (leon_cost): Likewise.
3336 (leon3_costs): Likewise.
3337 (sparclet_costs): Likewise.
3338 (ultrasparc_costs): Likewise.
3339 (ultrasparc_costs): Likewise.
3340 (niagara_costs): Likewise.
3341 (niagara2_costs): Likewise.
3342 (niagara3_costs): Likewise.
3343 (niagara4_costs): Likewise.
3344 (niagara7_costs): Likewise.
3345 (m8_costs): Likewise.
3346 (TARGET_CAN_FOLLOW_JUMP): Define.
3347 (pass_work_around_errata::gate): Minor tweak.
3348 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
3349 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
3350 Automaitcally clear MASK_FSMULD mask for V7 processors.
3351 (sparc_can_follow_jump): New static function.
3352 (output_ubranch): Deal with CROSSING_JUMP_P.
3353 (sparc_use_sched_lookahead): Rewrite using switch statement.
3354 (sparc_issue_rate): Reorder.
3355 (sparc_branch_cost): New function.
3357 2018-09-27 Martin Sebor <msebor@redhat.com>
3359 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
3360 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
3361 (int_fits_type_p): Same.
3363 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
3365 * config/i386/i386.md (FPCR_REG): Remove.
3366 (UNSPEC_FLDCW): Remove.
3367 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
3368 (x86_fldcw_1): Remove insn pattern.
3369 (fnstenv): Do not clobber FPCR_REG.
3371 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
3372 (CALL_USED_REGISTERS): Ditto.
3373 (REG_ALLOC_ORDER): Ditto.
3374 (REG_CLASS_CONTENTS): Ditto.
3375 (HI_REGISTER_NAMES): Ditto.
3376 (ADDITIONAL_REGISTER_NAMES): Use defines instead
3377 of numerical constants.
3378 * config/i386/i386.c (regclass_map): Remove fpsr register.
3379 (dbx_register_map): Ditto.
3380 (dbx64_register_map): Ditto.
3381 (svr4_dbx_register_map): Ditto.
3382 (print_reg): Do not handle FPCR_REG.
3384 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
3387 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
3388 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
3389 Delete, always treat as true.
3390 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
3391 Ditto. Simplify remaining code.
3392 * config/powerpcspe/powerpcspe.h: Ditto.
3393 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
3394 Simplify remaining code.
3395 (rs6000_expand_builtin): Ditto.
3396 * config/rs6000/rs6000.h: Ditto.
3397 * configure.ac: Ditto.
3398 * configure: Regenerate.
3400 2018-09-27 Martin Liska <mliska@suse.cz>
3402 * coverage.c (get_coverage_counts): Revert the formatting
3403 of missing profile opt info.
3405 2018-09-27 Richard Biener <rguenther@suse.de>
3409 * dwarf2out.c (set_block_origin_self): Do not mark outermost
3410 block as we do not output that.
3411 (gen_inlined_subroutine_die): Elide the originally outermost
3412 block, matching what we do for concrete instances.
3413 (decls_for_scope): Add parameter specifying whether to recurse
3416 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
3417 Tom de Vries <tom@codesourcery.com>
3421 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
3422 STORE_FLAG_VALUE == 1.
3424 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
3426 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
3427 constant definitions.
3428 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
3429 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
3430 ("speculation_barrier"): New expander definition.
3432 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
3434 PR gcov-profile/86957
3435 * common.opt: New warning option -Wmissing-profile.
3436 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
3437 * doc/invoke.texi: Document -Wmissing-profile.
3439 2018-09-26 Jim Wilson <jimw@sifive.com>
3441 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
3442 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
3444 2018-09-26 Martin Sebor <msebor@redhat.com>
3446 * tree.c (zerop): Change return type to bool.
3447 (integer_zerop, integer_onep, integer_each_onep): Same.
3448 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3449 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3450 (real_onep, real_minus_onep, chain_index): Same.
3451 (print_type_hash_statistics, type_list_equal): Same.
3452 * tree.h (zerop): Same.
3453 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
3454 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
3455 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
3456 (real_onep, real_minus_onep, chain_index): Same.
3457 (print_type_hash_statistics, type_list_equal): Same.
3459 2018-09-26 Jim Wilson <jimw@sifive.com>
3461 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
3463 2018-09-26 Jakub Jelinek <jakub@redhat.com>
3466 * config/i386/i386.c: Include debug.h and dwarf2out.h.
3467 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
3470 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
3472 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
3474 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3476 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
3477 and FP_SECOND_SSE_REGS.
3478 (REG_CLASS_NAMES): Ditto.
3479 (REG_CLASS_CONTENTS): Ditto.
3480 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
3481 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
3482 (ix86_preferred_output_reload_class): Ditto.
3483 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
3484 clobber constraint to "=&f".
3485 (fix_truncdi_i387): Ditto.
3486 (lrintxfdi2): Ditto.
3487 (fistdi2_<rounding>): Ditto.
3488 (fpremxf4_i387): Change "=u" constraint to "=f".
3489 (fprem1xf4_i387): Ditto.
3491 (fptanxf4_i387): Ditto.
3492 (fxtractxf3_i387): Ditto.
3493 (fscalexf4_i387): Ditto.
3494 (atan2xf3): Change "u" constraint to "f".
3495 (fyl2xxf3_i387): Ditto.
3496 (fyl2xp1xf3_i387): Ditto.
3498 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
3501 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
3502 for removed I387_MASK_PM entity.
3505 2018-09-26 Jeff Law <law@redhat.com>
3507 2018-09-26 Alexey Neyman <stilor@att.net>
3509 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3510 headers are no longer pulled in by <isl/val.h>.
3512 2018-09-26 Richard Biener <rguenther@suse.de>
3515 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
3516 or concrete instance DIE to the tree. Create abstract origin
3517 attributes also for concrete instances.
3519 2018-09-26 Alexey Neyman <stilor@att.net>
3521 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
3522 headers are no longer pulled in by <isl/val.h>.
3524 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
3526 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
3527 Use new helper functions.
3528 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
3529 Use new helper functions.
3530 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
3531 aarch_mm_needs_release): New declarations.
3532 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
3533 aarch_mm_needs_release): New.
3535 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
3537 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
3538 (arm32_output_mi_thunk): Deal with long calls.
3540 2018-09-26 Richard Biener <rguenther@suse.de>
3544 * tree-inline.c (expand_call_inline): When the location
3545 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
3546 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
3547 the inserted BLOCK to make inlined_function_outer_scope_p
3549 * dwarf2out.c (add_call_src_coords_attributes): Do not add
3550 coords for reserved locations.
3552 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3554 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
3555 (*call_indirect_nonlocal_sysv<mode>): Ditto.
3556 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
3557 (*sibcall_nonlocal_sysv<mode>): Ditto.
3558 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
3559 (<bd>_<mode>): Ditto.
3560 (<bd>tf_<mode>): Ditto.
3562 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3564 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
3565 control string as a list of templates instead of as C code.
3566 (*altivec_movti): Ditto.
3567 * config/rs6000/darwin.md (movdf_low_di): Ditto.
3569 2018-09-25 Jim Wilson <jimw@sifive.com>
3571 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
3572 when target symbol is weak.
3574 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
3577 * builtins.c (unterminated_array): Simplify.
3578 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
3579 where pointer arithmetic is safe.
3581 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
3584 * config/rs6000/altivec.md (altivec_vspltb): Use
3585 const_0_to_15_operand instead of u5bit_cint_operand.
3586 (*altivec_vspltb_internal): Ditto.
3587 (altivec_vspltb_direct): Ditto.
3588 (altivec_vsplth): Use const_0_to_7_operand instead of
3590 (*altivec_vsplth_internal): Ditto.
3591 (altivec_vsplth_direct): Ditto.
3592 (altivec_vspltw): Use const_0_to_3_operand instead of
3594 (*altivec_vspltw_internal): Ditto.
3595 (altivec_vspltw_direct): Ditto.
3596 (altivec_vspltsf): Ditto.
3597 (*altivec_vspltsf_internal): Ditto.
3598 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
3599 various splats with the proper size immediate. Reorder the various
3600 cases by ascending size of immediate, and put all such together.
3602 2018-09-25 Richard Biener <rguenther@suse.de>
3605 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
3607 (maybe_create_die_with_external_ref): Do not create
3608 DW_TAG_imported_unit here.
3609 (add_abstract_origin_attribute): Handle external BLOCK refs.
3610 (dwarf2out_abstract_function): Simplify LTO case.
3611 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
3612 rather than using maybe_create_die_with_external_ref.
3614 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
3617 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
3618 (frndintxf2_mask_pm_i387): Ditto.
3619 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
3620 Enable for !flag_trapping_math.
3621 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
3622 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
3623 Change operand 1 predicate to nonimmediate_operand.
3624 (attr "i387_cw"): Remove mask_pm.
3625 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
3626 (enum ix86_entity): Remove I387_MASK_PM.
3627 * config/i386/i386.c (ix86_i387_mode_needed): Do not
3628 handle I387_MASK_PM.
3629 (ix86_mode_needed): Ditto.
3630 (ix86_mode_after): Ditto.
3631 (ix86_mode_entry): Ditto.
3632 (ix86_mode_exit): Ditto.
3633 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
3635 2018-09-25 Jakub Jelinek <jakub@redhat.com>
3637 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
3638 to_update_switch_stmts to vNULL instead of calling create on them
3641 2018-09-25 Richard Biener <rguenther@suse.de>
3643 PR tree-optimization/87402
3644 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
3645 (visit_phi): Re-instantiate handling of supposed to be VARYING
3646 but non-VARYING backedge value.
3648 2018-09-25 Richard Biener <rguenther@suse.de>
3651 * dwarf2out.c (struct sym_off_pair): New.
3652 (external_die_map): New global.
3653 (lookup_decl_die): When in LTO create DIEs lazily from the
3655 (lookup_block_die): New function, create DIEs lazily in LTO.
3656 (equate_block_to_die): New function.
3657 (dwarf2out_die_ref_for_decl): During WPA get the association
3658 from the external DIE map.
3659 (dwarf2out_register_external_die): Record mapping into the
3661 (maybe_create_die_with_external_ref): New function split out from
3662 DIE generation part of old dwarf2out_register_external_die.
3663 (add_abstract_origin_attribute): Do not return the DIE. When
3664 in LTO reference externals directly.
3665 (dwarf2out_abstract_function): When in LTO ignore calls for
3666 decls with external DIEs (already present abstract instances).
3667 (gen_call_site_die): Adjust.
3668 (add_high_low_attributes): Likewise.
3669 (gen_lexical_block_die): Likewise.
3670 (gen_inlined_subroutine_die): Likewie.
3671 (gen_block_die): Likewise.
3672 (dwarf2out_inline_entry): Likewise.
3673 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
3676 2018-09-25 Martin Liska <mliska@suse.cz>
3678 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
3679 integers and not by a float value.
3681 2018-09-25 Martin Liska <mliska@suse.cz>
3684 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
3686 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
3687 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
3688 * godump.c (go_format_type): Remove extra parenthesis.
3690 2018-09-25 Martin Liska <mliska@suse.cz>
3692 * alias.c (set_dest_equal_p): Remove unused function.
3693 * config/i386/i386.c (def_builtin_pure2): Likewise.
3694 * diagnostic-show-locus.c (class layout): Remove
3696 (layout::layout): Likewise here.
3697 * dump-context.h (class temp_dump_context): Likewise.
3698 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
3699 (add_AT_loclistsptr): Likewise.
3700 (add_AT_offset): Likewise.
3701 (get_AT_hi_pc): Likewise.
3702 (is_comdat_die): Likewise.
3703 (type_is_enum): Likewise.
3704 (ceiling): Likewise.
3705 (add_AT_vms_delta): Likewise.
3706 (is_class_die): Likewise.
3707 * edit-context.c (class line_event): Remove unused field.
3708 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
3710 * ipa-cp.c (ipa_get_vr_lat): Likewise.
3711 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
3712 (ok_for_base_p_nonstrict): Likewise.
3713 * tree-chrec.c (is_not_constant_evolution): Likewise.
3714 (chrec_fold_poly_cst): Likewise.
3715 * tree-if-conv.c (has_pred_critical_p): Likewise.
3716 * tree-ssa-coalesce.c (print_exprs): Likewise.
3717 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
3718 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
3719 * tree-vrp.c (value_ranges_intersect_p): Likewise.
3720 (value_range_nonnegative_p): Likewise.
3722 2018-09-25 Martin Liska <mliska@suse.cz>
3724 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
3725 Do not handle "GNU Pascal".
3726 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
3728 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
3729 from documentation. Likewise.
3730 * dbxout.c (dbxout_range_type): Likewise.
3731 * doc/cpp.texi: Likewise.
3732 * doc/extend.texi: Likewise.
3733 * doc/frontends.texi: Likewise.
3734 * doc/invoke.texi: Remove Pascal entry.
3735 * tree.def (CLEANUP_POINT_EXPR): Likewise.
3736 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
3738 2018-09-25 Martin Liska <mliska@suse.cz>
3741 * doc/invoke.texi: Document all parameters and remove default
3744 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
3747 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
3748 contains HOST_WIDE_INTs when computing its size.
3750 2018-09-24 Jim Wilson <jimw@sifive.com>
3753 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
3755 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
3757 2018-09-24 Andrew Pinski <apinski@marvell.com>
3759 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
3760 access prev before checking it for NULLness in the
3761 AARCH64_FUSE_CMP_BRANCH case.
3763 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
3766 * config/i386/i386.c (rest_of_insert_endbranch): Set
3767 endbr_queued_at_entrance to true and don't insert ENDBR if
3768 x86_function_profiler will be called.
3769 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
3771 * config/i386/i386.h (machine_function): Add
3772 endbr_queued_at_entrance.
3774 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3776 * genattrtab.c (mk_attr_alt): Use alternative_mask.
3777 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
3779 (check_attr_test): Use alternative_mask.
3780 (get_attr_value): Likewise.
3781 (compute_alternative_mask): Use alternative_mask and XWINT.
3782 (make_alternative_compare): Use alternative_mask.
3783 (attr_alt_subset_p): Use XWINT.
3784 (attr_alt_subset_of_compl_p): Likewise.
3785 (attr_alt_intersection): Use alternative_mask and XWINT.
3786 (attr_alt_union): Likewise.
3787 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
3788 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
3789 (simplify_test_exp): Use alternative_mask and XWINT.
3790 (write_test_expr): Use alternative_mask and XWINT, adjust bit
3791 number calculation to support 64 bits. Generate code that
3792 checks 64-bit masks.
3793 (main): Use alternative_mask.
3794 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
3796 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
3799 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
3800 RETURN+USE when returning via %r14.
3802 2018-09-24 Martin Liska <mliska@suse.cz>
3804 * gcov.c (output_lines): Print colorization legend
3805 for both flag_use_colors and flag_use_hotness_colors.
3808 2018-09-24 Martin Liska <mliska@suse.cz>
3810 * coverage.c (get_coverage_counts): Use warning_at
3811 with current_function_decl location. Use %qD in warning
3814 2018-09-24 Martin Liska <mliska@suse.cz>
3816 * memory-block.h (memory_block_pool::release): Annotate with
3817 valgrind that the memory is not accessible.
3819 2018-09-24 Martin Liska <mliska@suse.cz>
3822 * asan.c: Make asan_handled_variables extern.
3824 * cfgexpand.c (expand_stack_vars): Make sure
3825 a representative is unpoison if another
3826 variable in the partition is handled by
3827 use-after-scope sanitization.
3829 2018-09-24 Richard Biener <rguenther@suse.de>
3831 PR tree-optimization/63155
3832 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
3833 the worklist when the edge of the respective argument isn't
3836 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3838 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
3839 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
3840 (MASK_CLASS_P): Update for rename.
3841 (MAYBE_MASK_CLASS_P): Ditto.
3842 (REG_CLASS_NAMES): Update.
3843 (REG_CLASS_CONTENT): Update.
3844 * config/i386/i386.c (regclass_map): Update for MASK_REG
3845 and ALL_MASK_REGS rename.
3846 * config/i386/constraints.md (Yk): Update for rename.
3849 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3851 * config/i386/i386.h (enum reg_class): Remove
3852 EVEX_SSE_REGS and MOD4_SSE_REGS.
3853 (REG_CLASS_NAMES): Update.
3854 (REG_CLASS_CONTENT): Update.
3855 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
3856 registers as ALL_SSE_REGS.
3857 (ix86_additional_allocno_class_p): Remove.
3858 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
3859 (ix86_register_priority): Lower priority of EVEX SSE registers.
3860 Use IN_RANGE macro where appropriate.
3861 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
3862 AVX-5124VNNIW checks.
3863 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
3864 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
3865 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
3866 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
3867 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
3868 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
3869 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
3870 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
3871 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
3872 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
3873 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
3874 * config/i386/constraints.md (Yh): Remove.
3876 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
3878 * config/i386/i386.c (regclass_map): Declare integer REX registers
3881 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
3883 * doc/service.texi (Service): Switch the fsf.org link to https.
3885 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
3888 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3889 Define to speculation_safe_value_not_needed.
3891 2018-09-21 Florian Weimer <fweimer@redhat.com>
3894 * doc/extend.texi (Common Function Attributes): Mention that
3895 noreturn suppresses tail call optimization.
3897 2018-09-21 Jeff Law <law@redhat.com>
3899 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
3900 vr_values::cleanup_edges_and_switches.
3901 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
3903 (identify_jump_threads): Remove EDGE_IGNORE handling.
3904 (execute_vrp): Move handling of to_remove_edges and
3905 to_update_switch_stmts into vr_values class member functions.
3906 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
3907 (to_update_switch_stmts): Likewise.
3908 * vr-values.c: Include cfghooks.h.
3909 (vr_values::vr_values): Initialize to_remove_edges and
3910 to_update_switch_stmts.
3911 (vr_values::~vr_values): Verify to_remove_edges and
3912 to_update_switch_stmts are empty.
3913 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
3914 (vr_values::cleanup_edges_and_switches): New member function.
3915 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
3916 function. Add new data members.
3918 2018-09-21 David Malcolm <dmalcolm@redhat.com>
3920 PR tree-optimization/87309
3921 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
3922 calls with pflags and alt_flags.
3923 (selftest::test_capture_of_dump_calls): Add test of interaction of
3924 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
3926 2018-09-21 Olivier Hainque <hainque@adacore.com>
3928 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
3930 2018-09-21 Olivier Hainque <hainque@adacore.com>
3932 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
3934 2018-09-21 Olivier Hainque <hainque@adacore.com>
3936 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
3937 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
3938 (VXWORKS_LIBS_RTP): Minor reordering.
3940 2018-09-21 Olivier Hainque <hainque@adacore.com>
3942 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
3943 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
3945 2018-09-21 Olivier Hainque <hainque@adacore.com>
3947 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
3948 (PTRDIFF_TYPE): Likewise.
3950 2018-09-21 Olivier Hainque <hainque@adacore.com>
3952 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
3953 triplet, similar to support for VxWorks7.
3954 * config/vxworks-dummy.h: Provide a default definition
3955 of TARGET_VXWORKS64 to 0.
3957 2018-09-21 Olivier Hainque <hainque@adacore.com>
3959 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
3960 * config/vxworks-dummy.h: here.
3962 2018-09-21 Olivier Hainque <hainque@adacore.com>
3964 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
3966 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
3967 Bo Zhou <zbo.zhou@hisilicon.com>
3969 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
3970 * config/aarch64/aarch64-tune.md: Regenerated.
3971 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
3972 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
3973 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
3975 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
3976 Julian Brown <julian@codesourcery.com>
3978 * builtins.c (get_builtin_sync_mem): Handle address spaces.
3980 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
3982 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
3983 if the call takes a static chain.
3985 2018-09-21 Martin Liska <mliska@suse.cz>
3987 * auto-profile.c (autofdo_source_profile::read): Do not
3989 (read_profile): Do not add working sets.
3990 (read_autofdo_file): Remove sum_all.
3991 (afdo_callsite_hot_enough_for_early_inline): Remove const
3993 * coverage.c (struct counts_entry): Remove gcov_summary.
3994 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
3995 do not support GCOV_TAG_PROGRAM_SUMMARY.
3996 (get_coverage_counts): Remove summary and expected
3998 * coverage.h (get_coverage_counts): Likewise.
3999 * doc/gcov-dump.texi: Remove -w option.
4000 * gcov-dump.c (dump_working_sets): Remove.
4001 (main): Do not support '-w' option.
4002 (print_usage): Likewise.
4003 (tag_summary): Likewise.
4004 * gcov-io.c (gcov_write_summary): Do not dump
4006 (gcov_read_summary): Likewise.
4007 (gcov_histo_index): Remove.
4008 (gcov_histogram_merge): Likewise.
4009 (compute_working_sets): Likewise.
4010 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
4012 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
4013 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
4014 (GCOV_HISTOGRAM_SIZE): Remove.
4015 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
4016 (struct gcov_summary): Simplify rapidly just
4017 to runs and sum_max fields.
4018 (gcov_histo_index): Remove.
4019 (NUM_GCOV_WORKING_SETS): Likewise.
4020 (compute_working_sets): Likewise.
4021 * gcov-tool.c (print_overlap_usage_message): Remove
4022 trailing empty line.
4023 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
4024 (output_lines): Remove program related line.
4025 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
4026 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
4028 (input_profile_summary): Do not read it.
4029 (merge_profile_summaries): And do not merge it.
4030 (input_symtab): Do not call removed function.
4031 * modulo-sched.c (sms_schedule): Do not print sum_max.
4032 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
4033 removed when histogram method was invented.
4034 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
4036 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
4038 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
4039 and dump selected value.
4040 * profile.c (add_working_set): Remove.
4041 (get_working_sets): Likewise.
4042 (find_working_set): Likewise.
4043 (get_exec_counts): Do not work with working sets.
4044 (read_profile_edge_counts): Do not inform as sum_max is removed.
4045 (compute_branch_probabilities): Likewise.
4046 (compute_value_histograms): Remove argument for call of
4047 get_coverage_counts.
4048 * profile.h: Do not make gcov_summary const.
4050 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
4052 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
4054 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
4056 PR tree-optimization/86990
4057 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
4058 Check that the entire merged store group is made of constants only for
4061 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
4063 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
4064 (VTABLE_VERIFICATION_SPEC): Likewise.
4065 (SANITIZER_EARLY_SPEC): Likewise.
4066 (SANITIZER_SPEC): Likewise.
4067 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
4068 * doc/invoke.texi (Link Options): Document -r.
4070 2018-09-20 Richard Biener <rguenther@suse.de>
4073 * gimplify.c (gimplify_expr): Retain alignment of
4074 addressable lvalue in dereference.
4076 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
4079 * configure.ac: Check for .loc is_stmt support.
4080 * configure, config.in: Rebuilt.
4081 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
4084 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4086 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
4087 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
4090 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
4092 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
4094 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
4097 2018-09-20 Marek Polacek <polacek@redhat.com>
4099 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
4101 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
4103 PR tree-optimization/87288
4104 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
4105 into account when determining PEELING_FOR_NITERS.
4107 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
4109 PR tree-optimization/86877
4110 * tree-vect-loop.c (vect_analyze_loop_2): Call
4111 vect_verify_datarefs_alignment.
4113 2018-09-19 Marek Polacek <polacek@redhat.com>
4115 * doc/invoke.texi: Document -Wclass-conversion.
4117 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
4119 * config/pa/pa.c (pa_adjust_priority): Delete.
4120 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
4122 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
4123 (atomic_storehi): Likewise.
4124 (atomic_storesi): Likewise.
4125 (atomic_loaddi): Restore compare and swap exchange loop code.
4127 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
4129 PR rtl-optimization/86902
4130 * combine.c (try_combine): When changing the CC mode used, don't change
4131 an unrelated mode in other_insn to that new CC mode.
4133 2018-09-19 David Malcolm <dmalcolm@redhat.com>
4135 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
4136 with %T in place of calls to dump_generic_expr.
4137 (prune_runtime_alias_test_list): Likewise.
4138 (create_runtime_alias_checks): Likewise.
4139 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
4140 (vect_analyze_data_ref_dependence): Likewise.
4141 (vect_slp_analyze_data_ref_dependence): Likewise.
4142 (vect_record_base_alignment): Likewise. Use %G in place of call
4143 to dump_gimple_stmt.
4144 (vect_compute_data_ref_alignment): Likewise.
4145 (verify_data_ref_alignment): Likewise.
4146 (vect_find_same_alignment_drs): Likewise.
4147 (vect_analyze_group_access_1): Likewise.
4148 (vect_analyze_data_ref_accesses): Likewise.
4149 (dependence_distance_ge_vf): Likewise.
4150 (dump_lower_bound): Likewise.
4151 (vect_prune_runtime_alias_test_list): Likewise.
4152 (vect_find_stmt_data_reference): Likewise.
4153 (vect_analyze_data_refs): Likewise.
4154 (vect_create_addr_base_for_vector_ref): Likewise.
4155 (vect_create_data_ref_ptr): Likewise.
4156 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
4157 (vect_can_advance_ivs_p): Likewise.
4158 (vect_update_ivs_after_vectorizer): Likewise.
4159 (vect_gen_prolog_loop_niters): Likewise.
4160 (vect_prepare_for_masked_peels): Likewise.
4161 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
4162 (vect_determine_vectorization_factor): Likewise.
4163 (vect_is_simple_iv_evolution): Likewise.
4164 (vect_analyze_scalar_cycles_1): Likewise.
4165 (vect_analyze_loop_operations): Likewise.
4166 (report_vect_op): Likewise.
4167 (vect_is_slp_reduction): Likewise.
4168 (check_reduction_path): Likewise.
4169 (vect_is_simple_reduction): Likewise.
4170 (vect_create_epilog_for_reduction): Likewise.
4171 (vect_finalize_reduction:): Likewise.
4172 (vectorizable_induction): Likewise.
4173 (vect_transform_loop_stmt): Likewise.
4174 (vect_transform_loop): Likewise.
4175 (optimize_mask_stores): Likewise.
4176 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
4177 (vect_split_statement): Likewise.
4178 (vect_recog_over_widening_pattern): Likewise.
4179 (vect_recog_average_pattern): Likewise.
4180 (vect_determine_min_output_precision_1): Likewise.
4181 (vect_determine_precisions_from_range): Likewise.
4182 (vect_determine_precisions_from_users): Likewise.
4183 (vect_mark_pattern_stmts): Likewise.
4184 (vect_pattern_recog_1): Likewise.
4185 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
4186 (vect_record_max_nunits): Likewise.
4187 (vect_build_slp_tree_1): Likewise.
4188 (vect_build_slp_tree_2): Likewise.
4189 (vect_print_slp_tree): Likewise.
4190 (vect_analyze_slp_instance): Likewise.
4191 (vect_detect_hybrid_slp_stmts): Likewise.
4192 (vect_detect_hybrid_slp_1): Likewise.
4193 (vect_slp_analyze_operations): Likewise.
4194 (vect_slp_analyze_bb_1): Likewise.
4195 (vect_transform_slp_perm_load): Likewise.
4196 (vect_schedule_slp_instance): Likewise.
4197 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
4198 (vect_mark_stmts_to_be_vectorized): Likewise.
4199 (vect_init_vector_1): Likewise.
4200 (vect_get_vec_def_for_operand): Likewise.
4201 (vect_finish_stmt_generation_1): Likewise.
4202 (vect_check_load_store_mask): Likewise.
4203 (vectorizable_call): Likewise.
4204 (vectorizable_conversion): Likewise.
4205 (vectorizable_operation): Likewise.
4206 (vectorizable_load): Likewise.
4207 (vect_analyze_stmt): Likewise.
4208 (vect_is_simple_use): Likewise.
4209 (vect_get_vector_types_for_stmt): Likewise.
4210 (vect_get_mask_type_for_stmt): Likewise.
4211 * tree-vectorizer.c (increase_alignment): Likewise.
4213 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
4215 * doc/rtl.texi: Adjust vec_select description.
4216 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
4217 non-constant selectors.
4219 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
4221 * config/aarch64/aarch64-protos.h
4222 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
4223 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
4224 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
4225 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
4226 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
4227 (AARCH64_FL_PROFILE): Move index so flags are ordered.
4228 (AARCH64_ISA_RCPC8_4): New flag.
4229 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
4230 to aarch64_offset_9bit_signed_unscaled_p.
4231 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
4233 * config/aarch64/constraints.md (Ust): New constraint.
4234 * config/aarch64/predicates.md.
4235 (aarch64_9bit_offset_memory_operand): New predicate.
4236 (aarch64_rcpc_memory_operand): New predicate.
4238 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
4240 PR rtl-optimization/87361
4241 * rtlanal.c (nonzero_bits1): Revert accidental change.
4243 2018-09-19 Richard Biener <rguenther@suse.de>
4245 PR tree-optimization/87349
4246 PR tree-optimization/87342
4247 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
4249 2018-09-18 Marek Polacek <polacek@redhat.com>
4251 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
4252 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
4254 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
4256 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
4258 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
4260 PR rtl-optimization/86882
4261 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
4263 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
4265 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
4266 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
4268 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
4271 * doc/invoke.texi (Link Options): Fix formatting and grammar.
4273 2018-09-18 Richard Biener <rguenther@suse.de>
4276 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
4277 (compute_samebase_partition_bases): Likewise.
4278 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
4279 (gimple_can_coalesce_p): Simplify.
4281 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
4283 Handle a library implementation of ffs calling __builtin_ffs.
4284 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
4285 (mmix_init_libfuncs): New function: make __builtin_ffs expand
4288 2018-09-17 David Malcolm <dmalcolm@redhat.com>
4290 * diagnostic-show-locus.c (class layout_range): Add field
4292 (layout_range::layout_range): Add "original_idx" param and use it
4293 to initialize new field.
4294 (make_range): Use 0 for original_idx.
4295 (layout::layout): Pass in index to calls to
4296 maybe_add_location_range.
4297 (layout::maybe_add_location_range): Add param "original_idx" and
4298 pass it on to layout_range.
4299 (layout::print_any_labels): Pass on range->m_original_idx to
4301 (gcc_rich_location::add_location_if_nearby): Use 0 for
4303 * gcc-rich-location.h (text_range_label::get_text): Update for new
4305 (range_label_for_type_mismatch::get_text): Likewise.
4307 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
4309 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
4311 2018-09-17 David Malcolm <dmalcolm@redhat.com>
4313 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
4314 format_string_diagnostic_t.
4315 (fmtwarn_n): Likewise.
4316 * substring-locations.c
4317 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
4318 (format_warning_n_va): Convert to...
4319 (format_string_diagnostic_t::emit_warning_n_va): ...this.
4320 (format_warning_va): Convert to...
4321 (format_string_diagnostic_t::emit_warning_va): ...this.
4322 (format_warning_at_substring): Convert to...
4323 (format_string_diagnostic_t::emit_warning): ...this.
4324 (format_warning_at_substring_n): Convert to...
4325 (format_string_diagnostic_t::emit_warning_n): ...this.
4326 * substring-locations.h (class format_string_diagnostic_t): New
4328 (format_warning_va): Convert to
4329 format_string_diagnostic_t::emit_warning_va.
4330 (format_warning_n_va): Convert to
4331 format_string_diagnostic_t::emit_warning_n_va.
4332 (format_warning_at_substring): Convert to
4333 format_string_diagnostic_t::emit_warning.
4334 (format_warning_at_substring_n): Convert to
4335 format_string_diagnostic_t::emit_warning_n.
4337 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
4338 Bernd Schmidt <bernds_cb1@t-online.de>
4340 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
4343 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
4345 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
4346 operand 0 predicate to nonimmediate operand.
4347 (rint<mode>2_frndint): Remove insn pattern.
4348 (rint<mode>2): Change operand 1 predicate to general_operand.
4349 Extend operand 1 to XFmode and generate rintxf2 insn.
4350 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
4351 Do not use X87MODEF mode macro.
4352 (frndintxf2_<rounding>_i387): Rename from
4353 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
4354 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
4355 to XFmode and generate significandxf3 insn.
4357 2018-09-17 Richard Biener <rguenther@suse.de>
4359 PR tree-optimization/87328
4360 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
4361 visiting unexecutable backedges when not iterating.
4362 (do_rpo_vn): Mark all edges not executable even when not
4365 2018-09-17 Martin Jambor <mjambor@suse.cz>
4368 * doc/invoke.texi (Warning Options): Likewise.
4370 2018-09-17 Richard Biener <rguenther@suse.de>
4372 PR tree-optimization/87301
4373 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
4374 clean EH info from leftover copy assignments.
4376 2018-09-17 Martin Liska <mliska@suse.cz>
4378 PR gcov-profile/85871
4379 * gcov.c (output_intermediate_file): Fix out of bounds
4382 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
4384 * config/arc/arc.c: Object attributes for core4 not reflected
4386 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
4389 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
4391 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
4393 2018-09-17 Martin Liska <mliska@suse.cz>
4395 * doc/gcov.texi: Document new option --use-hotness-colors.
4396 * gcov.c (struct source_info): Declare new field.
4397 (source_info::source_info): Set default for maximum_count.
4398 (print_usage): Add new -q option.
4399 (process_args): Process it.
4400 (accumulate_line_info): Save src->maximum_count.
4401 (output_line_beginning): Make color line number if
4402 flag_use_hotness_colors is set.
4403 (output_line_details): Pass default argument value.
4404 (output_lines): Pass src->maximum_count.
4406 2018-09-17 Martin Liska <mliska@suse.cz>
4408 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
4409 Use processor_names table.
4410 * config/i386/i386.c (ix86_default_align): Use
4411 processor_cost_table for alignment values.
4412 (ix86_option_override_internal): Use processor_names.
4413 (ix86_function_specific_print): Likewise.
4414 * config/i386/i386.h (struct processor_costs):
4415 Add alignment values.
4416 (struct ptt): Remove and replace with const char *.
4417 * config/i386/x86-tune-costs.h (struct processor_costs):
4418 Declare default alignments for all costs.
4420 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
4422 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
4423 symbolics or VR_VARYING ranges for ABS_EXPR.
4424 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
4425 when range will wrap.
4427 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
4430 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
4431 before and after a JUMP_TABLE_DATA.
4433 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
4436 * dojump.c (do_compare_and_jump): Canonicalize function pointers
4437 when one operand is a function pointer. Use POINTER_TYPE_P and
4438 FUNC_OR_METHOD_TYPE_P.
4439 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
4440 * fold-const.c (build_range_check): Likewise.
4441 * match.pd (simple_comparison): Likewise.
4443 2018-09-14 David Malcolm <dmalcolm@redhat.com>
4446 * spellcheck.c (get_edit_distance_cutoff): New function.
4447 (selftest::test_edit_distance_unit_test_oneway): Rename to...
4448 (selftest::test_get_edit_distance_one_way): ...this.
4449 (selftest::test_get_edit_distance_unit): Rename to...
4450 (selftest::test_get_edit_distance_both_ways): ...this.
4451 (selftest::test_edit_distances): Move tests to this new function,
4452 and test some more pairs of strings. Update for above renaming.
4453 (selftest::get_old_cutoff): New function.
4454 (selftest::test_get_edit_distance_cutoff): New function.
4455 (selftest::assert_suggested_for): New function.
4456 (ASSERT_SUGGESTED_FOR): New macro.
4457 (selftest::assert_not_suggested_for): New function.
4458 (ASSERT_NOT_SUGGESTED_FOR): New macro.
4459 (selftest::test_suggestions): New function.
4460 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
4461 tests to selftest::test_edit_distances and call it. Add calls to
4462 selftest::test_get_edit_distance_cutoff and
4463 selftest::test_suggestions.
4464 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
4465 (best_match::consider): Replace hard-coded cutoff calculation with
4467 (best_match::get_cutoff): New declaration.
4468 (best_match::get_best_meaningful_candidate): Likewise.
4470 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4472 * builtins.c (fold_builtin_strlen): Remove TODO comment.
4474 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
4477 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
4479 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
4480 terminated string literal.
4482 2018-09-14 Martin Sebor <msebor@redhat.com>
4484 * builtins.c (unterminated_array): Handle ARRAY_REF.
4485 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
4486 * builtins.h (unterminated_array): Declare extern.
4487 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
4489 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
4492 2018-09-14 Martin Sebor <msebor@redhat.com>
4493 Jeff Law <law@redhat.com>
4495 * builtins.c (unterminated_array): New.
4496 (expand_builtin_strcpy): Adjust.
4497 (expand_builtin_strcpy_args): Detect unterminated arrays.
4498 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
4499 unterminated arrays.
4500 * gimple-fold.h (get_maxval_strlen): Add argument.
4501 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
4503 * gimple-fold.c (get_range_strlen): Add argument.
4504 (get_maxval_strlen): Adjust.
4505 * gimple-fold.h (get_range_strlen): Add argument.
4507 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
4509 * config/i386/movdirintrin.h: Fix copyright year.
4511 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
4513 * reg-stack.c: Include regs.h.
4514 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
4515 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
4516 FIRST_STACK_REG, not DFmode.
4517 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
4518 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
4519 (change stack): Default register mode to the reg_raw_mode of
4520 FIRST_STACK_REG, not DFmode.
4521 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
4522 (*swapxf): Rename from swapxf.
4524 2018-09-14 Carl Love <cel@us.ibm.com>
4526 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
4527 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
4529 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
4532 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
4535 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
4538 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
4540 2018-09-14 Jason Merrill <jason@redhat.com>
4542 Fix --enable-gather-detailed-mem-stats.
4543 * hash-table.c (hash_table_usage): Change from variable to function.
4544 * hash-table.h: Adjust.
4545 * Makefile.in: Add missing dependencies on hash-table.h.
4547 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4549 PR tree-optimization/87259
4551 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
4552 execute_cse_reciprocals_1 has tried transforming.
4554 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
4556 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
4557 VR_VARYING for PLUS/MINUS_EXPR.
4559 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
4561 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
4564 2018-09-14 Richard Biener <rguenther@suse.de>
4567 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
4568 bits for the merged partition.
4570 2018-09-13 Martin Sebor <msebor@redhat.com>
4571 Bernd Edlinger <bernd.edlinger@hotmail.de>
4573 * builtins.h (c_srlen): Add argument.
4574 * builtins.c (warn_string_no_nul): New function.
4575 (c_strlen): Add argument and use it. Update recursive calls.
4576 Pass DECL argument to string_constant to get info on non
4577 terminated strings. Update *NONSTR as needed.
4578 (fold_builtin_strlen): Add argument to calls to c_strlen.
4579 Warn for unterminated arrays.
4580 (warn_string_no_null): Add prototype.
4581 * expr.c (string_constant): Update arguments. Update recursive
4582 calls appropriately. Detect missing NUL terminator and outermost
4583 declaration its missing in.
4584 Improve checks for arrays with nonzero lower bound or elements
4585 that are not a single byte. Simplify offset computation.
4586 Simplify checks for non-NUL terminated strings.
4587 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
4588 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
4590 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
4592 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
4594 * fold-const.c (c_getstr): Fix function comment. Remove unused third
4595 argument. Fix range checks.
4596 * fold-const.h (c_getstr): Adjust protoype.
4597 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
4598 string is constant but contains no NUL byte.
4600 * expr.c (string_constant): Adjust function comment.
4601 Remove bogus check for zero termination.
4603 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
4605 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
4606 (get_constant_size): Don't make STRING_CSTs larger than they are.
4607 (check_string_literal): New check function for STRING_CSTs.
4608 (output_constant): Use it.
4610 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
4613 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
4615 2018-09-13 Richard Biener <rguenther@suse.de>
4617 PR tree-optimization/87263
4618 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
4619 (struct unwind_state): Add max_rpo field.
4620 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
4621 Compute max_rpo, the max RPO number a block can be backwards reached
4622 from. Re-write non-iterating mode to a RPO ordered worklist approach,
4623 separating it from the iterating mode.
4625 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4627 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
4628 (rfs_decision): New scheduling decision.
4630 2018-09-13 Richard Biener <rguenther@suse.de>
4633 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
4634 (vn_nary_op_insert_pieces_predicated): Do not write useless
4635 valid_dominated_by_p entry outside of the allocated storage.
4637 2018-09-13 Omar Sandoval <osandov@osandov.com>
4638 Tom de Vries <tdevries@suse.de>
4641 * dwarf2out.c (is_c): New function.
4642 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
4644 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
4647 * config/aarch64/aarch64.md (*aarch64_bfxil):
4649 * config/aarch64/constraints.md (Ulc): Define.
4650 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
4652 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
4655 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
4657 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
4658 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
4659 aarch64_layout_frame call.
4660 (aarch64_expand_epilogue): Likewise.
4661 (aarch64_initial_elimination_offset): Likewise.
4662 (aarch64_get_separate_components): Likewise.
4663 (aarch64_use_return_insn_p): Likewise.
4664 (aarch64_layout_frame): Remove unneeded check.
4666 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4668 * configure.ac: Only append
4669 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
4670 gcc_config_arguments if it was never reconfigured or last reconfigure
4671 was with different arguments.
4672 * configure: Regenerated.
4674 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4675 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4678 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
4679 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
4681 2018-09-13 Jakub Jelinek <jakub@redhat.com>
4683 PR tree-optimization/87287
4684 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
4685 X % C == 0 to X % (unsigned) C == 0 optimization to ...
4686 * match.pd (X % C == 0): ... here. New optimization.
4688 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4691 * expr.h (maybe_optimize_mod_cmp): Declare.
4692 * expr.c (mod_inv): New function.
4693 (maybe_optimize_mod_cmp): New function.
4694 (do_store_flag): Use it.
4695 * cfgexpand.c (expand_gimple_cond): Likewise.
4697 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
4698 Julian Brown <julian@codesourcery.com>
4701 * gimplify.c (gimplify_scan_omp_clauses): Set
4702 target_firstprivatize_array_bases in OpenACC parallel and kernels
4703 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
4704 OpenACC data regions.
4706 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
4708 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
4709 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
4711 2018-09-12 Richard Biener <rguenther@suse.de>
4713 PR tree-optimization/87280
4714 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
4715 edge but unreachable target.
4716 (do_rpo_vn): For conservatively handling a PHI only mark
4717 the backedge executable but not the block reachable.
4719 2018-09-12 Richard Biener <rguenther@suse.de>
4721 PR tree-optimization/87266
4722 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
4725 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
4727 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
4729 ("trunc<BFP:mode><DFP_ALL:mode>2")
4730 ("trunc<DFP_ALL:mode><BFP:mode>2")
4731 ("extend<BFP:mode><DFP_ALL:mode>2")
4732 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
4733 according to the target operand type.
4735 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4736 Andreas Krebbel <krebbel@linux.ibm.com>
4738 PR tree-optimization/86844
4739 * gimple-ssa-store-merging.c
4740 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
4741 there are any overlapping stores in between them, make sure they are
4742 also coalesced or we give up completely.
4744 2018-09-12 Jakub Jelinek <jakub@redhat.com>
4747 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
4748 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
4750 2018-09-12 Tom de Vries <tdevries@suse.de>
4752 * common.opt (gdescribe-dies): Add option.
4753 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
4754 attribute for artifical and nameless decls.
4755 (dwarf2out_register_external_die): Add description attribute to
4756 external reference die.
4757 (add_desc_attribute): New functions.
4758 (gen_subprogram_die): Add description attribute to
4759 DW_TAG_call_site_parameter.
4760 * tree-pretty-print.c (print_generic_expr_to_str): New function.
4761 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
4762 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
4764 (@item -gdescribe-dies): Add.
4766 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4768 * tree-vrp.c (vrp_shift_undefined_p): Remove.
4769 (extract_range_from_binary_expr_1: Call
4770 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
4771 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
4774 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
4776 * gimple-ssa-warn-alloca.c
4777 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
4778 field for ALLOCA_BOUND_*_LARGE.
4780 2018-09-11 Nathan Sidwell <nathan@acm.org>
4782 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
4784 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4786 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
4787 for clobbers. Remove obsolete comment.
4789 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4791 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
4792 mpxchk, mpxld and mpxst types.
4793 (define_attr length_immediate): Remove all processing of mpx types.
4794 (define_attr prefix_0f): Ditto.
4795 (define_attr memory): Ditto.
4797 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
4799 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
4800 (log<mode>2): Change operand 1 predicate to general_operand.
4801 Extend operand 1 to XFmode and generate logxf3 insn.
4802 (log10<mode>2): Change operand 1 predicate to general_operand.
4803 Extend operand 1 to XFmode and generate log10xf3 insn.
4804 (log2<mode>2): Change operand 1 predicate to general_operand.
4805 Extend operand 1 to XFmode and generate log2xf3 insn.
4806 (fyl2xp1_extend<mode>xf3_i387): Remove.
4807 (log1p<mode>2): Change operand 1 predicate to general_operand.
4808 Extend operand 1 to XFmode and generate log1pxf3 insn.
4809 (fxtract_extend<mode>xf3_i387): Remove.
4810 (logb<mode>2): Change operand 1 predicate to general_operand.
4811 Extend operand 1 to XFmode and generate logbxf3 insn.
4812 (ilogb<mode>2): Change operand 1 predicate to general_operand.
4813 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
4814 (significand<mode>2): Change operand 1 predicate to general_operand.
4815 Extend operand 1 to XFmode and generate significandxf3 insn.
4817 2018-09-11 Nathan Sidwell <nathan@acm.org>
4819 * gcc.c (perror_with_name, pfatal_with_name): Delete.
4820 (load_specs): Use fatal_error.
4821 (DELETE_IF_ORDINARY, process_command): Use error.
4822 (execute, run_attempt): Use fatal_error.
4824 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
4826 * diagnostic-core.h (sorry_at): New prototype.
4827 * diagnostic.c (sorry_at): New function.
4829 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
4831 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
4832 by zero as VR_UNDEFINED.
4834 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4836 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
4837 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
4838 (<sincos>mode2): New expander.
4839 (sincos_extend<mode>xf3_i387): Remove insn pattern.
4840 (sincos -> sin, cos splitters): Remove splitter patterns.
4841 (sincos<mode>3): Change operand 2 predicate to general_operand.
4842 Extend operand 2 to XFmode and generate sincosxf3 insn.
4843 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
4844 Change operand 3 predicate to const1_operand.
4845 (fptan_extend<mode>xf4_i387): Remove insn pattern.
4846 (tanxf2): Update operands in the call to fptanxf4_i387.
4847 (tan<mode>2): Change operand 1 predicate to general_operand.
4848 Extend operand 1 to XFmode and generate tanxf3 insn.
4849 (atan2xf3): Rename from *fpatanxf3_i387.
4850 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
4851 (atan2xf3): Remove expander.
4852 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
4853 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
4854 (atan<mode>2): Change operand 1 predicate to general_operand.
4855 Extend operand 1 to XFmode and generate atanxf3 insn.
4857 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
4859 * config/i386/i386.md (x87/SSE constant load splitter): Use
4860 memory_operand instead of nonimmediate_operand for input operand
4863 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
4865 * config/i386/i386.md (float partial SSE register stall splitter): Move
4866 splitter near its instruction pattern.
4867 (float_extend partial SSE register stall splitter): Ditto.
4868 (float_truncate partial SSE register stall splitter): Ditto.
4870 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
4873 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
4874 to speculation_safe_value_not_needed.
4877 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
4878 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
4879 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
4880 leaf_function_p, instead use has_hard_reg_initial_val.
4882 2018-09-09 Nathan Sidwell <nathan@acm.org>
4884 * gcc.h (pfatal_with_name): Don't declare here.
4885 * gcc.c (pfatal_with_name): Make static.
4887 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
4889 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
4892 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
4894 PR rtl-optimization/85458
4895 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
4896 priority hook to reduce the priority of EXPR.
4898 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
4900 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
4901 DImode for x87 on 32bit targets. Conditionally disable x87 modes
4902 with X87_ENABLE_FLOAT. Remove preparation code.
4903 (*float<SWI48:mode><MODEF:mode>2): Rename from
4904 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
4905 math using "enabled" attribute.
4906 (*floatdi<MODEF:mode>2_i387): Rename from
4907 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
4908 enable for 32bit targets only.
4909 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
4911 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
4912 as operand 1 predicate. Rewrite as define_insn_and_split.
4913 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
4915 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
4917 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
4918 to fallthru to FLOAT case.
4920 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4923 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
4924 around folding of vec_sl to handle out of range shift values.
4926 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
4928 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
4929 Update callers to gen_fix_trunc<mode>_i387_fisttp
4930 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
4931 nonimmediate_operand.
4932 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
4933 and corresponding splitters.
4934 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
4935 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
4936 (fix_truncdi_i387_with_temp): Remove insn pattern
4937 and corresponding splitters.
4938 (fix_trunc<mode>_i387): Change operand 0 predicate to
4939 nonimmediate_operand.
4940 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
4941 and corresponding splitters.
4942 (*fistdi2_1): Remove.
4944 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
4945 (lrintxfdi2): Remove expander. Reimplement as define_insn.
4946 (*fist<mode>2_1): Remove.
4947 (fist<mode>2): Ditto.
4948 (fist<mode>2_with_temp): Remove insn pattern and corresponding
4950 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
4951 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
4952 (fistdi2_<rounding>): Change operand 0 predicate to
4953 nonimmediate_operand.
4954 (fistdi2_<rounding>_with_temp): Remove insn pattern
4955 and corresponding splitters.
4956 (fist<mode>2_<rounding>): Change operand 0 predicate to
4957 nonimmediate_operand.
4958 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
4959 and corresponding splitters.
4961 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
4963 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4965 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
4968 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4970 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4971 early gimple folding of vec_splat().
4972 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
4973 * gimple-fold.h: Add an extern define for tree_vec_extract().
4975 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
4977 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
4978 wrappers around TREE_TYPE comparisons.
4980 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4983 * config/s390/predicates.md: Add nonsym_memory_operand.
4984 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
4985 contains a SYMBOL_REF, load it into an intermediate pseudo.
4986 (s390_emit_compare_and_swap): Legitimize operand.
4987 * config/s390/s390.md: Use the new nonsym_memory_operand
4988 with UNSPECV_CAS patterns.
4990 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
4993 * config/s390/s390-passes.def: New file.
4994 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
4996 (make_pass_s390_early_mach): Add declaration.
4997 * config/s390/s390.c (make_pass_s390_early_mach):
4998 (s390_option_override): Remove dynamic registration.
4999 * config/s390/t-s390: Add s390-passes.def.
5001 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
5003 * config/s390/s390.c (s390_decompose_constant_pool_ref):
5004 Remove UNSPEC_LTREL_BASE check.
5005 (annotate_constant_pool_refs): Likewise.
5006 (find_constant_pool_ref): Likewise.
5007 (find_ltrel_base): Removed.
5008 (replace_ltrel_base): Removed.
5009 (s390_mainpool_finish): Remove replace_ltrel_base call.
5010 (s390_chunkify_start): Remove pending LTREL_BASE logic.
5011 (s390_chunkify_finish): Remove replace_ltrel_base call.
5012 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
5014 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
5017 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
5018 to speculation_safe_value_not_needed.
5020 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
5021 Bernd Schmidt <bernds_cb1@t-online.de>
5023 * config/nvptx/nvptx-opts.h: New file.
5024 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
5025 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
5027 (TARGET_SM35): New macro.
5028 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
5030 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
5032 (misa=): New option.
5033 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
5035 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
5037 * config/i386/i386.md (truncdfsf2): Remove expander.
5038 (truncdfsf2_with_temp): Ditto.
5039 (truncxf<mode>2): Ditto.
5040 (*truncdfsf_fast_mixed): Remove insn pattern.
5041 (*truncdfsf_fast_i387): Ditto.
5042 (*truncdfsf_mixed): Ditto.
5043 (*truncdfsf_i387): Ditto.
5044 (*truncdfsf2_i387_1): Ditto.
5045 (*truncxfsf2_mixed): Ditto.
5046 (*truncxfdf2_mixed): Ditto.
5047 (*truncxf<mode>2_i387_noop): Ditto. Update callers
5048 to call gen_truncxf<mode>2 instead.
5049 (*truncxf<mode>2_i387): Remove.
5050 (reg->reg splitters): Remove splitter pattern.
5051 (reg->mem splitters): Ditto.
5053 (truncdfsf2): New insn pattern.
5054 (truncxf<mode>2): Ditto.
5056 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5058 * tree-ssa-math-opts.c (is_mult_by): New function.
5059 (is_square_of): Use the above.
5060 (optimize_recip_sqrt): New function.
5061 (pass_cse_reciprocals::execute): Use the above.
5063 2018-09-05 Richard Biener <rguenther@suse.de>
5066 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
5067 to zero-init the emplaced vec.
5069 2018-09-05 Martin Liska <mliska@suse.cz>
5071 PR tree-optimization/87205
5072 * tree-switch-conversion.c (pass_lower_switch::execute):
5073 Group cases for switch statements.
5075 2018-09-05 Richard Biener <rguenther@suse.de>
5077 PR tree-optimization/87217
5078 * tree-ssa-sccvn.c (vuse_valueize): New.
5079 (vn_reference_lookup_pieces): Use it.
5080 (vn_reference_lookup): Likewise.
5082 2018-09-05 Nathan Sidwell <nathan@acm.org>
5085 * stor-layout.c (place_field): Scan forwards to check last
5086 bitfield when ms_bitfield_placement is in effect.
5088 2018-09-05 Richard Biener <rguenther@suse.de>
5091 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
5094 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
5095 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
5097 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
5098 * config.gcc (extra_objs): Build it.
5099 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
5101 * config/aarch64/aarch64-passes.def
5102 (pass_tag_collision_avoidance): New pass.
5103 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
5104 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
5105 (aarch64_classify_address): Remove static qualifier.
5106 (aarch64_address_info, aarch64_address_type): Move to...
5107 * config/aarch64/aarch64-protos.h: ... here.
5108 (make_pass_tag_collision_avoidance): New function.
5109 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
5112 2018-09-05 Martin Liska <mliska@suse.cz>
5114 * doc/gcov.texi: Update documentation of humar
5116 * gcov.c (format_count): Print one decimal place, it provides
5117 more fine number of situations like '1G' vs. '1.4G'.
5119 2018-09-05 Martin Liska <mliska@suse.cz>
5122 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
5123 * optc-gen.awk: Allow 'Var' for Deprecated options in order
5124 to generate a MASK value.
5126 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
5129 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
5130 if frame pointer isn't used.
5131 (compute_frame_pointer_to_fb_displacement): Likewise.
5133 2018-09-04 Jakub Jelinek <jakub@redhat.com>
5136 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
5137 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
5138 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
5139 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
5140 and OPTION_MASK_ISA_XSAVEC_UNSET.
5142 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
5144 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
5145 XOR operations in NAND case.
5147 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
5149 * wide-int-range.cc (wide_int_range_convert): New.
5150 * wide-int-range.h (wide_int_range_convert): New.
5151 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
5152 code into wide_int_range_convert.
5153 (extract_range_into_wide_ints): Do not munge anti range constants
5154 into the entire domain. Just return the range back.
5156 2018-09-04 Martin Liska <mliska@suse.cz>
5158 * genmatch.c (output_line_directive): Add new argument
5160 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
5162 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
5164 * doc/invoke.texi (Option Summary): Add whitespace.
5166 * doc/invoke.texi (Option Summary): Add -Waligned-new.
5168 2018-09-04 Richard Biener <rguenther@suse.de>
5170 PR tree-optimization/87211
5171 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
5172 backedge value we're supposed to treat as VARYING also number
5173 the PHI to VARYING in case it got a different value-number already.
5175 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
5177 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
5178 (extract_range_from_binary_expr_1): Do not call
5179 vrp_can_optimize_bit_op.
5180 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
5182 (wide_int_range_get_mask_and_bounds): New.
5183 (wide_int_range_optimize_bit_op): New.
5184 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
5185 (wide_int_range_bit_and): Same.
5186 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
5187 (wide_int_range_optimize_bit_op): New.
5188 (wide_int_range_get_mask_and_bounds): New.
5190 2018-09-04 Richard Biener <rguenther@suse.de>
5192 PR tree-optimization/87176
5193 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
5194 variable. When value-numbering a virtual PHI node make sure
5195 to not value-number to the backedge value.
5197 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
5199 * doc/extend.texi (Long Long, Hex Floats): Document support for
5200 long long and hex floats in more recent versions of ISO C++.
5202 2018-09-03 Richard Biener <rguenther@suse.de>
5204 PR tree-optimization/87177
5205 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
5208 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5210 * bb-reorder.c (edge_order): Convert to C-qsort-style
5211 tri-state comparator.
5212 (reorder_basic_blocks_simple): Change std::stable_sort to
5215 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5217 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
5218 tri-state comparator.
5219 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
5221 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5223 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
5224 (mergesort): ... here as maximum count for using netsort.
5225 (gcc_qsort): Set nlim to 3 if stable sort is requested.
5226 (gcc_stablesort): New.
5227 * system.h (gcc_stablesort): Declare.
5229 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
5231 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
5232 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
5233 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
5235 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
5237 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
5238 lxsdx and stxsdx alternatives.
5239 (*mov<mode>_hardfloat64): Ditto.
5240 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
5242 2018-09-03 Richard Biener <rguenther@suse.de>
5244 PR tree-optimization/87200
5245 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
5248 2018-09-03 Martin Liska <mliska@suse.cz>
5250 PR tree-optimization/87201
5251 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
5252 Fix parenthesis in an expression.
5254 2018-09-03 Richard Biener <rguenther@suse.de>
5256 PR tree-optimization/87197
5257 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
5258 visited. CSE the VN_INFO hashtable lookup.
5260 PR tree-optimization/87169
5261 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
5262 iterating make sure there's no extra backedges from irreducible
5263 regions feeding the header. Mark the destination block
5266 2018-09-03 Martin Liska <mliska@suse.cz>
5269 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
5270 * common/common-targhooks.c (default_get_valid_option_values):
5272 * common/common-targhooks.h (default_get_valid_option_values):
5274 * common/config/i386/i386-common.c: Move processor_target_table
5276 (ix86_get_valid_option_values): New function.
5277 (TARGET_GET_VALID_OPTION_VALUES): New macro.
5278 * config/i386/i386.c (struct ptt): Move to i386-common.c.
5279 (PTA_*): Move all defined masks into i386-common.c.
5280 (ix86_function_specific_restore): Use new processor_cost_table.
5281 * config/i386/i386.h (struct ptt): Moved from i386.c.
5282 (struct pta): Likewise.
5283 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
5284 * doc/tm.texi.in: Likewise.
5285 * opt-suggestions.c (option_proposer::suggest_option):
5286 Pass prefix to build_option_suggestions.
5287 (option_proposer::get_completions): Likewise.
5288 (option_proposer::build_option_suggestions): Use the new target
5290 * opts.c (struct option_help_tuple): New struct.
5291 (print_filtered_help): Use the new target hook.
5293 2018-09-03 Martin Liska <mliska@suse.cz>
5296 * predict.c (set_even_probabilities): Add likely_edges
5297 argument and handle cases where we have precisely one
5299 (combine_predictions_for_bb): Catch also likely_edges.
5300 (tree_predict_by_opcode): Handle gswitch statements.
5301 * tree-cfg.h (find_case_label_for_value): New declaration.
5302 (find_taken_edge_switch_expr): Likewise.
5303 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
5304 Find pivot in decision tree based on probabily, not by number of
5307 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
5309 * doc/standards.texi (Standards): Update Objective-C reference.
5311 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5313 * doc/install.texi (Prerequisites): Update link for MPC.
5315 2018-09-01 Michael Matz <matz@suse.de>
5317 PR tree-optimization/87074
5318 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
5319 PHIs for outer-loop uses.
5321 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5323 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
5324 * doc/invoke.texi (C Dialect Options): Ditto.
5326 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
5328 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
5330 2018-08-31 Richard Biener <rguenther@suse.de>
5332 PR tree-optimization/87168
5333 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
5334 (rpo_elim::eliminate_avail): When OP was not visited it must
5337 2018-08-31 David Malcolm <dmalcolm@redhat.com>
5339 * tree-vrp.c (copy_value_range): Convert param "from" from
5340 "value_range *" to "const value_range *".
5341 (range_is_null): Likewise for param "vr".
5342 (range_int_cst_p): Likewise.
5343 (range_int_cst_singleton_p): Likewise.
5344 (symbolic_range_p): Likewise.
5345 (value_ranges_intersect_p): Likewise for both params.
5346 (value_range_nonnegative_p): Likewise for param "vr".
5347 (value_range_constant_singleton): Likewise.
5348 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
5349 (extract_range_into_wide_ints): Likewise for param "vr".
5350 (extract_range_from_multiplicative_op): Likewise for params "vr0"
5352 (vrp_can_optimize_bit_op): Likewise.
5353 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
5355 (extract_range_from_unary_expr): Likewise.
5356 (debug_value_range): Likewise for param "vr".
5357 (value_range::dump): Add "const" qualifier.
5358 (vrp_prop::check_array_ref): Convert local "vr" from
5359 "value_range *" to "const value_range *".
5360 (vrp_prop::check_mem_ref): Likewise.
5361 (vrp_prop::visit_stmt): Likewise for local "old_vr".
5362 (vrp_intersect_ranges_1): Likewise for param "vr_1".
5363 (vrp_intersect_ranges): Likewise.
5364 (simplify_stmt_for_jump_threading): Likewise for local "vr".
5365 (vrp_prop::vrp_finalize): Likewise.
5366 * tree-vrp.h (value_range::dump): Add "const" qualifier.
5367 (vrp_intersect_ranges): Add "const" qualifier to params as above.
5368 (extract_range_from_unary_expr): Likewise.
5369 (value_range_constant_singleton): Likewise.
5370 (symbolic_range_p): Likewise.
5371 (copy_value_range): Likewise.
5372 (extract_range_from_binary_expr_1): Likewise.
5373 (range_int_cst_p): Likewise.
5374 (vrp_set_zero_nonzero_bits): Likewise.
5375 (range_int_cst_singleton_p): Likewise.
5377 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
5379 * config/aarch64/arm_neon.h (vabsd_s64): New.
5380 (vnegd_s64): Likewise.
5382 2018-08-31 Martin Jambor <mjambor@suse.cz>
5384 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
5386 2018-08-31 Martin Liska <mliska@suse.cz>
5388 * ipa-icf.c (sem_item::add_type): Use
5389 sem_item::m_type_hash_cache.
5390 * ipa-icf.h: Move the cache from sem_item_optimizer
5393 2018-08-31 Nathan Sidwell <nathan@acm.org>
5395 * doc/extend.texi (Backwards Compatibility): Remove implicit
5396 extern C leeway of () being (...).
5398 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5400 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
5402 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
5406 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
5408 2018-08-31 Jakub Jelinek <jakub@redhat.com>
5411 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
5412 gen_int_mode. Formatting fixes.
5414 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
5416 * target.def (custom_function_descriptors): Improve documentation.
5417 * doc/tm.texi.in (Trampolines): Expand discussion of function
5418 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
5419 beginning of the section.
5420 * doc/tm.texi: Regenerated.
5422 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
5424 * cfg.h (class auto_edge_flag): Spell out the template-id of the
5425 base class in the initializer list. This is a workaround for
5426 building with older GCC.
5427 (class auto_bb_flag): Likewise.
5429 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
5431 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
5432 (altivec_vcmpequ<VI_char>_p): Remove star.
5433 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
5435 (expand_strncmp_vec_sequence): New function.
5436 (emit_final_str_compare_vec): New function.
5437 (expand_strn_compare): Add support for vector strncmp.
5438 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
5439 length specification to bytes.
5440 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
5441 (vcmpnezb_p): New pattern.
5442 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
5443 for option -mstring-compare-inline-limit.
5445 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
5447 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
5448 (PTA_SKYLAKE): Add PTA_AES.
5449 (PTA_GOLDMONT): Likewise.
5451 2018-08-29 Jan Hubicka <jh@suse.cz>
5454 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
5455 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
5457 2018-08-29 Jan Hubicka <jh@suse.cz>
5459 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
5461 (hash_tree): Do not visit TYPE_STUB_DECL.
5462 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
5463 stream TYPE_STUB_DECL.
5464 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
5465 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
5466 after free_lang_data.
5467 (type_in_anonymous_namespace_p): Likewise.
5469 2018-08-29 Jan Hubicka <jh@suse.cz>
5471 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5472 comment that it has to be even number.
5473 (class sreal): Change m_sig type to int32_t.
5474 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5475 int64_t for temporary calculations.
5476 (sreal_verify_basics): Drop one bit from minimum and maximum.
5478 2018-08-30 Richard Biener <rguenther@suse.de>
5480 PR tree-optimization/87147
5481 * tree-ssa-sccvn.c (SSA_VISITED): New function.
5482 (visit_phi): When the degenerate result is from the backedge and
5483 we didn't visit its definition yet drop to VARYING.
5484 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
5486 2018-08-29 Jan Hubicka <jh@suse.cz>
5488 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
5490 (hash_tree): Likewise.
5492 2018-08-29 Jan Hubicka <jh@suse.cz>
5494 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
5495 and TYPE_NEXT_REF_TO.
5497 2018-08-29 Jan Hubicka <jh@suse.cz>
5499 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
5500 comment that it has to be even number.
5501 (class sreal): Change m_sig type to int32_t.
5502 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
5503 int64_t for temporary calculations.
5504 (sreal_verify_basics): Drop one bit from minimum and maximum.
5506 2018-08-30 Tamar Christina <tamar.christina@arm.com>
5508 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
5510 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
5513 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
5514 if to_add is negative.
5516 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
5519 * builtins.c (c_strlen): Improve range checks.
5521 2018-08-29 Martin Sebor <msebor@redhat.com>
5522 Jeff Law <law@redhat.com>
5524 PR tree-optimization/86714
5525 PR tree-optimization/86711
5526 * builtins.c (c_strlen): Add arguments to call to string_constant.
5527 * expr.c (string_constant): Add argument. Detect missing nul
5528 terminator and outermost declaration it's missing in.
5529 * expr.h (string_constant): Add argument.
5530 * fold-const.c (read_from_constant_string): Add arguments to call to
5532 (c_getstr): Likewise.
5533 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5535 * tree-ssa-strlen.c (get_stridx): Likewise.
5537 2018-08-29 Jan Hubicka <jh@suse.cz>
5539 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
5540 Do not stream DECL_VINDEX.
5541 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
5542 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
5543 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
5545 2018-08-29 Richard Biener <rguenther@suse.de>
5547 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
5548 virtual operands that are not default defs to honor region
5550 (rpo_vn_valueize): Remove ineffective code here.
5552 2018-08-29 Richard Biener <rguenther@suse.de>
5554 PR tree-optimization/87132
5555 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
5556 when skipping defs reachable over backedges.
5558 2018-08-29 Richard Biener <rguenther@suse.de>
5560 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
5561 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
5562 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
5563 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
5564 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
5565 (vn_reference_lookup_3): Remove use of const_parms.
5566 (free_rpo_vn): Do not free const_parms.
5567 (do_rpo_vn): Do not call init_const_parms.
5568 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
5569 SSA_NAME_POINTS_TO_READONLY_MEMORY.
5570 (call_may_clobber_ref_p_1): Likewise.
5572 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
5575 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
5577 (-ftree-scev-cprop): Document.
5579 2018-08-29 Jan Hubicka <jh@suse.cz>
5581 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
5583 (sreal constructor): Update.
5584 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
5585 sreal:operator/): Update.
5587 2018-08-29 Martin Liska <mliska@suse.cz>
5589 * tree-switch-conversion.c (switch_conversion::expand):
5590 Strenghten assumption about gswitch statements.
5592 2018-08-29 Richard Biener <rguenther@suse.de>
5594 PR tree-optimization/87117
5595 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
5596 re-value-number released SSA VDEFs.
5598 2018-08-29 Richard Biener <rguenther@suse.de>
5600 PR tree-optimization/87126
5601 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
5603 2018-08-28 Jim Wilson <jimw@sifive.com>
5605 * config/riscv/pic.md: Rewrite.
5606 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
5608 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
5609 (SOFTF, default_load, softload, softstore): New.
5611 2018-08-28 Jeff Law <law@redhat.com>
5613 * fold-const.c (fold_binary_loc): Remove recently added assert.
5615 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
5617 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
5618 to OP parmeter of generated function.
5620 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
5622 PR tree-optimization/87009
5623 * match.pd: Add boolean optimizations.
5625 2018-08-28 Martin Sebor <msebor@redhat.com>
5628 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
5629 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
5630 (pass_walloca::gate): Use it.
5631 (alloca_call_type): Same.
5632 (pass_walloca::execute): Same.
5633 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
5635 2018-08-28 David Malcolm <dmalcolm@redhat.com>
5637 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
5638 GCC_VERSION for usage of "__gcc_dump_printf__" format from
5641 2018-08-28 Richard Biener <rguenther@suse.de>
5643 PR tree-optimization/87124
5644 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
5645 constants before looking up avail.
5647 2018-08-28 Jakub Jelinek <jakub@redhat.com>
5650 * calls.c (maybe_warn_nonstring_arg): Punt early if
5651 warn_stringop_overflow is zero. Don't call get_range_strlen
5652 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
5653 Swap comparison operands to have constants on rhs. Only use
5654 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
5655 increment lenrng[0].
5657 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
5659 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
5660 use of tree_to_shwi. Remove duplicated test for the size being
5661 a whole number of bytes.
5663 2018-08-28 Richard Biener <rguenther@suse.de>
5665 PR tree-optimization/87117
5666 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
5667 Handle removed stmt without LHS (GIMPLE_NOP).
5669 2018-08-28 Richard Biener <rguenther@suse.de>
5671 PR tree-optimization/87117
5672 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
5673 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
5675 2018-08-28 Richard Biener <rguenther@suse.de>
5677 PR tree-optimization/87117
5678 * tree-ssa-pre.c (compute_avail): Do not make expressions
5679 with predicated values available.
5680 (get_expr_value_id): Assert we do not run into predicated value
5683 2018-08-28 Richard Biener <rguenther@suse.de>
5685 PR tree-optimization/87117
5686 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
5687 get virtual operands.
5688 (get_expr_operands): Handle STRING_CST like other decls.
5690 2018-08-28 Martin Liska <mliska@suse.cz>
5692 * tree.h: Update documentation of fndecl_built_in_p
5696 2018-08-27 Jeff Law <law@redhat.com>
5697 PR tree-optimization/87110
5698 * tree-ssa-dse.c (compute_trims): Handle non-constant
5701 2018-08-27 Martin Sebor <msebor@redhat.com>
5703 PR tree-optimization/86914
5704 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
5706 2018-08-27 Martin Sebor <msebor@redhat.com>
5708 PR tree-optimization/87112
5709 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
5710 the type of the bound argument.
5712 2018-08-27 Jeff Law <law@redhat.com>
5714 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
5715 type does not have a TYPE_SIZE_UNIT.
5717 2018-08-27 Steve Ellcey <sellcey@cavium.com>
5719 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
5720 with include of backend.h.
5722 2018-08-27 Richard Biener <rguenther@suse.de>
5724 PR tree-optimization/86927
5725 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
5726 use const cond reduction code.
5728 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
5730 PR tree-optimization/85758
5731 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
5733 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5736 * diagnostic-show-locus.c (class layout_range): Update for
5737 conversion of show_caret_p to a tri-state.
5738 (layout_range::layout_range): Likewise.
5739 (make_range): Likewise.
5740 (layout::maybe_add_location_range): Likewise.
5741 (layout::should_print_annotation_line_p): Don't show annotation
5742 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
5743 (layout::get_state_at_point): Update for conversion of
5744 show_caret_p to a tri-state. Bail out early for
5745 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
5746 underlining or source colorization.
5747 (gcc_rich_location::add_location_if_nearby): Update for conversion
5748 of show_caret_p to a tri-state.
5749 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
5750 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
5752 (selftest::test_one_liner_labels): Likewise.
5753 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
5754 conversion of show_caret_p to a tri-state.
5755 * pretty-print.c (text_info::set_location): Likewise.
5756 * pretty-print.h (text_info::set_location): Likewise.
5757 * substring-locations.c (format_warning_n_va): Likewise.
5758 * tree-diagnostic.c (default_tree_printer): Likewise.
5759 * tree-pretty-print.c (newline_and_indent): Likewise.
5761 2018-08-27 David Malcolm <dmalcolm@redhat.com>
5764 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
5765 line above for line-insertion fix-it hints.
5766 (selftest::test_fixit_insert_containing_newline): Update the
5767 expected results, and add a test with line-numbering enabled.
5769 2018-08-27 Martin Liska <mliska@suse.cz>
5772 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
5773 params with DECL_HAS_VALUE_EXPR_P.
5775 2018-08-27 Martin Liska <mliska@suse.cz>
5777 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
5778 selected expansion strategy.
5780 2018-08-27 Martin Liska <mliska@suse.cz>
5782 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
5783 * builtins.c (is_builtin_fn): Likewise.
5784 * attribs.c (diag_attr_exclusions): Use new function
5785 fndecl_built_in_p and remove check for FUNCTION_DECL if
5787 (builtin_mathfn_code): Likewise.
5788 (fold_builtin_expect): Likewise.
5789 (fold_call_expr): Likewise.
5790 (fold_builtin_call_array): Likewise.
5791 (fold_call_stmt): Likewise.
5792 (set_builtin_user_assembler_name): Likewise.
5793 (is_simple_builtin): Likewise.
5794 * calls.c (gimple_alloca_call_p): Likewise.
5795 (maybe_warn_nonstring_arg): Likewise.
5796 * cfgexpand.c (expand_call_stmt): Likewise.
5797 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
5798 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
5799 (cgraph_node::verify_node): Likewise.
5800 * cgraphclones.c (build_function_decl_skip_args): Likewise.
5801 (cgraph_node::create_clone): Likewise.
5802 * config/arm/arm.c (arm_insert_attributes): Likewise.
5803 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
5804 * dse.c (scan_insn): Likewise.
5805 * expr.c (expand_expr_real_1): Likewise.
5806 * fold-const.c (operand_equal_p): Likewise.
5807 (fold_binary_loc): Likewise.
5808 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
5809 * gimple-low.c (lower_stmt): Likewise.
5810 * gimple-pretty-print.c (dump_gimple_call): Likewise.
5811 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
5813 * gimple.c (gimple_build_call_from_tree): Likewise.
5814 (gimple_call_builtin_p): Likewise.
5815 (gimple_call_combined_fn): Likewise.
5816 * gimplify.c (gimplify_call_expr): Likewise.
5817 (gimple_boolify): Likewise.
5818 (gimplify_modify_expr): Likewise.
5819 (gimplify_addr_expr): Likewise.
5820 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
5821 * ipa-cp.c (determine_versionability): Likewise.
5822 * ipa-fnsummary.c (compute_fn_summary): Likewise.
5823 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
5824 * ipa-split.c (visit_bb): Likewise.
5825 (split_function): Likewise.
5826 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
5827 * lto-cgraph.c (input_node): Likewise.
5828 * lto-streamer-out.c (write_symbol): Likewise.
5829 * omp-low.c (setjmp_or_longjmp_p): Likewise.
5830 (lower_omp_1): Likewise.
5831 * predict.c (strip_predict_hints): Likewise.
5832 * print-tree.c (print_node): Likewise.
5833 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
5834 * trans-mem.c (is_tm_irrevocable): Likewise.
5835 (is_tm_load): Likewise.
5836 (is_tm_simple_load): Likewise.
5837 (is_tm_store): Likewise.
5838 (is_tm_simple_store): Likewise.
5839 (is_tm_abort): Likewise.
5840 (tm_region_init_1): Likewise.
5841 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
5842 * tree-cfg.c (verify_gimple_call): Likewise.
5843 (move_stmt_r): Likewise.
5844 (stmt_can_terminate_bb_p): Likewise.
5845 * tree-eh.c (lower_eh_constructs_2): Likewise.
5846 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
5847 * tree-inline.c (remap_gimple_stmt): Likewise.
5848 (copy_bb): Likewise.
5849 (estimate_num_insns): Likewise.
5850 (fold_marked_statements): Likewise.
5851 * tree-sra.c (scan_function): Likewise.
5852 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
5853 (optimize_stack_restore): Likewise.
5854 (pass_fold_builtins::execute): Likewise.
5855 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
5856 (mark_all_reaching_defs_necessary_1): Likewise.
5857 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
5858 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
5859 (pass_forwprop::execute): Likewise.
5860 * tree-ssa-loop-im.c (stmt_cost): Likewise.
5861 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5862 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
5863 * tree-ssa-strlen.c (get_string_length): Likewise.
5864 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
5865 (find_func_aliases_for_call): Likewise.
5866 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
5867 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
5868 * tree-tailcall.c (find_tail_calls): Likewise.
5869 * tree.c (need_assembler_name_p): Likewise.
5870 (free_lang_data_in_decl): Likewise.
5871 (get_call_combined_fn): Likewise.
5872 * ubsan.c (is_ubsan_builtin_p): Likewise.
5873 * varasm.c (incorporeal_function_p): Likewise.
5874 * tree.h (DECL_BUILT_IN): Remove and replace with
5876 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
5877 (fndecl_built_in_p): New.
5879 2018-08-27 Martin Liska <mliska@suse.cz>
5881 PR tree-optimization/86847
5882 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
5883 Dump also subtree probability.
5884 (switch_decision_tree::do_jump_if_equal): New function.
5885 (switch_decision_tree::emit_case_nodes): Handle special
5886 situations in balanced tree that can be emitted much simpler.
5887 Fix calculation of probabilities that happen in tree expansion.
5888 * tree-switch-conversion.h (struct cluster): Add
5890 (struct simple_cluster): Likewise.
5891 (struct case_tree_node): Add new function has_child.
5892 (do_jump_if_equal): New.
5894 2018-08-27 Martin Liska <mliska@suse.cz>
5896 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
5897 Add new argument to bit_test_cluster constructor.
5898 (bit_test_cluster::emit): Set bits really number of values
5900 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
5901 probability argument.
5902 * tree-switch-conversion.h (struct bit_test_cluster):
5903 Add m_handles_entire_switch.
5905 2018-08-27 Martin Liska <mliska@suse.cz>
5907 PR tree-optimization/86702
5908 * tree-switch-conversion.c (jump_table_cluster::emit):
5909 Make probabilities even for values in jump table
5910 according to number of cases handled.
5911 (switch_decision_tree::compute_cases_per_edge): Pass
5912 argument to reset_out_edges_aux function.
5913 (switch_decision_tree::analyze_switch_statement): Likewise.
5914 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
5917 2018-08-27 Martin Liska <mliska@suse.cz>
5919 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
5920 cfun argument explicitly.
5921 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
5922 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
5923 function gimple_switch_default_bb.
5924 (convert_switch_statements):
5926 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
5927 * stmt.c (label_to_block_fn): Use label_to_block and pass
5928 cfun argument explicitly and use gimple_switch_label_bb.
5929 (expand_case): Likewise.
5930 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
5931 cfun argument explicitly. Likewise.
5932 (make_edges_bb): Likewise.
5933 (make_cond_expr_edges): Likewise.
5934 (get_cases_for_edge): Likewise.
5935 (make_gimple_switch_edges): Likewise.
5936 (label_to_block_fn): Likewise.
5937 (label_to_block): Likewise.
5938 (make_goto_expr_edges): Likewise.
5939 (make_gimple_asm_edges): Likewise.
5940 (main_block_label): Likewise.
5941 (group_case_labels_stmt): Likewise.
5942 (find_taken_edge_computed_goto): Likewise.
5943 (find_taken_edge_switch_expr): Likewise.
5944 (gimple_verify_flow_info): Likewise.
5945 (gimple_redirect_edge_and_branch): Likewise.
5946 (gimple_switch_label_bb): New function.
5947 (gimple_switch_default_bb): Likewise.
5948 (gimple_switch_edge): Likewise.
5949 (gimple_switch_default_edge): Likewise.
5950 * tree-cfg.h (label_to_block_fn): Remove and replace ...
5951 (label_to_block): ... with this.
5952 (gimple_switch_label_bb): New.
5953 (gimple_switch_default_bb): Likewise.
5954 (gimple_switch_edge): Likewise.
5955 (gimple_switch_default_edge): Likewise.
5956 * tree-cfgcleanup.c (convert_single_case_switch): Use
5957 new gimple functions and pass new argument to label_to_block.
5958 (cleanup_control_flow_bb):
5959 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
5960 cfun argument explicitly.
5961 (make_eh_edges): Likewise.
5962 (redirect_eh_dispatch_edge): Likewise.
5963 (lower_resx): Likewise.
5964 (lower_eh_dispatch): Likewise.
5965 (maybe_remove_unreachable_handlers): Likewise.
5966 (unsplit_eh): Likewise.
5967 (cleanup_empty_eh): Likewise.
5968 (verify_eh_edges): Likewise.
5969 (verify_eh_dispatch_edge): Likewise.
5970 * tree-ssa-dom.c (record_edge_info): Likewise.
5971 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
5972 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
5973 (thread_through_normal_block): Likewise.
5974 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
5975 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
5976 * tree-switch-conversion.c (switch_conversion::collect): Use new
5978 (switch_conversion::check_final_bb): Likewise.
5979 (switch_conversion::gather_default_values): Pass new argument
5981 (switch_conversion::build_constructors): Likewise.
5982 (switch_decision_tree::compute_cases_per_edge): Use new
5983 gimple_switch_edge function.
5984 (switch_decision_tree::analyze_switch_statement): Pass new argument
5986 (switch_decision_tree::try_switch_expansion): Use
5987 gimple_switch_default_edge.
5988 * tree-vrp.c (find_switch_asserts): Pass new argument
5990 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
5991 (vr_values::simplify_switch_using_ranges): Likewise.
5993 2018-08-27 Richard Biener <rguenther@suse.de>
5995 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
5996 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
5998 * tree-ssa-sccvn.h (struct vn_pval): New structure.
5999 (struct vn_nary_op_s): Add unwind_to member. Add
6000 predicated_values flag and put result into a union together
6001 with a linked list of vn_pval.
6002 (struct vn_ssa_aux): Add name member to make maintaining
6003 a map of SSA name to vn_ssa_aux possible. Remove no longer
6004 needed info, dfsnum, low, visited, on_sccstack, use_processed
6005 and range_info_anti_range_p members.
6006 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
6007 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
6009 (vn_valueize): New global.
6010 (vn_context_bb): Likewise.
6011 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
6012 VN_INFO_PTR_INFO): Remove.
6013 * tree-ssa-sccvn.c: ... (rewrite)
6014 (pass_fre::execute): For -O2+ initialize loops and run
6015 RPO VN in optimistic mode (iterating). For -O1 and -Og
6016 run RPO VN in non-optimistic mode.
6017 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
6018 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
6019 * doc/invoke.texi (sccvn-max-scc-size): Remove.
6020 (rpo-vn-max-loop-depth): Document.
6021 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
6022 when valuezing the VUSE signals we walked out of the region.
6023 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
6024 (phi_translate): Set VN context block to use for availability
6026 (compute_avail): Likewise.
6027 (pre_valueize): New function.
6028 (pass_pre::execute): Adjust to the RPO VN API.
6030 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
6031 (propagate_constants_for_unrolling): Remove.
6032 (tree_unroll_loops_completely): Perform value-numbering
6033 on the unrolled bodies loop parent.
6035 2018-08-27 Richard Biener <rguenther@suse.de>
6037 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
6038 for partial antic compute.
6040 2018-08-27 Jakub Jelinek <jakub@redhat.com>
6042 PR rtl-optimization/87065
6043 * combine.c (simplify_if_then_else): Formatting fix.
6044 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
6046 (known_cond): Don't return const_true_rtx for vector modes. Use
6047 CONST0_RTX instead of const0_rtx. Formatting fixes.
6049 2018-08-27 Martin Liska <mliska@suse.cz>
6051 PR gcov-profile/87069
6052 * gcov.c (process_file): Record files already processed
6053 and warn about a file being processed multiple times.
6055 2018-08-27 Martin Liska <mliska@suse.cz>
6058 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6059 Set default values for x_aarch64_*_string strings.
6060 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
6061 prefix. For -mabi do not print '=ABI' in help and use
6062 <option_value> format for -msve-vector-bits and -moverride
6065 2018-08-26 Jeff Law <law@redhat.com>
6067 * config/mips/frame-header-opt.c: Include "backend.h" rather than
6070 2018-08-26 Marek Polacek <polacek@redhat.com>
6072 PR c++/87029, Implement -Wredundant-move.
6073 * doc/invoke.texi: Document -Wredundant-move.
6075 2018-08-25 Martin Sebor <msebor@redhat.com>
6077 PR tree-optimization/87059
6078 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
6079 to the same type as the other.
6080 * fold-const.c (fold_binary_loc): Assert expectation.
6082 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
6084 * config/darwin.c (machopic_legitimize_pic_address): Clean up
6085 extraneous parentheses, dead code section and formatting.
6087 2018-08-24 David Malcolm <dmalcolm@redhat.com>
6090 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
6091 wide enough for jumps in the line-numbering to be visible.
6092 (layout::print_gap_in_line_numbering): New member function.
6093 (layout::calculate_line_spans): When using line numbering, merge
6094 line spans that are only 1 line apart.
6095 (diagnostic_show_locus): When printing line numbers, show gaps in
6096 line numbering directly, rather than printing headers.
6097 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
6098 line-numbering with multiple line spans.
6099 (selftest::test_fixit_insert_containing_newline_2): Add test of
6100 line-numbering, in which the spans are close enough to be merged.
6102 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
6104 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
6105 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
6106 * tree-vrp.c (range_is_nonnull): Remove.
6107 (range_includes_zero_p): Accept value_range instead of min/max.
6108 (extract_range_from_binary_expr_1): Do not early bail on
6110 Use range_includes_zero_p instead of range_is_nonnull.
6111 (extract_range_from_unary_expr): Use range_includes_zero_p instead
6112 of range_is_nonnull.
6113 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
6114 special case VR_ANTI_RANGE.
6115 (vrp_finalize): Same.
6116 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
6118 (range_is_nonnull): Remove.
6119 * vr-values.c (vrp_stmt_computes_nonzero): Use
6120 range_includes_zero_p instead of range_is_nonnull.
6121 (extract_range_basic): Pass value_range to range_includes_zero_p
6122 instead of range_is_nonnull.
6124 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
6126 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
6127 * emit-rtl.h (rtl_data): Remove return_bnd.
6128 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
6129 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
6130 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
6131 (POINTER_BOUNDS_MODE): Remove definition.
6132 (make_pointer_bounds_mode): Remove.
6133 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
6134 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
6135 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
6136 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
6137 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
6138 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
6139 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
6141 * config/i386/i386-modes.def (BND32, BND64): Remove.
6142 * config/i386/i386.c (dbx_register_map): Remove bound registers.
6143 (dbx64_register_map): Ditto.
6144 (svr4_dbx_register_map): Ditto.
6145 (indirect_thunk_bnd_needed): Remove.
6146 (indirect_thunks_bnd_used): Ditto.
6147 (indirect_return_bnd_needed): Ditto.
6148 (indirect_return_via_cx_bnd): Ditto.
6149 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
6150 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
6151 (output_indirect_thunk): Ditto. Remove need_prefix argument.
6152 (output_indirect_thunk_function): Remove handling of
6153 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
6154 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
6155 (ix86_save_reg): Remove handling of crtl->return_bnd.
6156 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
6157 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
6158 and UNSPEC_BNDLX_ADDR.
6159 (ix86_output_indirect_branch_via_reg): Remove handling of
6160 indirect_thunk_prefix_bnd.
6161 (ix86_output_indirect_branch_via_push): Ditto.
6162 (ix86_output_function_return): Ditto.
6163 (ix86_output_indirect_function_return): Ditto.
6164 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
6165 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
6166 (CALL_USED_REGISTERS): Ditto.
6167 (REG_ALLOC_ORDER): Update for removal of bound registers.
6168 (HI_REGISTER_NAMES): Ditto.
6169 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
6170 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
6171 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
6172 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
6173 (FIRST_PSEUDO_REG): Update.
6174 (BND): Remove mode iterator.
6175 * config/i386/predicates.md (bnd_mem_operator): Remove.
6177 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
6179 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
6182 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
6184 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
6185 the case in which the permute needs only a single element and
6186 repeats for every vector of the result. Extend that case to
6187 handle variable-length vectors.
6188 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
6190 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
6193 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
6194 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
6196 2018-08-24 Richard Biener <rguenther@suse.de>
6198 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
6199 bb_flags_allocated members.
6200 (auto_flag): New RAII class for allocating flags.
6201 (auto_edge_flag): New RAII class for allocating edge flags.
6202 (auto_bb_flag): New RAII class for allocating bb flags.
6203 * cfgloop.c (verify_loop_structure): Allocate temporary edge
6205 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
6206 in favor of temporarily allocated BB flag.
6207 * hsa-brig.c: Re-order includes.
6208 * hsa-dump.c: Likewise.
6209 * hsa-regalloc.c: Likewise.
6210 * print-rtl.c: Likewise.
6211 * profile-count.c: Likewise.
6213 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
6216 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
6219 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
6222 * wide-int-range.cc (wide_int_range_div): Do not ignore result
6223 from wide_int_range_multiplicative_op.
6225 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6227 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
6230 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
6232 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
6235 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
6237 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
6238 full GENERIC expression used for replacement.
6240 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
6242 * tree-vrp.c (abs_extent_range): Remove.
6243 (extract_range_into_wide_ints): Pass wide ints by reference.
6244 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
6245 Pass wide ints by reference in all calls to
6246 extract_range_into_wide_ints.
6247 * wide-int-range.cc (wide_int_range_div): New.
6248 * wide-int-range.h (wide_int_range_div): New.
6249 (wide_int_range_includes_zero_p): New.
6250 (wide_int_range_zero_p): New.
6252 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
6254 * config/aarch64/aarch64.md (arches): New enum.
6255 (arch): New enum attr.
6256 (arch_enabled): New attr.
6257 (enabled): Now uses arch_enabled only.
6258 (simd, sve, fp16): Removed attribute.
6259 (fp): Attr now defined in terms of 'arch'.
6260 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
6261 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
6262 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
6263 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
6264 attributes into 'arch'.
6265 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
6266 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
6267 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
6268 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
6269 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
6270 'simd' attribute into 'arch'.
6271 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
6272 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
6273 Convert use of 'fp' attribute to 'arch'.
6274 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
6275 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
6277 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
6278 (different modes) Merge 'fp' and 'simd' into 'arch'.
6279 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
6282 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
6284 PR rtl-optimization/87026
6285 * expmed.c (canonicalize_comparison): If we can no longer create
6286 pseudoregisters, don't.
6288 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
6291 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
6293 * config/arm/arm.c (speculation_barrier_libfunc): New static
6295 (arm_init_libfuncs): Initialize it.
6296 (arm_emit_speculation_barrier): New function.
6297 * config/arm/arm.md (speculation_barrier): Call
6298 arm_emit_speculation_barrier for architectures that do not have
6300 (speculation_barrier_insn): Only match on Armv7 or later.
6302 2018-08-23 Richard Biener <rguenther@suse.de>
6305 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
6308 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
6310 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
6311 of single-vector TBLs.
6312 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
6315 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
6318 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
6319 aarch64_evpc_tbl guard.
6321 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
6323 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
6326 2018-08-22 Martin Sebor <msebor@redhat.com>
6329 * tree-pretty-print.c (pretty_print_string): Add argument.
6330 (dump_generic_node): Call to pretty_print_string with string size.
6332 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
6334 PR rtl-optimization/86771
6335 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
6336 of two SETs into those two SETs, one to be placed at i2, if that SETs
6337 destination is modified between i2 and i3.
6339 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6341 PR tree-optimization/86725
6342 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
6344 (vect_analyze_scalar_cycles_1): Check it.
6346 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6348 PR tree-optimization/86725
6349 * tree-vect-loop.c (vect_is_simple_reduction): When treating
6350 an outer loop phi as a double reduction, make sure that the
6351 single user of the phi result is an inner loop phi.
6353 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6355 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
6356 grouped stores with gaps to a strided group.
6358 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
6360 * tree-vect-stmts.c (get_group_load_store_type)
6361 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
6362 first statement in a group.
6364 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6366 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
6367 the sequence used in gcc/gcc.c.
6369 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6372 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
6375 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6377 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
6378 Darwin10-specific unwinder-shim.
6379 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
6380 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
6381 New to cater for Darwin10 Rosetta.
6383 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6385 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
6386 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
6388 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
6393 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
6394 (dwarf2out_switch_text_section): Generate a local label for the
6395 second function sub-section and apply it as the second FDE start
6397 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
6398 second sub-section start.
6400 2018-08-22 Richard Biener <rguenther@suse.de>
6402 PR tree-optimization/86988
6403 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
6405 2018-08-22 Richard Biener <rguenther@suse.de>
6407 PR tree-optimization/86945
6408 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
6410 2018-08-22 Alexandre Oliva <oliva@adacore.com>
6412 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
6413 a comment about how uses of r2 for .sdata2 come about.
6415 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
6417 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
6419 2018-08-21 Marek Polacek <polacek@redhat.com>
6421 PR c++/86981, Implement -Wpessimizing-move.
6422 * doc/invoke.texi: Document -Wpessimizing-move.
6424 2018-08-21 Jan Hubicka <jh@suse.cz>
6426 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
6427 * tree.h (is_redundant_typedef): Remove.
6428 * dwarf2out.c (is_redundant_typedef): Turn into static function.
6430 2018-08-21 Jan Hubicka <jh@suse.cz>
6432 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
6435 2018-08-21 Tamar Christina <tamar.christina@arm.com>
6437 * expmed.c (extract_low_bits): Reject invalid subregs early.
6439 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6442 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
6445 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6447 * config/vxworks.h: Guard vxworks_asm_out_constructor and
6448 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
6449 * config/vxworks.c: Likewise.
6451 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6453 * config/vxworks.c: Set targetm.have_ctors_dtors
6454 if HAVE_INITFINI_ARRAY_SUPPORT.
6455 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
6456 if HAVE_INITFINI_ARRAY_SUPPORT.
6458 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
6460 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
6461 default search path for VxWorks < 7.
6463 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
6465 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
6466 (get_format_string): Refer to c_getstr.
6468 2018-08-21 Tom de Vries <tdevries@suse.de>
6470 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6471 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6472 (debuginfo_early_stop): Declare.
6473 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
6474 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
6475 (debuginfo_early_stop): New function.
6476 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
6477 and debuginfo_early_stop.
6478 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
6479 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
6480 (general_init): Call debuginfo_early_init.
6481 (finalize): Call debuginfo_fini.
6482 (do_compile): Call debuginfo_init.
6483 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
6485 (@item -fdump-debug, @item -fdump-earlydebug): Add.
6487 2018-08-21 Tom de Vries <tdevries@suse.de>
6489 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
6490 flag_dump_noaddr and flag_dump_unnumbered.
6492 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
6494 * wide-int-range.cc (wide_int_range_abs): New.
6495 (wide_int_range_order_set): Rename from wide_int_range_min_max.
6496 * wide-int-range.h (wide_int_range_abs): New.
6497 (wide_int_range_min_max): New.
6498 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
6499 case to call wide_int_range_abs.
6500 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
6501 (extract_range_from_abs_expr): Delete.
6503 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
6506 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
6507 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
6510 2018-08-20 Nathan Sidwell <nathan@acm.org>
6511 Jeff Law <law@redhat.com>
6513 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
6514 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
6516 2018-08-20 David Malcolm <dmalcolm@redhat.com>
6519 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
6520 (decl_attributes): Likewise.
6521 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
6523 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
6524 * diagnostic-core.h (class auto_diagnostic_group): New class.
6525 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
6526 (diagnostic_report_diagnostic): Handle the first diagnostics within
6528 (emit_diagnostic): Add auto_diagnostic_group instance.
6530 (inform_n): Likewise.
6531 (warning): Likewise.
6532 (warning_at): Likewise.
6533 (warning_n): Likewise.
6534 (pedwarn): Likewise.
6535 (permerror): Likewise.
6537 (error_n): Likewise.
6538 (error_at): Likewise.
6540 (fatal_error): Likewise.
6541 (internal_error): Likewise.
6542 (internal_error_no_backtrace): Likewise.
6543 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
6544 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
6545 * diagnostic.h (struct diagnostic_context): Add fields
6546 "diagnostic_group_nesting_depth",
6547 "diagnostic_group_emission_count", "begin_group_cb",
6549 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
6550 Add auto_diagnostic_group instance(s).
6551 (find_explicit_erroneous_behavior): Likewise.
6552 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
6553 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
6554 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
6555 (gimplify_va_arg_expr): Likewise.
6556 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
6557 (HSA_SORRY_AT): Likewise.
6558 * ipa-devirt.c (compare_virtual_tables): Likewise.
6559 (warn_odr): Likewise.
6560 * multiple_target.c (expand_target_clones): Likewise.
6561 * opts-common.c (cmdline_handle_error): Likewise.
6562 * reginfo.c (globalize_reg): Likewise.
6563 * substring-locations.c (format_warning_n_va): Likewise.
6564 * tree-inline.c (expand_call_inline): Likewise.
6565 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
6566 * tree-ssa-loop-niter.c
6567 (do_warn_aggressive_loop_optimizations): Likewise.
6568 * tree-ssa-uninit.c (warn_uninit): Likewise.
6569 * tree.c (warn_deprecated_use): Likewise.
6571 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
6574 * config/i386/i386.md (eh_return): Always update EH return
6575 address in word_mode.
6577 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
6579 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
6580 TARGET_SPLIT_COMPLEX_ARG is defined.
6582 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6584 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
6586 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
6589 * expr.c (expand_assignment): Assert that bitpos is positive.
6590 (store_field): Likewise
6591 (expand_expr_real_1): Make sure that bitpos is positive.
6592 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
6595 2018-08-20 Nathan Sidwell <nathan@acm.org>
6597 * Makefile.in (CPP_ID_DATA_H): Delete.
6598 (CPP_INTERNAL_H): Don't add it.
6599 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
6600 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
6602 2018-08-20 Richard Biener <rguenther@suse.de>
6604 PR tree-optimization/78655
6605 * tree-vrp.c (extract_range_from_binary_expr_1): Make
6606 pointer + offset nonnull if either operand is nonnull work.
6608 2018-08-20 Tom de Vries <tdevries@suse.de>
6610 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
6611 unless the referenced die describes the added property using
6612 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
6613 Otherwise, add a DW_AT_location to the referenced die.
6615 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
6618 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
6619 register_operand when calling ix86_set_reg_reg_cost.
6620 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
6621 Set *total to 0 for operands that satisfy x86_64_immediate_operand
6622 predicate and to 1 otherwise.
6624 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
6626 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
6627 emit a diagnostic that it is not supported and reset the option.
6628 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
6629 supported and consume the option. (ASM_FINAL_SPEC): New.
6631 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6633 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
6636 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
6638 C-SKY port: Documentation
6640 * doc/extend.texi (C-SKY Function Attributes): New section.
6641 * doc/invoke.texi (Option Summary): Add C-SKY options.
6642 (C-SKY Options): New section.
6643 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
6645 2018-08-17 Jojo <jijie_rong@c-sky.com>
6646 Huibin Wang <huibin_wang@c-sky.com>
6647 Sandra Loosemore <sandra@codesourcery.com>
6648 Chung-Lin Tang <cltang@codesourcery.com>
6650 C-SKY port: Backend implementation
6652 * config/csky/*: New.
6653 * common/config/csky/*: New.
6655 2018-08-17 Jojo <jijie_rong@c-sky.com>
6656 Huibin Wang <huibin_wang@c-sky.com>
6657 Sandra Loosemore <sandra@codesourcery.com>
6658 Chung-Lin Tang <cltang@codesourcery.com>
6659 Andrew Jenner <andrew@codesourcery.com>
6661 C-SKY port: Configury
6663 * config.gcc (csky-*-*): New.
6664 * configure.ac: Add csky to targets for dwarf2 debug_line support.
6665 * configure: Regenerated.
6667 2018-08-17 David Malcolm <dmalcolm@redhat.com>
6669 * dump-context.h: Include "dumpfile.h".
6670 (dump_context::dump_printf_va): Convert final param from va_list
6671 to va_list *. Convert from ATTRIBUTE_PRINTF to
6672 ATTRIBUTE_GCC_DUMP_PRINTF.
6673 (dump_context::dump_printf_loc_va): Likewise.
6674 * dumpfile.c: Include "stringpool.h".
6675 (make_item_for_dump_printf_va): Delete.
6676 (make_item_for_dump_printf): Delete.
6677 (class dump_pretty_printer): New class.
6678 (dump_pretty_printer::dump_pretty_printer): New ctor.
6679 (dump_pretty_printer::emit_items): New member function.
6680 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
6682 (dump_pretty_printer::emit_item): New member function.
6683 (dump_pretty_printer::stash_item): New member function.
6684 (dump_pretty_printer::format_decoder_cb): New member function.
6685 (dump_pretty_printer::decode_format): New member function.
6686 (dump_context::dump_printf_va): Reimplement in terms of
6687 dump_pretty_printer.
6688 (dump_context::dump_printf_loc_va): Convert final param from va_list
6690 (dump_context::begin_scope): Reimplement call to
6691 make_item_for_dump_printf.
6692 (dump_printf): Update for change to dump_printf_va.
6693 (dump_printf_loc): Likewise.
6694 (selftest::test_capture_of_dump_calls): Convert "stmt" from
6695 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
6696 with %T, %E, and %G.
6697 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
6698 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
6699 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
6700 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
6701 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
6702 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
6703 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
6704 within a dump_printf_loc call to "%wu".
6705 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
6706 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
6707 missing space after "=".
6708 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
6709 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
6710 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
6711 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
6712 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
6713 duplicate "vectorized" from message.
6715 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
6717 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
6718 polyNxK_t element's TYPE_STRING_FLAG.
6720 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
6722 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
6723 (they were unnamed before). Fix comments.
6725 2018-08-17 Nathan Sidwell <nathan@acm.org>
6727 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
6729 2018-08-17 Richard Biener <rguenther@suse.de>
6731 PR tree-optimization/86841
6732 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
6734 2018-08-17 Martin Liska <mliska@suse.cz>
6736 * common.opt: Remove Warn, Init and Report for options with
6737 Ignore/Deprecated flag. Warning is done automatically for
6739 * config/i386/i386.opt: Likewise.
6740 * config/ia64/ia64.opt: Likewise.
6741 * config/rs6000/rs6000.opt: Likewise.
6742 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
6743 Remove usage of flag_check_pointer_bounds.
6744 * lto-wrapper.c (merge_and_complain): Do not handle
6745 OPT_fcheck_pointer_bounds.
6746 (append_compiler_options): Likewise.
6747 * opt-functions.awk: Do not handle Deprecated.
6748 * optc-gen.awk: Check that Var, Report and Init are not
6749 used for an option with Ignore/Deprecated flag.
6750 * opts-common.c (decode_cmdline_option): Do not report
6752 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
6754 * opts.h (struct cl_option): Remove cl_deprecated flag.
6755 (CL_ERR_DEPRECATED): Remove error enum value.
6757 2018-08-17 Richard Biener <rguenther@suse.de>
6760 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
6761 across a va-arg-pack using call adjust its return value accordingly.
6763 2018-08-16 Martin Sebor <msebor@redhat.com>
6765 PR tree-optimization/86853
6766 * gimple-ssa-sprintf.c (struct format_result): Rename member.
6767 (struct fmtresult): Add member and initialize it in ctors.
6768 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
6769 (format_string): Handle %S the same as %ls. Set MAYFAIL.
6770 (format_directive): Set POSUNDER4K when MAYFAIL is set.
6771 (parse_directive): Handle %C same as %c.
6772 (sprintf_dom_walker::compute_format_length): Adjust.
6773 (is_call_safe): Adjust.
6775 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
6777 * builtins.c (c_strlen): Add new parameter eltsize. Use it
6778 for determining how to count the elements.
6779 * builtins.h (c_strlen): Adjust prototype.
6780 * expr.c (string_constant): Add new parameter mem_size.
6781 Set *mem_size appropriately.
6782 * expr.h (string_constant): Adjust protoype.
6783 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
6784 * gimple-fold.h (get_range_strlen): Adjust prototype.
6785 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
6786 (format_string): Call get_string_length with eltsize.
6788 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6790 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
6791 to emit the span, rather than setting it as the prefix.
6793 2018-08-16 David Malcolm <dmalcolm@redhat.com>
6795 * diagnostic-show-locus.c (layout::start_annotation_line): Add
6796 "margin_char" parameter, defaulting to space. Use it in place
6797 of pp_space for the initial part of the margin.
6798 (layout::print_leading_fixits): Use '+' when filling the margin
6799 of line-insertion fix-it hints.
6801 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6803 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
6806 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6808 * config/rs6000/altivec.md: Don't set length attribute to the default
6810 * config/rs6000/darwin.md: Ditto.
6811 * config/rs6000/dfp.md: Ditto.
6812 * config/rs6000/htm.md: Ditto.
6813 * config/rs6000/rs6000.md: Ditto.
6814 * config/rs6000/sync.md: Ditto.
6815 * config/rs6000/vsx.md: Ditto.
6817 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6819 * config/rs6000/altivec.md: Don't set length attribute to the default
6820 value, for branch instructions.
6821 * config/rs6000/darwin.md: Ditto.
6822 * config/rs6000/rs6000.md: Ditto.
6824 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
6826 * config/rs6000/rs6000.md (length): Always define as const_int 4.
6827 (unnamed conditional branch define_insn): Set length to 4 or 8
6828 depending on offset.
6829 (<bd>_<mode>): Similar, for alternative 0.
6830 (<bd>tf_<mode>): Ditto.
6832 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6834 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
6836 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
6838 * doc/rtl.texi: Replace old RTX class names with new names.
6841 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
6843 * expmed.h (canonicalize_comparison): New declaration.
6844 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
6845 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
6846 * optabs.c (prepare_cmp_insn): Likewise.
6847 * rtl.h (unsigned_condition_p): New function which checks if a
6848 comparison operator is unsigned.
6850 2018-08-16 Nathan Sidwell <nathan@acm.org>
6852 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
6853 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
6855 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6858 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
6859 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
6860 (mov<mov>): ..this and enable unconditionally.
6862 2018-08-16 Tamar Christina <tamar.christina@arm.com>
6864 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
6866 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
6868 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
6869 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
6870 "Common" with "Target".
6872 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
6874 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
6875 * doc/invoke.texi (mmitigate-rop): Remove.
6876 * config/i386/i386.c: Do not include "regrename.h".
6877 (ix86_rop_should_change_byte_p, reg_encoded_number)
6878 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
6880 (ix86_reorg): Remove call to ix86_mitigate_rop.
6881 * config/i386/i386.md (attr "modrm_class"): Remove.
6882 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
6883 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
6884 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
6886 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
6888 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
6889 allow folding of mergeh() and mergel() for the float and double types.
6890 (fold_mergehl_helper): Rework to handle building a permute tree
6893 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
6895 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
6898 2018-08-15 David Malcolm <dmalcolm@redhat.com>
6900 * common.opt (fdiagnostics-show-labels): New option.
6901 * diagnostic-show-locus.c (class layout_range): Add field
6903 (class layout): Add field "m_show_labels_p".
6904 (layout_range::layout_range): Add param "label" and use it to
6906 (make_range): Pass in NULL for new "label" param of layout_range's
6908 (layout::layout): Initialize m_show_labels_p.
6909 (layout::maybe_add_location_range): Pass in loc_range->m_label
6910 when constructing layout_range instances.
6911 (struct line_label): New struct.
6912 (layout::print_any_labels): New member function.
6913 (layout::print_line): Call it if label-printing is enabled.
6914 (selftest::test_one_liner_labels): New test.
6915 (selftest::test_diagnostic_show_locus_one_liner): Call it.
6916 * diagnostic.c (diagnostic_initialize): Initialize
6917 context->show_labels_p.
6918 * diagnostic.h (struct diagnostic_context): Add field
6920 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
6921 -fno-diagnostics-show-labels.
6922 * dwarf2out.c (gen_producer_string): Add
6923 OPT_fdiagnostics_show_labels to the ignored options.
6924 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
6926 (gcc_rich_location::maybe_add_expr): Likewise.
6927 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
6928 label" param, defaulting to NULL.
6929 (gcc_rich_location::add_expr): Add "label" param.
6930 (gcc_rich_location::maybe_add_expr): Likewise.
6931 (class text_range_label): New class.
6932 (class range_label_for_type_mismatch): New class.
6933 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
6934 of format_warning_va.
6935 (fmtwarn_n): Likewise for new params of format_warning_n_va.
6936 * lto-wrapper.c (merge_and_complain): Add
6937 OPT_fdiagnostics_show_labels to the "pick one setting" options.
6938 (append_compiler_options): Likewise to the dropped options.
6939 (append_diag_options): Likewise to the passed-on options.
6940 * opts.c (common_handle_option): Handle the new option.
6941 * selftest-diagnostic.c
6942 (test_diagnostic_context::test_diagnostic_context): Enable
6944 * substring-locations.c: Include "gcc-rich-location.h".
6945 (format_warning_n_va): Add "fmt_label" and "param_label" params
6946 and use them as appropriate.
6947 (format_warning_va): Add "fmt_label" and "param_label" params,
6948 passing them on to format_warning_n_va.
6949 (format_warning_at_substring): Likewise.
6950 (format_warning_at_substring_n): Likewise.
6951 * substring-locations.h (format_warning_va): Add "fmt_label" and
6952 "param_label" params.
6953 (format_warning_n_va): Likewise.
6954 (format_warning_at_substring): Likewise.
6955 (format_warning_at_substring_n): Likewise.
6956 * toplev.c (general_init): Initialize global_dc->show_labels_p.
6958 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
6961 * builtins.c (expand_builtin_memcmp): Do not expand the call
6962 when overflow is detected.
6964 2018-08-15 Martin Sebor <msebor@redhat.com>
6966 PR tree-optimization/71625
6967 * config/aarch64/aarch64-builtins.c
6968 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
6970 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
6972 * config/s390/s390.c (s390_reorg): Remove loop.
6974 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
6977 (darwin_function_switched_text_sections): Delete.
6978 * gcc/config/darwin.h
6979 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
6981 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
6984 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
6985 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
6986 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
6988 2018-08-15 Martin Liska <mliska@suse.cz>
6990 PR tree-optimization/86925
6991 * predict.c (expr_expected_value_1): When taking
6992 later predictor, assign also probability.
6993 Use fold_build2_initializer_loc in order to fold
6994 the expression in -frounding-math.
6996 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
6998 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
7000 (expand_vec_perm_1): Try the new method.
7002 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
7005 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
7006 Check whether lra_live_max_point is 0 before dividing.
7008 2018-08-14 Martin Sebor <msebor@redhat.com>
7010 PR tree-optimization/86650
7011 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
7012 (vrp_prop::check_mem_ref): Same.
7014 2018-08-13 Liu Hao <lh_mouse@126.com>
7016 * pretty-print.c (eat_esc_sequence): Swap the foreground and
7017 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
7018 and clear it thereafter, as it only works for DBCS.
7020 2018-08-13 Liu Hao <lh_mouse@126.com>
7022 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
7023 handle returned by _get_osf_handle().
7025 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7027 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
7028 for folding vec_perm.
7030 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7032 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
7033 Add support for gimple-folding of vec_pack() and vec_unpack()
7036 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7038 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
7039 vec_xst variants to the list.
7040 (rs6000_gimple_fold_builtin): Add support for folding unaligned
7041 vector loads and stores.
7043 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
7045 * config.gcc (rs6000-ibm-aix4.x): Delete.
7046 (rs6000-ibm-aix5.1): Delete.
7047 (rs6000-ibm-aix5.2): Delete.
7048 (rs6000-ibm-aix5.3): Delete.
7049 * config/rs6000/aix43.h: Delete.
7050 * config/rs6000/aix51.h: Delete.
7051 * config/rs6000/aix52.h: Delete.
7052 * config/rs6000/t-aix43: Delete.
7054 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
7056 * config/s390/s390.c (s390_decompose_constant_pool_ref):
7058 (s390_decompose_address): Factor out constant pool ref
7061 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7063 * config/nds32/nds32-predicates.c
7064 (nds32_can_use_bclr_p): Change return type as bool.
7065 (nds32_can_use_bset_p): Ditto.
7066 (nds32_can_use_btgl_p): Ditto.
7067 (nds32_can_use_bitci_p): Ditto.
7068 * config/nds32/nds32-protos.h
7069 (nds32_can_use_bclr_p): Change declaration.
7070 (nds32_can_use_bset_p): Ditto.
7071 (nds32_can_use_btgl_p): Ditto.
7072 (nds32_can_use_bitci_p): Ditto.
7074 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7076 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
7077 Support -msched-prolog-epilog option.
7078 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
7080 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7082 * common/config/nds32/nds32-common.c
7083 (nds32_option_optimization_table): Enalbe -malways-align.
7085 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
7087 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
7089 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
7090 OPT_misr_secure_ case.
7091 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
7092 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
7093 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
7095 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
7096 (nds32_isr_info): New field security_level.
7097 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
7098 * config/nds32/nds32.md (return_internal): Consider critical attribute.
7099 * config/nds32/nds32.opt (misr-secure): New option.
7100 * config/nds32/nds32_init.inc: New file.
7101 * config/nds32/nds32_isr.h: New file.
7103 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
7105 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
7106 Update comment for atomic instructions.
7107 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
7108 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
7110 (atomic_loaddi): Revise fence expansion to only emit fence prior to
7111 load for __ATOMIC_SEQ_CST model.
7112 (atomic_loaddi_1): Remove float register target.
7113 (atomic_storedi): Handle CONST_INT values.
7114 (atomic_storedi_1): Remove float register source. Add special case
7116 (memory_barrier): New expander and insn.
7118 2018-08-11 Jakub Jelinek <jakub@redhat.com>
7120 PR tree-optimization/86835
7121 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
7122 new_stmt after def_gsi, make sure to insert new_square_stmt after
7123 that stmt, not 2 stmts before it.
7125 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
7128 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
7131 2018-08-10 Martin Liska <mliska@suse.cz>
7134 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
7136 * builtins.c (expand_builtin_expect_with_probability):
7138 (expand_builtin_expect_with_probability): New function.
7139 (build_builtin_expect_predicate): Add new argumnet probability
7140 for BUILT_IN_EXPECT_WITH_PROBABILITY.
7141 (fold_builtin_expect):
7144 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
7145 * builtins.h (fold_builtin_expect): Set new argument.
7146 * doc/extend.texi: Document __builtin_expect_with_probability.
7147 * doc/invoke.texi: Likewise.
7148 * gimple-fold.c (gimple_fold_call): Pass new argument.
7149 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
7150 also BUILT_IN_EXPECT_WITH_PROBABILITY.
7151 * predict.c (get_predictor_value): New function.
7152 (expr_expected_value): Add new argument probability. Assume
7153 that predictor and probability are always non-null.
7154 (expr_expected_value_1): Likewise. For __builtin_expect and
7155 __builtin_expect_with_probability set probability. Handle
7156 combination in binary expressions.
7157 (tree_predict_by_opcode): Simplify code by simply calling
7158 get_predictor_value.
7159 (pass_strip_predict_hints::execute): Add handling of
7160 BUILT_IN_EXPECT_WITH_PROBABILITY.
7161 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
7163 * tree.h (DECL_BUILT_IN_P): New function.
7165 2018-08-10 Martin Liska <mliska@suse.cz>
7167 PR tree-optimization/85799
7168 * passes.def: Add argument for pass_strip_predict_hints.
7169 * predict.c (class pass_strip_predict_hints): Add new argument
7171 (strip_predictor_early): New function.
7172 (pass_strip_predict_hints::execute): Call the function to
7174 (strip_predict_hints): New function.
7175 * predict.def: Fix comment.
7177 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
7179 * Makefile.in: Clarify which tm.texi to copy over to assert the
7180 right to grant a GFDL license for all.
7182 2018-08-09 Jeff Law <law@redhat.com>
7184 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
7187 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
7189 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
7192 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
7194 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
7195 reductions for variable-length vectors.
7197 2018-08-09 David Malcolm <dmalcolm@redhat.com>
7200 * common.opt (fdiagnostics-show-line-numbers): New option.
7201 * diagnostic-show-locus.c (class layout): Add fields
7202 "m_show_line_numbers_p" and "m_linenum_width";
7203 (num_digits): New function.
7204 (test_num_digits): New function.
7205 (layout::layout): Initialize new fields. Update m_x_offset
7206 logic to handle any left margin.
7207 (layout::print_source_line): Print line number when requested.
7208 (layout::start_annotation_line): New member function.
7209 (layout::print_annotation_line): Call it.
7210 (layout::print_leading_fixits): Likewise.
7211 (layout::print_trailing_fixits): Likewise. Update calls to
7212 move_to_column for new parameter.
7213 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
7214 it to potentially call start_annotation_line.
7215 (layout::show_ruler): Call start_annotation_line.
7216 (selftest::test_line_numbers_multiline_range): New selftest.
7217 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
7218 and selftest::test_line_numbers_multiline_range.
7219 * diagnostic.c (diagnostic_initialize): Initialize
7220 show_line_numbers_p.
7221 * diagnostic.h (struct diagnostic_context): Add field
7222 "show_line_numbers_p".
7223 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
7224 -fno-diagnostics-show-line-numbers.
7225 * dwarf2out.c (gen_producer_string): Add
7226 OPT_fdiagnostics_show_line_numbers to the ignored options.
7227 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
7228 one setting" options.
7229 (append_compiler_options): Likewise to the dropped options.
7230 (append_diag_options): Likewise to the passed-on options.
7231 * opts.c (common_handle_option): Handle the new option.
7232 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
7234 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
7236 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
7237 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
7238 third argument of type "const signed char" to descriptions of
7239 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
7240 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
7241 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
7242 __builtin_bcdsub_ov functions.
7244 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
7246 PR tree-optimization/86858
7247 * tree-vect-loop.c (vect_is_simple_reduction): Restore
7248 flow_bb_inside_loop_p calls.
7250 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
7252 PR tree-optimization/86871
7253 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
7254 instead of gimple_assign_lhs.
7256 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
7259 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
7260 register constraint to operand 0.
7261 (add<mode>3_carryinC): Likewise.
7262 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
7264 2018-08-09 Martin Liska <mliska@suse.cz>
7267 * common.opt: Remove extra line.
7269 2018-08-09 Martin Liska <mliska@suse.cz>
7271 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
7272 at the end of a line, make first letter capital and end up
7273 a sentence with a dot.
7274 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
7275 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
7276 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
7277 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
7278 (PARAM_MAX_ISL_OPERATIONS): Likewise.
7279 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
7280 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
7281 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
7282 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
7283 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
7284 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
7285 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
7286 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
7287 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
7288 (PARAM_TREE_REASSOC_WIDTH): Likewise.
7289 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
7290 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
7291 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
7293 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
7296 * config/s390/s390.c (s390_option_override_internal): Reduce the
7297 stack-clash-protection-probe-interval param if it would be too big
7300 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
7303 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
7304 * config/m68k/m68k.c (handle_move_double): Don't call
7305 m68k_final_prescan_insn.
7306 (m68k_adjust_decorated_operand): Renamed from
7307 m68k_final_prescan_insn, remove first and third operand and
7309 (print_operand): Call it.
7310 (print_operand_address): Call it.
7312 2018-08-08 Nathan Sidwell <nathan@acm.org>
7314 * diagnostic.c (diagnostic_report_current_module): Use
7315 linemap_included_from & linemap_included_from_linemap.
7317 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
7319 * config/aarch64/aarch64-cores.def: Add phecda core.
7320 * config/aarch64/aarch64-tune.md: Regenerate.
7321 * doc/invoke.texi: Add phecda core.
7323 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
7326 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
7328 * config/s390/s390.md ("movti"): Add more alternatives for
7329 constant to GPR copies.
7331 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
7333 * config/s390/s390.c: Fix whitespace damage throughout the file.
7334 * config/s390/s390.h: Likewise.
7335 * config/s390/tpf.h: Likewise.
7337 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
7339 * config/s390/s390.c (s390_loadrelative_operand_p):
7340 Remove TARGET_CPU_ZARCH usages.
7341 (s390_rtx_costs): Likewise.
7342 (s390_legitimate_constant_p): Likewise.
7343 (s390_cannot_force_const_mem): Likewise.
7344 (legitimate_reload_constant_p): Likewise.
7345 (s390_preferred_reload_class): Likewise.
7346 (legitimize_pic_address): Likewise.
7347 (legitimize_tls_address): Likewise.
7348 (s390_split_branches): Removed.
7349 (s390_add_execute): Removed.
7350 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
7351 (s390_mainpool_start): Likewise.
7352 (s390_mainpool_finish): Likewise.
7353 (s390_mainpool_cancel): Removed.
7354 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
7355 (s390_chunkify_cancel): Likewise.
7356 (s390_return_addr_rtx): Likewise.
7357 (s390_register_info): Remove split_branches_pending_p uages.
7358 (s390_optimize_register_info): Likewise.
7359 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
7360 split_branches_pending_p usages.
7361 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
7362 (s390_load_got): Likewise.
7363 (s390_expand_split_stack_prologue): Likewise.
7364 (output_asm_nops): Likewise.
7365 (s390_function_profiler): Likewise.
7366 (s390_emit_call): Likewise.
7367 (s390_conditional_register_usage): Likewise.
7368 (s390_optimize_prologue): Likewise.
7369 (s390_reorg): Remove TARGET_CPU_ZARCH and
7370 split_branches_pending_p usages.
7371 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
7373 (s390_output_indirect_thunk_function): Likewise.
7374 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
7375 (TARGET_CPU_ZARCH_P): Removed.
7376 (struct machine_function): Remove split_branches_pending_p.
7377 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
7379 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
7381 * common/config/s390/s390-common.c (processor_flags_table):
7383 * config.gcc: Remove with_arch/with_tune support.
7384 * config/s390/2064.md: Remove cpu attribute comparisons.
7385 * config/s390/driver-native.c (s390_host_detect_local_cpu):
7387 * config/s390/linux.h (ASM_SPEC):
7388 Remove -march support.
7389 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
7390 Use a table to get an arch level.
7391 * config/s390/s390-opts.h (enum processor_type):
7393 * config/s390/s390.c
7394 (processor_table): Remove entries, add arch_level values.
7395 (s390_issue_rate): Remove cases.
7396 (s390_option_override): Adjust
7397 s390_option_override_internal() call.
7398 (s390_option_override_internal): Remove deprecation warning.
7399 (s390_valid_target_attribute_tree): Adjust
7400 s390_option_override_internal() call.
7401 * config/s390/s390.h (struct s390_processor):
7402 Share with s390-c.c, add arch_level field.
7403 * config/s390/s390.md:
7404 Remove occurrences in cpu attribute.
7405 * config/s390/s390.opt: Remove -march/-mtune support.
7406 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
7407 * doc/invoke.texi: Remove deprecation warning.
7409 2018-08-08 Luis Machado <luis.machado@linaro.org>
7411 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
7413 (qdf24xx_tunings): Set vector cost structure to
7414 qdf24xx_vector_cost.
7416 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
7417 <register_sextend>: Set to 3.
7419 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
7422 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
7423 * config/aarch64/aarch64-simd.md
7424 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
7425 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
7426 (aarch64_frecpx<mode>): ...this new pattern.
7427 * config/aarch64/aarch64-simd-builtins.def: Remove comment
7428 about aarch64_frecp<FRECP:frecp_suffix><mode>.
7430 2018-08-07 Martin Liska <mliska@suse.cz>
7433 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
7434 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
7435 * predict.def (PRED_MALLOC_NONNULL): New predictor.
7436 * doc/extend.texi: Document that malloc attribute adds
7439 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
7442 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7443 Define to speculation_safe_value_not_needed.
7445 2018-08-06 Jeff Law <law@redhat.com>
7447 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
7448 the vr_values instance to cprop_into_stmt.
7449 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
7450 (cprop_operand): Also query EVRP to determine if OP is a constant.
7452 2018-08-06 Nathan Sidwell <nathan@acm.org>
7454 * diagnostic.c (diagnostic_report_current_module): Reroll
7455 included-at loop. Translate text.
7457 2018-08-06 David Malcolm <dmalcolm@redhat.com>
7459 * function-tests.c (selftest::test_expansion_to_rtl): Call
7460 free_after_compilation.
7462 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7464 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
7466 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
7468 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
7469 loops with memory block operations from getting unrolled.
7471 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
7474 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7475 Define to speculation_safe_value_not_needed.
7477 2018-08-06 Jeff Law <law@redhat.com>
7479 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
7482 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
7485 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
7486 with all enabled __intN types.
7488 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
7490 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7492 * alias.c (record_set): Check for clobber high.
7493 * cfgexpand.c (expand_gimple_stmt): Likewise.
7494 * combine-stack-adj.c (single_set_for_csa): Likewise.
7495 * combine.c (find_single_use_1): Likewise.
7496 (set_nonzero_bits_and_sign_copies): Likewise.
7497 (get_combine_src_dest): Likewise.
7498 (is_parallel_of_n_reg_sets): Likewise.
7499 (try_combine): Likewise.
7500 (record_dead_and_set_regs_1): Likewise.
7501 (reg_dead_at_p_1): Likewise.
7502 (reg_dead_at_p): Likewise.
7503 * dce.c (deletable_insn_p): Likewise.
7504 (mark_nonreg_stores_1): Likewise.
7505 (mark_nonreg_stores_2): Likewise.
7506 * df-scan.c (df_find_hard_reg_defs): Likewise.
7507 (df_uses_record): Likewise.
7508 (df_get_call_refs): Likewise.
7509 * dwarf2out.c (mem_loc_descriptor): Likewise.
7510 * haifa-sched.c (haifa_classify_rtx): Likewise.
7511 * ira-build.c (create_insn_allocnos): Likewise.
7512 * ira-costs.c (scan_one_insn): Likewise.
7513 * ira.c (equiv_init_movable_p): Likewise.
7514 (rtx_moveable_p): Likewise.
7515 (interesting_dest_for_shprep): Likewise.
7516 * jump.c (mark_jump_label_1): Likewise.
7517 * postreload-gcse.c (record_opr_changes): Likewise.
7518 * postreload.c (reload_cse_simplify): Likewise.
7519 (struct reg_use): Add source expr.
7520 (reload_combine): Check for clobber high.
7521 (reload_combine_note_use): Likewise.
7522 (reload_cse_move2add): Likewise.
7523 (move2add_note_store): Likewise.
7524 * print-rtl.c (print_pattern): Likewise.
7525 * recog.c (decode_asm_operands): Likewise.
7526 (store_data_bypass_p): Likewise.
7527 (if_test_bypass_p): Likewise.
7528 * regcprop.c (kill_clobbered_value): Likewise.
7529 (kill_set_value): Likewise.
7530 * reginfo.c (reg_scan_mark_refs): Likewise.
7531 * reload1.c (maybe_fix_stack_asms): Likewise.
7532 (eliminate_regs_1): Likewise.
7533 (elimination_effects): Likewise.
7534 (mark_not_eliminable): Likewise.
7535 (scan_paradoxical_subregs): Likewise.
7536 (forget_old_reloads_1): Likewise.
7537 * reorg.c (find_end_label): Likewise.
7538 (try_merge_delay_insns): Likewise.
7539 (redundant_insn): Likewise.
7540 (own_thread_p): Likewise.
7541 (fill_simple_delay_slots): Likewise.
7542 (fill_slots_from_thread): Likewise.
7543 (dbr_schedule): Likewise.
7544 * resource.c (update_live_status): Likewise.
7545 (mark_referenced_resources): Likewise.
7546 (mark_set_resources): Likewise.
7547 * rtl.c (copy_rtx): Likewise.
7548 * rtlanal.c (reg_referenced_p): Likewise.
7549 (single_set_2): Likewise.
7550 (noop_move_p): Likewise.
7551 (note_stores): Likewise.
7552 * sched-deps.c (sched_analyze_reg): Likewise.
7553 (sched_analyze_insn): Likewise.
7555 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7557 * cse.c (invalidate_reg): New function extracted from...
7558 (invalidate): ...here.
7559 (canonicalize_insn): Check for clobber high.
7560 (invalidate_from_clobbers): invalidate clobber highs.
7561 (invalidate_from_sets_and_clobbers): Likewise.
7562 (count_reg_usage): Check for clobber high.
7563 (insn_live_p): Likewise.
7564 * cselib.c (cselib_expand_value_rtx_1):Likewise.
7565 (cselib_invalidate_regno): Check for clobber in setter.
7566 (cselib_invalidate_rtx): Pass through setter.
7567 (cselib_invalidate_rtx_note_stores):
7568 (cselib_process_insn): Check for clobber high.
7569 * cselib.h (cselib_invalidate_rtx): Add operand.
7571 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7573 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
7574 (mark_not_eliminable): Likewise.
7575 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
7576 * lra-lives.c (process_bb_lives): Check for clobber high.
7577 * lra.c (new_insn_reg): Remember clobber highs.
7578 (collect_non_operand_hard_regs): Check for clobber high.
7579 (lra_set_insn_recog_data): Likewise.
7580 (add_regs_to_insn_regno_info): Likewise.
7581 (lra_update_insn_regno_info): Likewise.
7583 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7585 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
7586 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
7588 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7590 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
7591 (copy_insn_1): Likewise.
7592 (gen_hard_reg_clobber_high): New gen function.
7593 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
7594 * genemit.c (gen_exp): Likewise.
7595 (gen_emit_seq): Pass through info.
7596 (gen_insn): Check for CLOBBER_HIGH.
7597 (gen_expand): Pass through info.
7598 (gen_split): Likewise.
7599 (output_add_clobbers): Likewise.
7600 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
7601 (remove_clobbers): Likewise.
7602 * rtl.h (gen_hard_reg_clobber_high): New declaration.
7604 2018-08-06 Alan Hayward <alan.hayward@arm.com>
7606 * doc/rtl.texi (clobber_high): Add.
7607 (parallel): Add in clobber high
7608 * rtl.c (rtl_check_failed_code3): Add function.
7609 * rtl.def (CLOBBER_HIGH): Add expression.
7610 * rtl.h (RTL_CHECKC3): Add macro.
7611 (rtl_check_failed_code3): Add declaration.
7612 (XC3EXP): Add macro.
7614 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
7617 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
7618 cfun->machine->max_used_stack_alignment if needed.
7620 2018-08-04 Martin Sebor <msebor@redhat.com>
7622 PR tree-optimization/86571
7623 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
7626 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
7629 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7632 2018-08-03 Jeff Law <law@redhat.com>
7635 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7636 Define to speculation_safe_value_not_needed.
7638 2018-08-03 David Malcolm <dmalcolm@redhat.com>
7640 * doc/gcov.texi (-x): Remove duplicate "to".
7641 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
7642 (-Wif-not-aligned): Remove duplicate "is".
7643 (-flto): Remove duplicate "the".
7644 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
7645 duplicate "v5.00.b".
7646 (MSP430 Options): Remove duplicate "and" from the description
7647 of "-mgprel-sec=regexp".
7648 (x86 Options): Remove duplicate copies of "vmldLog102" and
7649 vmlsLog104 from description of "-mveclibabi=type".
7651 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
7653 * internal-fn.h (first_commutative_argument): Declare.
7654 * internal-fn.c (first_commutative_argument): New function.
7655 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
7656 restrictions for pattern statements. Use first_commutative_argument
7657 to look for commutative operands in calls to internal functions.
7659 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
7661 * Makefile.in (wide-int-range.o): New.
7662 * tree-vrp.c: Move all the wide_int_* functions to...
7663 * wide-int-range.cc: ...here.
7664 * tree-vrp.h: Move all the wide_int_* prototypes to...
7665 * wide-int-range.h: ...here.
7667 2018-08-03 Tom de Vries <tdevries@suse.de>
7669 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
7671 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
7672 * except.c (output_function_exception_table): Do early exit if
7673 targetm_common.except_unwind_info (&global_options) == UI_NONE.
7675 2018-08-03 Martin Liska <mliska@suse.cz>
7677 * predict.c (dump_prediction): Change to 2 digits
7680 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
7682 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
7684 (falkor_am_1_gtov_gtov): ... a new insn reservation.
7686 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
7688 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
7689 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
7690 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
7691 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
7692 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
7694 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7696 * diagnostic-show-locus.c (diagnostic_show_locus): Use
7697 pp_take_prefix when saving the existing prefix.
7698 * diagnostic.c (diagnostic_append_note): Likewise.
7699 * langhooks.c (lhd_print_error_function): Likewise.
7700 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
7701 param's type. Free the existing prefix.
7702 (pp_take_prefix): New function.
7703 (pretty_printer::pretty_printer): Drop the prefix parameter.
7704 Rename the length parameter to match the comment.
7705 (pretty_printer::~pretty_printer): Free the prefix.
7706 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
7708 (struct pretty_printer): Drop the "const" from "prefix" field's
7709 type and clarify memory management.
7710 (pp_set_prefix): Drop the "const" from the 2nd param.
7711 (pp_take_prefix): New decl.
7713 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
7715 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
7716 for word_mode_ok here instead of passing as argument.
7717 (expand_block_compare): Change select_block_compare_mode() call.
7718 (expand_strncmp_gpr_sequence): New function.
7719 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
7721 2018-08-02 Jeff Law <law@redhat.com>
7724 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7725 Define to speculation_safe_value_not_needed.
7728 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7729 Define to speculation_safe_value_not_needed.
7731 2018-08-02 Tom de Vries <tdevries@suse.de>
7734 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
7735 function. Return UI_TARGET unconditionally.
7736 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
7737 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
7739 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7741 * genemit.c (print_overload_test): Fix typo.
7743 2018-08-02 Richard Biener <rguenther@suse.de>
7745 PR tree-optimization/86816
7746 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
7747 which checks for value availability before querying it.
7748 (gvn_uses_equal): Use it.
7749 (same_succ_hash): Likewise.
7750 (gimple_equal_p): Likewise.
7752 2018-08-02 Nick Clifton <nickc@redhat.com>
7755 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7756 Define to speculation_safe_value_not_needed.
7759 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7760 Define to speculation_safe_value_not_needed.
7763 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7764 Define to speculation_safe_value_not_needed.
7767 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7768 Define to speculation_safe_value_not_needed.
7771 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7772 Define to speculation_safe_value_not_needed.
7775 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7776 Define to speculation_safe_value_not_needed.
7779 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7780 Define to speculation_safe_value_not_needed.
7783 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7784 Define to speculation_safe_value_not_needed.
7787 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
7788 speculation_safe_value_not_needed.
7791 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
7792 to speculation_safe_value_not_needed.
7794 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7796 * doc/md.texi: Expand the documentation of instruction names
7797 to mention port-local uses. Document '@' in pattern names.
7798 * read-md.h (overloaded_instance, overloaded_name): New structs.
7800 (md_reader::handle_overloaded_name): New member function.
7801 (md_reader::get_overloads): Likewise.
7802 (md_reader::m_first_overload): New member variable.
7803 (md_reader::m_next_overload_ptr): Likewise.
7804 (md_reader::m_overloads_htab): Likewise.
7805 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
7806 m_next_overload_ptr and m_overloads_htab.
7807 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
7808 (get_mode_token, get_code_token, get_int_token): New functions.
7809 (map_attr_string): Add an optional argument that passes back
7810 the associated iterator.
7811 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
7812 (md_reader::handle_overloaded_name, add_overload_instance): New
7814 (apply_iterators): Handle '@' names. Report an error if '@'
7815 is used without iterators.
7816 (initialize_iterators): Initialize the new iterator_group fields.
7817 * genopinit.c (handle_overloaded_code_for)
7818 (handle_overloaded_gen): New functions.
7819 (main): Use them to print declarations of maybe_code_for_* and
7820 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
7821 * genemit.c (print_overload_arguments, print_overload_test)
7822 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
7823 (main): Use it to print definitions of maybe_code_for_* and
7824 maybe_gen_* functions.
7825 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
7826 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
7827 instead of explicit mode checks.
7828 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
7829 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
7830 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
7831 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
7832 (aarch64_expand_compare_and_swap): Likewise
7833 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
7834 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
7835 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
7836 (aarch64_constant_pool_reload_icode): Delete.
7837 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
7838 instead of aarch64_constant_pool_reload_icode. Use
7839 code_for_aarch64_reload_mov instead of explicit mode checks.
7840 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
7841 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
7842 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
7843 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
7844 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
7845 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
7846 (aarch64_atomic_load_op_code): Delete.
7847 (aarch64_emit_atomic_load_op): Likewise.
7848 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
7849 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
7850 instead of aarch64_emit_atomic_load_op.
7851 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
7852 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
7853 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
7854 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
7855 character before the pattern name.
7856 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
7857 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
7858 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
7859 (aarch64_frecps<mode>): Likewise.
7860 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
7861 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
7862 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
7863 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
7864 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
7866 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
7868 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
7869 Allow HFmode constants if TARGET_FP_F16INST.
7871 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
7874 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
7875 No longer check last store for clobber of address register.
7877 2018-08-02 Martin Liska <mliska@suse.cz>
7879 PR gcov-profile/86817
7880 * gcov.c (process_all_functions): New function.
7882 (process_file): Move functions processing to
7883 process_all_functions.
7885 2018-08-02 David Malcolm <dmalcolm@redhat.com>
7887 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
7888 "const" to the "gimple *" and "rtx_insn *" parameters.
7889 * dumpfile.h (dump_user_location_t::dump_user_location_t):
7891 (dump_location_t::dump_location_t): Likewise.
7893 2018-08-01 Martin Sebor <msebor@redhat.com>
7895 PR tree-optimization/86650
7896 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
7897 rather than a "gcall *". Directly pass the data of interest
7898 to percent_K_format, rather than building a temporary CALL_EXPR
7900 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
7901 (gimple_fold_builtin_strncat): Adjust.
7902 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
7903 gcall* argument with gimple*.
7904 * gimple-ssa-warn-restrict.c (check_call): Same.
7905 (wrestrict_dom_walker::before_dom_children): Same.
7906 (builtin_access::builtin_access): Same.
7907 (check_bounds_or_overlap): Same
7908 (maybe_diag_overlap): Same.
7909 (maybe_diag_offset_bounds): Same.
7910 * tree-diagnostic.c (default_tree_printer): Move usage of
7911 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
7913 * tree-pretty-print.c (percent_K_format): Add argument.
7914 * tree-pretty-print.h: Add argument.
7915 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
7916 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
7917 (maybe_diag_stxncpy_trunc): Same.
7918 (handle_builtin_stxncpy): Same.
7919 (handle_builtin_strcat): Same.
7921 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7923 * match.pd: Optimise pointer range checks.
7925 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7927 PR tree-optimization/86758
7928 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
7929 to remove pattern statements.
7931 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7933 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
7934 result of dfs_enumerate_from when constructing stmt_vec_infos,
7935 instead of additionally calling get_loop_body.
7937 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7939 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
7941 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
7942 When creating an iv, assert that the step is not known to be zero.
7943 (vect_setup_realignment): Update call accordingly.
7944 * tree-vect-stmts.c (vectorizable_store): Likewise.
7945 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
7947 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7949 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
7950 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
7951 (vectorizable_reduction): Likewise.
7952 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
7953 (vect_detect_hybrid_slp_stmts): Likewise.
7954 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
7956 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
7958 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
7959 (wide_int_set_zero_nonzero_bits): ...this.
7960 (zero_nonzero_bits_from_vr): Rename to...
7961 (vrp_set_zero_nonzero_bits): ...this.
7962 (extract_range_from_multiplicative_op_1): Abstract wide int
7964 (wide_int_range_multiplicative_op): ...here.
7965 (extract_range_from_binary_expr_1): Extract wide int binary
7966 operations into their own functions.
7967 (wide_int_range_lshift): New.
7968 (wide_int_range_can_optimize_bit_op): New.
7969 (wide_int_range_shift_undefined_p): New.
7970 (wide_int_range_bit_xor): New.
7971 (wide_int_range_bit_ior): New.
7972 (wide_int_range_bit_and): New.
7973 (wide_int_range_trunc_mod): New.
7974 (extract_range_into_wide_ints): New.
7975 (vrp_shift_undefined_p): New.
7976 (extract_range_from_multiplicative_op): New.
7977 (vrp_can_optimize_bit_op): New.
7978 * tree-vrp.h (value_range::dump): New.
7979 (wide_int_range_multiplicative_op): New.
7980 (wide_int_range_lshift):New.
7981 (wide_int_range_shift_undefined_p): New.
7982 (wide_int_range_bit_xor): New.
7983 (wide_int_range_bit_ior): New.
7984 (wide_int_range_bit_and): New.
7985 (wide_int_range_trunc_mod): New.
7986 (zero_nonzero_bits_from_bounds): Rename to...
7987 (wide_int_set_zero_nonzero_bits): ...this.
7988 (zero_nonzero_bits_from_vr): Rename to...
7989 (vrp_set_zero_nonzero_bits): ...this.
7990 (range_easy_mask_min_max): Rename to...
7991 (wide_int_range_can_optimize_bit_op): this.
7992 * vr-values.c (simplify_bit_ops_using_ranges): Rename
7993 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
7995 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
7997 * tree-vectorizer.h (vect_orig_stmt): New function.
7998 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
7999 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
8000 (vect_create_epilog_for_reduction): Likewise.
8001 (vectorizable_live_operation): Likewise.
8002 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
8003 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
8004 * tree-vect-stmts.c (vectorizable_call): Likewise.
8005 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
8007 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8009 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
8011 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
8012 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
8013 (vect_transform_loop): Likewise.
8014 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8016 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8018 * tree-vectorizer.h (vect_schedule_slp): Return void.
8019 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
8020 (vect_schedule_slp): Likewise.
8022 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8024 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
8026 (vect_transform_loop): Update calls accordingly. Schedule SLP
8027 instances before the main loop, if any exist.
8029 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
8031 PR tree-optimization/86749
8032 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
8033 If the lhs is used in a COND_EXPR, check that it is being used
8034 as the "then" or "else" value.
8036 2018-08-01 Tom de Vries <tdevries@suse.de>
8039 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
8040 speculation_safe_value_not_needed.
8042 2018-08-01 Richard Biener <rguenther@suse.de>
8044 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
8047 2018-08-01 Martin Liska <mliska@suse.cz>
8049 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
8050 format how successful transformation is dumped.
8051 (gimple_mod_pow2_value_transform): Likewise.
8052 (gimple_mod_subtract_transform): Likewise.
8053 (gimple_stringops_transform): Likewise.
8055 2018-08-01 Martin Liska <mliska@suse.cz>
8058 * value-prof.c (interesting_stringop_to_profile_p):
8059 Simplify the code and add BUILT_IN_MEMMOVE.
8060 (gimple_stringops_transform): Likewise.
8062 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
8064 * config/aarch64/aarch64-simd.md
8065 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
8066 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
8067 use GPI iterator instead of SI mode.
8069 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
8071 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
8072 rs6000_speculation_barrier.
8073 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
8074 new barrier pattern name.
8076 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
8078 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
8079 (speculation_barrier): New insn.
8081 2018-08-01 Richard Biener <rguenther@suse.de>
8084 * graphite.h: Include isl/id.h and isl/space.h to allow build
8087 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
8090 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
8091 mode for COFF targets.
8092 * defaults.h (TARGET_COFF): Define.
8093 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
8094 TARGET_COFF): Define.
8095 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
8096 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
8097 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
8099 2018-07-31 Alexandre Oliva <oliva@adacore.com>
8101 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
8102 * gimple-streamer-out.c (output_bb): Save it.
8103 * lto-streamer-in.c (input_struct_function_base): Restore
8104 instance discriminator if available. Create map on demand.
8105 * lto-streamer-out.c (output_struct_function_base): Save it if
8107 * final.c (decl_to_instance_map): Document LTO strategy.
8109 2018-07-31 Alexandre Oliva <oliva@adacore.com>
8110 Olivier Hainque <hainque@adacore.com>
8112 * debug.h (decl_to_instance_map_t): New type.
8113 (decl_to_instance_map): Declare.
8114 (maybe_create_decl_to_instance_map): New inline function.
8115 * final.c (bb_discriminator, last_bb_discriminator): New statics,
8116 to track basic block discriminators.
8117 (final_start_function_1): Initialize them.
8118 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
8120 (decl_to_instance_map): New variable.
8121 (map_decl_to_instance, maybe_set_discriminator): New functions.
8122 (notice_source_line): Set discriminator.
8124 2018-07-31 Ian Lance Taylor <iant@golang.org>
8126 * targhooks.c (default_have_speculation_safe_value): Add
8129 2018-07-31 David Malcolm <dmalcolm@redhat.com>
8131 * dump-context.h: Include "pretty-print.h".
8132 (dump_context::refresh_dumps_are_enabled): New decl.
8133 (dump_context::emit_item): New decl.
8134 (class dump_context): Add fields "m_test_pp" and
8136 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
8137 (temp_dump_context::get_dumped_text): New decl.
8138 (class temp_dump_context): Add field "m_pp".
8139 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
8140 (dump_context::refresh_dumps_are_enabled): ...and add a test for
8142 (set_dump_file): Update for above change.
8143 (set_alt_dump_file): Likewise.
8144 (dump_loc): New overload, taking a pretty_printer *.
8145 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
8146 to any test pretty-printer.
8147 (make_item_for_dump_gimple_stmt): New function, adapted from
8148 optinfo::add_gimple_stmt.
8149 (dump_context::dump_gimple_stmt): Call it, and use the result,
8150 eliminating the direct usage of dump_file and alt_dump_file in
8151 favor of indirectly using them via emit_item.
8152 (make_item_for_dump_gimple_expr): New function, adapted from
8153 optinfo::add_gimple_expr.
8154 (dump_context::dump_gimple_expr): Call it, and use the result,
8155 eliminating the direct usage of dump_file and alt_dump_file in
8156 favor of indirectly using them via emit_item.
8157 (make_item_for_dump_generic_expr): New function, adapted from
8159 (dump_context::dump_generic_expr): Call it, and use the result,
8160 eliminating the direct usage of dump_file and alt_dump_file in
8161 favor of indirectly using them via emit_item.
8162 (make_item_for_dump_printf_va): New function, adapted from
8163 optinfo::add_printf_va.
8164 (make_item_for_dump_printf): New function.
8165 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
8166 and use the result, eliminating the direct usage of dump_file and
8167 alt_dump_file in favor of indirectly using them via emit_item.
8168 (make_item_for_dump_dec): New function.
8169 (dump_context::dump_dec): Call it, and use the result,
8170 eliminating the direct usage of dump_file and alt_dump_file in
8171 favor of indirectly using them via emit_item.
8172 (make_item_for_dump_symtab_node): New function, adapted from
8173 optinfo::add_symtab_node.
8174 (dump_context::dump_symtab_node): Call it, and use the result,
8175 eliminating the direct usage of dump_file and alt_dump_file in
8176 favor of indirectly using them via emit_item.
8177 (dump_context::begin_scope): Reimplement, avoiding direct usage
8178 of dump_file and alt_dump_file in favor of indirectly using them
8180 (dump_context::emit_item): New member function.
8181 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
8182 Set up test pretty-printer on the underlying context. Call
8183 refresh_dumps_are_enabled.
8184 (temp_dump_context::~temp_dump_context): Call
8185 refresh_dumps_are_enabled.
8186 (temp_dump_context::get_dumped_text): New member function.
8187 (selftest::verify_dumped_text): New function.
8188 (ASSERT_DUMPED_TEXT_EQ): New macro.
8189 (selftest::test_capture_of_dump_calls): Run all tests twice, with
8190 and then without optinfo enabled. Add uses of
8191 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
8192 * dumpfile.h: Update comment for the dump_* API.
8193 * optinfo-emit-json.cc
8194 (selftest::test_building_json_from_dump_calls): Update for new
8195 param for temp_dump_context ctor.
8196 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
8197 and "m_owned" field.
8198 (optinfo_item::~optinfo_item): Likewise.
8199 (optinfo::add_item): New member function.
8200 (optinfo::emit): Update comment.
8201 (optinfo::add_string): Delete.
8202 (optinfo::add_printf): Delete.
8203 (optinfo::add_printf_va): Delete.
8204 (optinfo::add_gimple_stmt): Delete.
8205 (optinfo::add_gimple_expr): Delete.
8206 (optinfo::add_tree): Delete.
8207 (optinfo::add_symtab_node): Delete.
8208 (optinfo::add_dec): Delete.
8209 * optinfo.h (class dump_context): New forward decl.
8210 (optinfo::add_item): New decl.
8211 (optinfo::add_string): Delete.
8212 (optinfo::add_printf): Delete.
8213 (optinfo::add_printf_va): Delete.
8214 (optinfo::add_gimple_stmt): Delete.
8215 (optinfo::add_gimple_expr): Delete.
8216 (optinfo::add_tree): Delete.
8217 (optinfo::add_symtab_node): Delete.
8218 (optinfo::add_dec): Delete.
8219 (optinfo::add_poly_int): Delete.
8220 (optinfo_item::optinfo_item): Remove "owned" param.
8221 (class optinfo_item): Remove field "m_owned".
8223 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
8226 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
8227 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
8228 requested variable alignment.
8229 (expand_one_ssa_partition): Likewise.
8230 (expand_one_var): Likewise.
8232 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8234 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
8235 to speculation_safe_value_not_needed.
8237 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8239 * targhooks.h (speculation_safe_value_not_needed): New prototype.
8240 * targhooks.c (speculation_safe_value_not_needed): New function.
8241 * target.def (have_speculation_safe_value): Update documentation.
8242 * doc/tm.texi: Regenerated.
8244 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8246 * config/aarch64/iterators.md (ALLI_TI): New iterator.
8247 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
8249 (despeculate_copy<ALLI:mode>_insn): New insn.
8250 (despeculate_copyti_insn): New insn.
8251 (despeculate_simple<ALLI:mode>): New insn
8252 (despeculate_simpleti): New insn.
8253 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
8255 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
8256 aarch64_speculation_safe_value.
8257 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
8259 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8261 * config/aarch64/aarch64-speculation.cc: New file.
8262 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
8263 before pass_reorder_blocks.
8264 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
8266 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
8267 X14 and X15 when tracking speculation.
8268 * config/aarch64/aarch64.md (register name constants): Add
8269 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
8270 (unspec): Add UNSPEC_SPECULATION_TRACKER.
8271 (speculation_barrier): New insn attribute.
8272 (cmp<mode>): Allow SP in comparisons.
8273 (speculation_tracker): New insn.
8274 (speculation_barrier): Add speculation_barrier attribute.
8275 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
8276 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
8277 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
8279 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8281 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
8282 aarch64_track_speculation is true.
8283 (tb<optab><mode>1): Likewise.
8284 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
8285 generate CB[N]Z when tracking speculation.
8286 (aarch64_split_compare_and_swap): Likewise.
8287 (aarch64_split_atomic_op): Likewise.
8289 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8291 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
8293 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8295 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
8296 (speculation_barrier): New insn.
8298 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8300 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
8301 * config/arm/arm.md (speculation_barrier): New expand.
8302 (speculation_barrier_insn): New pattern.
8304 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
8306 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
8307 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
8308 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
8309 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
8311 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
8312 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
8313 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
8314 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
8315 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
8316 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
8317 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
8318 * builtins.c (expand_speculation_safe_value): New function.
8319 (expand_builtin): Call it.
8320 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
8321 * doc/extend.texi: Document __builtin_speculation_safe_value.
8322 * doc/md.texi: Document "speculation_barrier" pattern.
8323 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
8324 TARGET_HAVE_SPECULATION_SAFE_VALUE.
8325 * doc/tm.texi: Regenerated.
8326 * target.def (have_speculation_safe_value, speculation_safe_value): New
8328 * targhooks.c (default_have_speculation_safe_value): New function.
8329 (default_speculation_safe_value): New function.
8330 * targhooks.h (default_have_speculation_safe_value): Add prototype.
8331 (default_speculation_safe_value): Add prototype.
8333 2018-07-31 David Malcolm <dmalcolm@redhat.com>
8335 * dump-context.h (dump_context::dump_loc): New decl.
8336 * dumpfile.c (dump_context::dump_loc): New member function.
8337 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
8338 and dump_gimple_stmt.
8339 (dump_context::dump_gimple_expr_loc): Likewise, using
8341 (dump_context::dump_generic_expr_loc): Likewise, using
8343 (dump_context::dump_printf_loc_va): Likewise, using
8345 (dump_context::begin_scope): Explicitly using the global function
8346 "dump_loc", rather than the member function.
8348 2018-07-31 Martin Sebor <msebor@redhat.com>
8350 PR tree-optimization/86741
8351 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
8353 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
8355 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
8356 depend on whether prefetch instructions will be emitted or not.
8357 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
8358 will be emitted or not.
8359 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
8360 (TARGET_SETMEM_PFD): New macros.
8362 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8364 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
8365 (NULL_STMT_VEC_INFO): Delete.
8366 (stmt_vec_info::operator*): Likewise.
8367 (stmt_vec_info::operator gimple *): Likewise.
8368 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
8369 of NULL_STMT_VEC_INFO.
8370 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
8371 (vect_reassociating_reduction_p): Likewise.
8372 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
8373 (vectorizable_store): Likewise.
8374 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
8375 (vec_info::free_stmt_vec_infos): Likewise.
8377 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8379 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
8380 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
8382 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8384 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
8385 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
8386 (vec_info::free_stmt_vec_info): New private member functions.
8387 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
8388 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
8389 * tree-parloops.c (gather_scalar_reductions): Remove calls to
8390 set_stmt_vec_info_vec and free_stmt_vec_infos.
8391 * tree-vect-loop.c (_loop_vec_info): Remove call to
8392 set_stmt_vec_info_vec.
8393 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
8394 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
8395 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
8396 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
8397 (vec_info::free_stmt_vec_info): ...these new functions. Remove
8398 assignments in {vec_info::,}new_stmt_vec_info that are redundant
8399 with the clearing in the xcalloc.
8400 (stmt_vec_info_vec): Delete.
8401 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
8402 (vectorize_loops): Likewise.
8403 (vec_info::~vec_info): Remove argument from call to
8404 free_stmt_vec_infos.
8405 (vec_info::add_stmt): Remove vinfo argument from call to
8408 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8410 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
8411 rather than a gimple stmt.
8412 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
8413 information for pattern statements when passed the original
8414 statement; instead wait to be passed the pattern statement itself.
8415 Don't call set_vinfo_for_stmt here.
8416 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
8417 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
8418 stmt_vec_infos here.
8419 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
8420 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
8421 stmt_vec_infos entry.
8423 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8425 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
8426 * tree-vectorizer.c (vec_info::replace_stmt): New function.
8427 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
8428 * tree-vect-stmts.c (vectorizable_call): Likewise.
8429 (vectorizable_simd_clone_call): Likewise.
8431 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8433 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
8434 * tree-vectorizer.c (vec_info::remove_stmt): New function.
8435 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
8436 * tree-vect-loop.c (vect_transform_loop): Likewise.
8437 * tree-vect-slp.c (vect_schedule_slp): Likewise.
8438 * tree-vect-stmts.c (vect_remove_stores): Likewise.
8440 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8442 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
8443 (vect_dr_stmt): Delete.
8444 * tree-vectorizer.c (vec_info::lookup_dr): New function.
8445 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
8447 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
8448 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
8449 (vect_verify_datarefs_alignment, vect_peeling_supportable)
8450 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8451 (vect_analyze_data_refs): Likewise.
8452 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
8454 (vect_find_same_alignment_drs): Likewise.
8455 (vect_slp_analyze_node_dependences): Update calls accordingly.
8456 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8457 instead of DR_VECT_AUX.
8458 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
8459 of a vector data references. Use vec_info::lookup_dr instead of
8461 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
8462 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
8463 instead of DR_VECT_AUX.
8465 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8467 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
8469 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8471 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8472 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8473 (vect_gen_prolog_loop_niters): Likewise.
8475 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8477 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
8478 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
8479 (vect_known_alignment_in_bytes, vect_dr_behavior)
8480 (vect_get_scalar_dr_size): Take references as dr_vec_infos
8481 instead of data_references. Update calls to other routines for
8482 which the same change has been made.
8483 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
8484 dr_vec_infos instead of stmt_vec_infos.
8485 (vect_analyze_data_ref_dependence): Update call accordingly.
8486 (vect_slp_analyze_data_ref_dependence)
8487 (vect_record_base_alignments): Use DR_VECT_AUX.
8488 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
8489 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
8490 (vector_alignment_reachable_p, vect_get_data_access_cost)
8491 (vect_peeling_supportable, vect_analyze_group_access_1)
8492 (vect_analyze_group_access, vect_analyze_data_ref_access)
8493 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
8494 (vect_compile_time_alias, vect_small_gap_p)
8495 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
8496 (vect_supportable_dr_alignment): Take references as dr_vec_infos
8497 instead of data_references. Update calls to other routines for
8498 which the same change has been made.
8499 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
8500 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8501 (vect_slp_analyze_and_verify_node_alignment)
8502 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
8503 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8504 (vect_setup_realignment): Use dr_vec_infos. Update calls after
8506 (_vect_peel_info::dr): Replace with...
8507 (_vect_peel_info::dr_info): ...this new field.
8508 (vect_peeling_hash_get_most_frequent)
8509 (vect_peeling_hash_choose_best_peeling): Update accordingly.
8510 (vect_peeling_hash_get_lowest_cost):
8511 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
8512 routines for which the same change has been made.
8513 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
8515 * tree-vect-loop-manip.c (get_misalign_in_elems)
8516 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
8518 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8519 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
8520 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
8521 (get_group_load_store_type, get_negative_load_store_type)
8522 (vect_get_data_ptr_increment, vectorizable_store)
8523 (vectorizable_load): Likewise.
8524 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
8525 Update calls to other routines for which the same change has been made.
8527 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8529 * tree-vectorizer.h (vec_info::move_dr): New member function.
8530 (dataref_aux): Rename to...
8531 (dr_vec_info): ...this and add "dr" and "stmt" fields.
8532 (_stmt_vec_info::dr_aux): Update accordingly.
8533 (_stmt_vec_info::data_ref_info): Delete.
8534 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
8535 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
8536 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
8537 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
8539 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
8540 (STMT_VINFO_DR_INFO): New macro.
8541 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
8542 (set_dr_misalignment): Update after rename of dataref_aux.
8543 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
8544 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
8545 initialization of STMT_VINFO_DATA_REF.
8546 * tree-vectorizer.c (vec_info::move_dr): New function.
8547 * tree-vect-patterns.c (vect_recog_bool_pattern)
8548 (vect_recog_mask_conversion_pattern)
8549 (vect_recog_gather_scatter_pattern): Use it.
8550 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
8551 the "dr" and "stmt" fields of dr_vec_info instead of
8552 STMT_VINFO_DATA_REF.
8554 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8556 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
8557 (is_pattern_stmt_p): Use it.
8558 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
8559 on pattern statements.
8561 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8563 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
8564 original stmt as a stmt_vec_info rather than a gimple stmt.
8565 (vect_pattern_recog_1): Take the statement directly as a
8566 stmt_vec_info, rather than via a gimple_stmt_iterator.
8567 Update call to vect_mark_pattern_stmts.
8568 (vect_pattern_recog): Update calls accordingly.
8570 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8572 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
8573 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
8574 a vect_def_type for the first argument.
8575 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
8576 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
8577 operand if it isn't defined by a vectorized statement.
8578 (vect_build_gather_load_calls): Remove the mask_dt argument and
8579 update calls to vect_get_vec_def_for_stmt_copy.
8580 (vectorizable_bswap): Likewise the dt argument.
8581 (vectorizable_call): Update calls to vectorizable_bswap and
8582 vect_get_vec_def_for_stmt_copy.
8583 (vectorizable_simd_clone_call, vectorizable_assignment)
8584 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
8585 (vectorizable_comparison): Update calls to
8586 vect_get_vec_def_for_stmt_copy.
8587 (vectorizable_store): Likewise. Remove now-unnecessary calls to
8589 (vect_get_loop_based_defs): Remove dt argument and update call
8590 to vect_get_vec_def_for_stmt_copy.
8591 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
8592 and vect_get_vec_def_for_stmt_copy.
8593 (vectorizable_load): Update calls to vect_build_gather_load_calls
8594 and vect_get_vec_def_for_stmt_copy.
8595 * tree-vect-loop.c (vect_create_epilog_for_reduction)
8596 (vectorizable_reduction, vectorizable_live_operation): Update calls
8597 to vect_get_vec_def_for_stmt_copy.
8599 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8601 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
8602 and gimple stmt arguments with a stmt_vec_info.
8603 (vect_record_base_alignments): Update calls accordingly.
8604 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
8605 and gimple stmt arguments with a stmt_vec_info.
8606 (vect_build_slp_tree_1): Remove vinfo argument and update call
8607 to vect_record_max_nunits.
8608 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
8609 and vect_record_max_nunits.
8611 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8613 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
8614 file and take a stmt_vec_info instead of a gimple stmt.
8615 (supportable_widening_operation, vect_finish_replace_stmt)
8616 (vect_finish_stmt_generation, vect_get_store_rhs)
8617 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
8618 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
8619 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
8620 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
8621 (vect_create_data_ref_ptr, bump_vector_ptr)
8622 (vect_permute_store_chain, vect_setup_realignment)
8623 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
8624 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
8625 (vectorizable_reduction, vectorizable_induction)
8626 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
8627 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
8628 than gimple stmts as arguments.
8629 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
8630 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
8631 (can_group_stmts_p, vect_check_gather_scatter)
8632 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
8633 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
8634 (vect_permute_load_chain, vect_shift_permute_load_chain)
8635 (vect_transform_grouped_load)
8636 (vect_record_grouped_load_vectors): Likewise.
8637 * tree-vect-loop.c (vect_fixup_reduc_chain)
8638 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
8639 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
8640 (vectorizable_reduction, vectorizable_induction)
8641 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
8642 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
8643 (vect_get_load_store_mask): Likewise.
8644 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
8645 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
8646 * tree-vect-stmts.c (vect_mark_relevant)
8647 (is_simple_and_all_uses_invariant)
8648 (exist_non_indexing_operands_for_use_p, process_use)
8649 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
8650 (vect_get_vec_def_for_operand, vect_get_vec_defs)
8651 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
8652 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
8653 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
8654 (get_negative_load_store_type, get_load_store_type)
8655 (vect_check_load_store_mask, vect_check_store_rhs)
8656 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
8657 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8658 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8659 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8660 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
8661 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
8662 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
8663 (supportable_widening_operation): Likewise.
8665 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8667 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
8668 a stmt_vec_info instead of a gcall.
8669 (vect_check_gather_scatter): Update call accordingly.
8670 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
8672 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
8673 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
8674 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
8675 instead of a gimple stmt.
8676 (vect_transform_loop): Update calls accordingly.
8677 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
8678 stmt_vec_infos instead of gimple stmts.
8679 (vect_analyze_slp_instance): Update use accordingly.
8680 * tree-vect-stmts.c (read_vector_array, write_vector_array)
8681 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
8682 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
8683 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
8684 (vect_gen_widened_results_half, vect_get_loop_based_defs)
8685 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
8686 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
8687 down to subroutines.
8689 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8691 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
8692 of the worklist from a vector of gimple stmts to a vector of
8694 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8695 (vect_mark_stmts_to_be_vectorized): Likewise
8697 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8699 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
8700 statement before passing it to vect_analyze_stmt.
8701 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
8702 the chain of phi vector definitions. Track the exit phi via its
8704 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
8705 STMT_VINFO_REDUC_DEF.
8706 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
8707 stmt_vec_infos to handle the statement chains.
8708 (vect_get_slp_defs): Record the first statement in the node
8709 using a stmt_vec_info.
8710 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
8711 statements here and pass their stmt_vec_info down to subroutines.
8712 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
8713 down to vect_finish_stmt_generation.
8714 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
8715 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
8716 stmt_vec_infos to subroutines.
8717 (vect_remove_stores): Use stmt_vec_infos to handle the statement
8720 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8722 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
8723 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
8724 (vect_permute_store_chain, vect_setup_realignment)
8725 (vect_permute_load_chain, vect_shift_permute_load_chain)
8726 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
8727 stmts internally, and when passing values to other vectorizer routines.
8728 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8729 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
8730 (vect_analyze_loop_operations, get_initial_def_for_reduction)
8731 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
8732 (vectorizable_reduction, vectorizable_induction)
8733 (vectorizable_live_operation, vect_transform_loop_stmt)
8734 (vect_transform_loop): Likewise.
8735 * tree-vect-patterns.c (vect_reassociating_reduction_p)
8736 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
8737 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
8738 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8739 (vect_slp_analyze_node_operations_1): Likewise.
8740 * tree-vect-stmts.c (vect_mark_relevant, process_use)
8741 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
8742 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
8743 (vect_finish_stmt_generation_1, get_group_load_store_type)
8744 (get_load_store_type, vect_build_gather_load_calls)
8745 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
8746 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
8747 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8748 (vectorizable_store, vectorizable_load, vectorizable_condition)
8749 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
8750 (supportable_widening_operation): Likewise.
8751 (vect_get_vector_types_for_stmt): Likewise.
8752 * tree-vectorizer.h (vect_dr_behavior): Likewise.
8754 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8756 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8757 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
8758 (vect_permute_store_chain, vect_permute_load_chain)
8759 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
8760 repeated stmt_vec_info lookups.
8761 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
8762 (vect_update_ivs_after_vectorizer): Likewise.
8763 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
8764 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
8765 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
8766 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8767 (vect_bb_slp_scalar_cost): Likewise.
8768 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
8770 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8772 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
8773 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
8774 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
8775 to get gassigns and gcalls, rather than operating on generc gimple
8777 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
8778 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
8779 (vectorizable_load, vect_analyze_stmt): Likewise.
8780 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
8782 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8784 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
8785 return stmt_vec_infos rather than gimple stmts. Do not accept
8787 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
8789 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
8790 Update use of get_later_stmt.
8791 (vect_get_constant_vectors): Update call accordingly.
8792 (vect_schedule_slp_instance): Likewise
8793 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
8794 (vect_slp_analyze_instance_dependence): Likewise.
8795 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
8797 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8799 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
8800 (stmt_info_for_cost::stmt_info): ...this new field.
8801 (add_stmt_costs): Update accordingly.
8802 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
8803 (vect_get_known_peeling_cost): Likewise.
8804 (vect_estimate_min_profitable_iters): Likewise.
8805 * tree-vect-stmts.c (record_stmt_cost): Likewise.
8807 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8809 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
8810 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8811 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
8813 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
8815 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8817 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
8818 a gimple stmt to a stmt_vec_info.
8819 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
8821 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8823 * tree-vectorizer.h (vec_info::grouped_stores): Change from
8824 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8825 (_loop_vec_info::reduction_chains): Likewise.
8826 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
8828 * tree-vect-slp.c (vect_analyze_slp): Likewise.
8830 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8832 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
8833 a gimple stmt to a stmt_vec_info.
8834 (_stmt_vec_info::next_element): Likewise.
8835 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
8836 (vect_slp_analyze_and_verify_node_alignment)
8837 (vect_analyze_group_access_1, vect_analyze_group_access)
8838 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
8839 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
8840 (vect_supportable_dr_alignment): Update accordingly.
8841 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
8842 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
8843 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
8844 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
8845 * tree-vect-slp.c (vect_build_slp_tree_1)
8846 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
8847 (vect_split_slp_store_group, vect_analyze_slp_instance)
8848 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
8849 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
8850 (get_group_load_store_type, get_load_store_type)
8851 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
8852 (vect_transform_stmt, vect_remove_stores): Likewise.
8854 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8856 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
8858 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
8859 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
8860 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
8861 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
8862 (vector_alignment_reachable_p, vect_get_data_access_cost)
8863 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
8864 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
8865 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
8866 (vect_analyze_group_access_1, vect_analyze_group_access)
8867 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
8868 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
8869 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
8870 result of vect_dr_stmt and use the stmt_vec_info instead of
8871 the associated gimple stmt.
8872 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
8873 (vect_gen_prolog_loop_niters): Likewise.
8874 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
8876 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8878 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
8879 to a vec<stmt_vec_info>.
8880 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
8881 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
8883 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
8884 to a vec<stmt_vec_info>.
8885 (bst_traits::value_type, bst_traits::value_type): Likewise.
8886 (bst_traits::hash): Update accordingly.
8887 (vect_get_and_check_slp_defs): Change the stmts parameter from
8888 a vec<gimple *> to a vec<stmt_vec_info>.
8889 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
8890 (vect_build_slp_tree): Likewise.
8891 (vect_build_slp_tree_2): Likewise. Update uses of
8892 SLP_TREE_SCALAR_STMTS.
8893 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
8894 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
8895 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
8896 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
8897 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
8898 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
8899 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
8900 (vect_get_constant_vectors, vect_get_slp_defs)
8901 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
8902 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
8903 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
8904 instead of gimple stmts.
8905 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
8906 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
8907 (vect_slp_analyze_instance_dependence): Update uses of
8908 SLP_TREE_SCALAR_STMTS.
8909 (vect_slp_analyze_and_verify_node_alignment): Likewise.
8910 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
8911 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
8912 (get_initial_defs_for_reduction): Likewise.
8913 (vect_create_epilog_for_reduction): Likewise.
8914 (vectorize_fold_left_reduction): Likewise.
8915 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
8916 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
8917 (can_vectorize_live_stmts): Likewise.
8919 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8921 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
8922 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
8923 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
8925 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
8927 (gather_scalar_reductions): Update after above interface changes.
8928 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
8929 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
8931 (vect_force_simple_reduction): Likewise.
8932 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
8933 LOOP_VINFO_REDUCTIONS.
8934 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
8936 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8938 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
8939 a gimple stmt to a stmt_vec_info.
8940 * tree-vect-loop.c (vect_active_double_reduction_p)
8941 (vect_force_simple_reduction, vectorizable_reduction): Update
8944 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8946 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
8947 vec<gimple *> to a vec<stmt_vec_info>.
8948 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
8949 the reduction_phis argument from a vec<gimple *> to a
8951 (vectorizable_reduction): Likewise the phis local variable that
8952 is passed to vect_create_epilog_for_reduction. Update for new type
8953 of SLP_TREE_VEC_STMTS.
8954 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
8955 (vectorizable_live_operation): Likewise.
8956 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
8957 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
8959 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8961 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
8962 a gimple stmt to a stmt_vec_info.
8963 (vectorizable_condition, vectorizable_live_operation)
8964 (vectorizable_reduction, vectorizable_induction): Pass back the
8965 vectorized statement as a stmt_vec_info.
8966 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
8967 use of STMT_VINFO_VEC_STMT.
8968 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
8969 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
8970 as stmt_vec_infos rather than gimple stmts.
8971 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
8973 (vectorizable_live_operation): Likewise.
8974 (vectorizable_reduction, vectorizable_induction): Likewise,
8975 updating use of STMT_VINFO_VEC_STMT.
8976 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
8977 of STMT_VINFO_VEC_STMT.
8978 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
8979 (vectorizable_simd_clone_call, vectorizable_conversion)
8980 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
8981 (vectorizable_store, vectorizable_load, vectorizable_condition)
8982 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
8983 from a gimple stmt to a stmt_vec_info.
8984 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
8985 pointer to a stmt_vec_info to the vectorizable_* routines.
8987 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
8989 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
8990 a gimple stmt to a stmt_vec_info.
8991 (is_pattern_stmt_p): Update accordingly.
8992 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
8993 (vect_record_grouped_load_vectors): Likewise.
8994 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
8995 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
8996 (vect_model_reduction_cost): Likewise.
8997 (vect_create_epilog_for_reduction): Likewise.
8998 (vectorizable_reduction, vectorizable_induction): Likewise.
8999 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9000 Return the stmt_vec_info for the pattern statement.
9001 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
9002 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
9003 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
9004 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
9005 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
9006 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
9007 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
9008 (free_stmt_vec_info, vect_is_simple_use): Likewise.
9010 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9012 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
9013 (vect_finish_stmt_generation): Likewise.
9014 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
9015 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
9016 (vect_build_gather_load_calls): Use the return value of the above
9017 functions instead of a separate call to vinfo_for_stmt. Use narrow
9018 scopes for the input gimple stmt and wider scopes for the associated
9019 stmt_vec_info. Use vec_info::lookup_def when setting these
9020 stmt_vec_infos from an SSA_NAME definition.
9021 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
9022 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
9023 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
9024 (vectorizable_store, vectorizable_load, vectorizable_condition)
9025 (vectorizable_comparison): Likewise.
9026 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
9027 (vectorizable_reduction): Likewise.
9029 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9031 * tree-vectorizer.h (vect_is_simple_use): Add an optional
9032 stmt_vec_info * parameter before the optional gimple **.
9033 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
9034 (process_use, vect_get_vec_def_for_operand_1): Update callers.
9035 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
9036 * tree-vect-loop.c (vectorizable_reduction): Likewise.
9037 (vectorizable_live_operation): Likewise.
9038 * tree-vect-patterns.c (type_conversion_p): Likewise.
9039 (vect_look_through_possible_promotion): Likewise.
9040 (vect_recog_rotate_pattern): Likewise.
9041 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
9043 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9045 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
9046 a typedef to a wrapper class.
9047 (NULL_STMT_VEC_INFO): New macro.
9048 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
9049 (stmt_vec_info::operator*): New function.
9050 (stmt_vec_info::operator gimple *): Likewise.
9051 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
9052 (add_stmt_costs): Likewise.
9053 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
9054 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
9055 (vect_get_known_peeling_cost): Likewise.
9056 (vect_estimate_min_profitable_iters): Likewise.
9057 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9058 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
9059 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
9060 (vectorizable_store, free_stmt_vec_infos): Likewise.
9061 (new_stmt_vec_info): Change return type of xcalloc to
9064 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9066 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
9067 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
9068 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
9069 a single_imm_use-based sequence.
9070 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
9072 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9074 * tree-vectorizer.h (vec_info::lookup_def): Declare.
9075 * tree-vectorizer.c (vec_info::lookup_def): New function.
9076 * tree-vect-patterns.c (vect_get_internal_def): Use it.
9077 (vect_widened_op_tree): Likewise.
9078 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
9079 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
9080 (vectorizable_reduction): Likewise.
9081 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
9083 (vect_is_slp_reduction): Update calls accordingly. Use
9084 vec_info::lookup_def.
9085 (vect_is_simple_reduction): Likewise
9086 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
9088 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9090 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
9091 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
9092 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
9094 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
9095 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
9096 (vect_update_vf_for_slp, vect_analyze_loop_operations)
9097 (vect_is_slp_reduction, vectorizable_induction)
9098 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
9099 * tree-vect-patterns.c (vect_init_pattern_stmt):
9100 (vect_determine_min_output_precision_1, vect_determine_precisions)
9101 (vect_pattern_recog): Likewise.
9102 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
9103 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
9104 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
9105 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
9106 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
9107 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
9108 info field from a loop to a loop_vec_info.
9110 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9112 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
9113 (vec_info::add_stmt): Declare.
9114 * tree-vectorizer.c (vec_info::add_stmt): New function.
9115 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
9116 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
9117 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
9118 (vectorizable_induction): Likewise.
9119 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
9120 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
9121 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
9122 (vectorizable_load): Likewise.
9123 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9124 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
9125 (vect_recog_gather_scatter_pattern): Likewise.
9126 (append_pattern_def_seq): Likewise. Remove a check that is
9127 performed by add_stmt itself.
9129 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9131 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
9132 which make_ssa_name was called with new_stmt before new_stmt
9135 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9137 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
9139 (vect_is_slp_reduction): ...here...
9140 (vect_is_simple_reduction): ...and here. Remove repetition of tests
9141 that are already known to be false.
9143 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9145 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
9146 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
9147 STMT_VINFO_NUM_SLP_USES when it's true.
9148 (vect_free_slp_instance): Add a final_p parameter and pass it to
9150 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
9151 (vect_analyze_slp_instance): Likewise.
9152 (vect_slp_analyze_operations): Likewise.
9153 (vect_slp_analyze_bb_1): Likewise.
9154 * tree-vectorizer.c (vec_info): Likewise.
9155 * tree-vect-loop.c (vect_transform_loop): Likewise.
9157 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9159 * tree-vect-loop.c (vectorizable_reduction): Assert that the
9160 function is not called for second and subsequent members of
9163 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9165 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
9166 cases for nested loops from here to ...
9167 (vect_create_epilog_for_reduction): ...here. Only call
9168 vect_is_simple_use for inner-loop reductions.
9170 2018-07-31 Martin Liska <mliska@suse.cz>
9172 PR gcov-profile/85338
9173 PR gcov-profile/85350
9174 PR gcov-profile/85372
9175 * profile.c (struct location_triplet): New.
9176 (struct location_triplet_hash): Likewise.
9177 (output_location): Do not output a BB that
9178 is already recorded for a line.
9179 (branch_prob): Use streamed_locations.
9181 2018-07-31 Martin Liska <mliska@suse.cz>
9183 PR gcov-profile/85370
9184 * coverage.c (coverage_begin_function): Do not mark target
9185 clones as artificial functions.
9187 2018-07-31 Martin Liska <mliska@suse.cz>
9189 PR gcov-profile/83813
9190 PR gcov-profile/84758
9191 PR gcov-profile/85217
9192 PR gcov-profile/85332
9193 * profile.c (branch_prob): Do not record GOTO expressions
9194 for GIMPLE statements which locations are already streamed.
9196 2018-07-31 Olivier Hainque <hainque@adacore.com>
9198 * gcc.c (handle_spec_function): Accept a soft_matched_part
9199 argument, as do_spec_1. Pass it down to ...
9200 (eval_spec_function): Accept a soft_matched_part argument,
9201 and pass it down to ...
9202 (do_spec_2): Accept a soft_matched_part argument, and pass
9203 it down to do_spec_1.
9204 (do_spec_1): Pass soft_matched_part to handle_spec_function.
9205 (handle_braces): Update call to handle_spec_function.
9206 (driver::set_up_specs): Update calls to do_spec_2.
9207 (compare_debug_dump_opt_spec_function): Likewise.
9208 (compare_debug_self_opt_spec_function): Likewise.
9210 2018-07-31 Olivier Hainque <hainque@adacore.com>
9212 * common.opt (nolibc): New option.
9213 * doc/invoke.texi (Link Options): Document it.
9214 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
9215 * config/alpha/linux.h: Likewise.
9216 * config/arc/elf.h: Likewise.
9217 * config/arm/uclinux-elf.h: Likewise.
9218 * config/arm/unknown-elf.h: Likewise.
9219 * config/avr/avrlibc.h: Likewise.
9220 * config/bfin/bfin.h: Likewise.
9221 * config/bfin/linux.h: Likewise.
9222 * config/bfin/uclinux.h: Likewise.
9223 * config/darwin.h: Likewise.
9224 * config/darwin10.h: Likewise.
9225 * config/darwin12.h: Likewise.
9226 * config/gnu-user.h: Likewise.
9227 * config/lm32/uclinux-elf.h: Likewise.
9228 * config/pa/pa-hpux11.h: Likewise.
9229 * config/pa/pa64-hpux.h: Likewise.
9230 * config/sparc/sparc.h: Likewise.
9232 2018-07-31 Olivier Hainque <hainque@adacore.com>
9234 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
9235 undefined variables.
9237 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
9240 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
9243 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
9245 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
9246 terminated string literal.
9248 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
9250 PR rtl-optimization/85160
9251 * combine.c (is_just_move): New function.
9252 (try_combine): Allow combining two instructions into two if neither of
9253 the original instructions was a move.
9255 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
9258 * doc/extend.texi (Global Register Variables): Discourage use of type
9260 (Local Register Variables): Likewise.
9262 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
9264 PR tree-optimization/86506
9265 * hwint.h (ceil_log2): Resync with hwint.c implementation.
9267 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
9270 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
9271 hard_regno, make sure no insn between `from` and `to` clobbers it.
9273 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
9274 Tom de Vries <tdevries@suse.de>
9276 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
9277 (PTX_DEFAULT_RUNTIME_DIM): ... this.
9278 (nvptx_goacc_validate_dims): Set default worker and gang dims to
9279 PTX_DEFAULT_RUNTIME_DIM.
9280 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
9282 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
9284 * config/pa/pa.c (pa_output_addr_vec): Align address table.
9285 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
9286 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
9288 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
9290 * config/rs6000/constraints.md (wG constraint): Delete, no longer
9292 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
9293 predicate to reflect toc fusion has been deleted.
9294 (toc_fusion_mem_raw): Delete, no longer used.
9295 (toc_fusion_mem_wrapped): Likewise.
9296 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
9298 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
9299 Delete, no longer used.
9300 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
9301 meant to be used for toc fusion.
9302 (rs6000_debug_print_mode): Delete toc fusion debugging.
9303 (rs6000_debug_reg_global): Likewise.
9304 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
9305 fusion and secondary reload support that were never used.
9306 (rs6000_option_override_internal): Delete TOC fusion, that was only
9307 partially defined, and it did not work unless you also used the
9309 (rs6000_legitimate_address_p): Delete TOC fusion support.
9310 (rs6000_opt_masks): Likewise.
9311 (fusion_wrap_memory_address): Delete function, no longer used.
9312 (fusion_split_address); Delete TOC fusion support.
9313 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
9314 longer used with toc fusion being deleted.
9315 (TARGET_TOC_FUSION_FP): Likewise.
9316 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
9318 (toc fusion spliter): Delete TOC fusion support.
9319 (toc_fusionload_<mode>): Likewise.
9320 (toc_fusionload_di): Likewise.
9321 (fusion_gpr_load_<mode>): Delete generator function, this insn no
9322 longer needs to be named. Rename predicate to delete TOC fusion.
9323 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
9324 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
9325 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
9326 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
9327 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
9329 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
9331 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
9332 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
9333 __int128 in built-in function prototypes.
9334 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
9335 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
9337 2018-07-27 Martin Sebor <msebor@redhat.com>
9339 PR tree-optimization/86696
9340 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
9341 types, including enums.
9342 (handle_char_store): Be prepared for the above function to fail.
9344 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
9346 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
9347 when optimization level is lower than 2 or optimize for size.
9349 2018-07-26 Martin Sebor <msebor@redhat.com>
9351 PR tree-optimization/86043
9352 PR tree-optimization/86042
9353 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
9354 (get_string_cst_length): Rename...
9355 (get_min_string_length): ...to this. Add argument.
9356 (handle_char_store): Extend to handle multi-character stores by
9358 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
9359 * tree.h (initializer_zerop): Add argument.
9361 2018-07-26 Jakub Jelinek <jakub@redhat.com>
9364 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
9365 declare target to variables if they have always,{to,from,tofrom} map
9368 2018-07-26 Martin Liska <mliska@suse.cz>
9371 * lto-wrapper.c: Add linker_output as prefix
9372 for ltrans_output_file.
9374 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
9376 PR rtl-optimization/85805
9377 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
9378 value for hard registers if that was written in the same mode.
9380 2018-07-26 Martin Liska <mliska@suse.cz>
9382 PR gcov-profile/86536
9383 * gcov.c (format_gcov): Use printf format %.*f directly
9384 and do not handle special values.
9386 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9388 * common/config/arc/arc-common.c (arc_option_optimization_table):
9389 Update default optimizations for size.
9391 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9393 * config/arc/arc.md (movsf_insn): Add short instruction selection.
9394 * config/arc/constraints.md (CfZ): New constraint.
9395 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
9396 (subsf3_fpu): Likewise.
9397 (cmpsf_fpu): Likewise.
9398 (cmpsf_fpu_uneq): Likewise.
9400 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9402 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
9404 (arc_is_uncached_mem_p): uncached applies to both the variable and
9407 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
9409 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
9412 2018-07-25 David Malcolm <dmalcolm@redhat.com>
9414 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
9415 field "m_scopes" from vec to auto_vec.
9417 2018-07-25 Martin Liska <mliska@suse.cz>
9419 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
9422 2018-07-25 Richard Biener <rguenther@suse.de>
9425 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
9426 special wrt context_die late.
9427 (gen_subprogram_die): Re-use DIEs in local scope.
9429 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
9431 PR tree-optimization/86644
9432 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
9434 2018-07-25 Martin Liska <mliska@suse.cz>
9437 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
9438 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
9440 2018-07-25 Martin Liska <mliska@suse.cz>
9443 * params.def: Explain ASan abbreviation and provide
9444 a documentation link.
9446 2018-07-24 Martin Sebor <msebor@redhat.com>
9448 PR tree-optimization/86622
9449 PR tree-optimization/86532
9450 * builtins.h (string_length): Declare.
9451 * builtins.c (c_strlen): Correct handling of non-constant offsets.
9452 (check_access): Be prepared for non-constant length ranges.
9453 (string_length): Make extern.
9454 * expr.c (string_constant): Only handle the minor non-constant
9455 array index. Use string_constant to compute the length of
9456 a generic string constant.
9458 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
9460 PR tree-optimization/86618
9461 * tree-vect-stmts.c (vectorizable_call): Don't take the address
9462 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
9464 2018-07-24 David Malcolm <dmalcolm@redhat.com>
9466 PR tree-optimization/86636
9467 * json.cc (json::object::set): Fix comment. Add assertions.
9468 (json::array::append): Move here from json.h. Add comment and an
9470 (json::string::string): Likewise.
9471 * json.h (json::array::append): Move to json.cc.
9472 (json::string::string): Likewise.
9473 * optinfo-emit-json.cc
9474 (optrecord_json_writer::impl_location_to_json): Assert that we
9475 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
9476 wrapper around it. Expand the location once, rather than three
9478 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
9479 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
9481 (optrecord_json_writer::optinfo_to_json): Likewise, in four
9482 places. Fix some overlong lines.
9484 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
9486 * config/aarch64/aarch64-simd.md
9487 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
9488 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
9489 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
9490 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
9491 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
9492 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
9493 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
9494 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
9495 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
9497 2018-07-24 Jakub Jelinek <jakub@redhat.com>
9500 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
9501 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
9502 and abs_d == d, do the power of two handling if profitable.
9504 2018-07-24 Richard Biener <rguenther@suse.de>
9506 * match.pd: Add BIT_FIELD_REF canonicalizations.
9508 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9511 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
9513 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9515 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
9516 terminated STRING_CST object.
9518 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
9520 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
9522 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9524 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
9525 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
9526 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
9527 the elements into a register.
9528 (rs6000_split_v4si_init_di_reg): Delete.
9529 (rs6000_split_v4si_init): Delete.
9530 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
9531 (vsx_init_v4si): Rewrite as a define_expand.
9533 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
9535 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
9536 zero_extend argument from memory): New.
9538 2018-07-22 Martin Sebor <msebor@redhat.com>
9541 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
9542 diagnosing calls with unknown arguments unless -Walloca-larger-than
9543 is restricted to less than PTRDIFF_MAX bytes.
9545 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
9547 * doc/gcov.texi (Invoking Gcov): Editorial changes.
9549 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9551 * pretty-print.c (text_info::set_location): Remove redundant
9552 "line_table" parameter from call to rich_location::set_range.
9554 2018-07-20 Martin Sebor <msebor@redhat.com>
9557 * builtins.c (expand_builtin_alloca): Adjust.
9558 * calls.c (alloc_max_size): Simplify.
9559 * cgraphunit.c (cgraph_node::expand): Adjust.
9560 * common.opt (larger_than_size, warn_frame_larger_than): Remove
9562 (frame_larger_than_size): Same.
9563 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
9564 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
9566 * doc/invoke.texi (GCC Command Options): Document option arguments.
9567 Explain byte-size arguments and suffixes.
9568 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
9569 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
9570 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
9571 * doc/options.texi (UInteger): Expand.
9572 (Host_Wide_Int, ByteSize): Document new properties.
9573 * final.c (final_start_function_1): Include sizes in an error message.
9574 * function.c (frame_offset_overflow): Same.
9575 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
9576 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
9577 Diagnose unbounded alloca calls only for limits of less than
9579 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
9580 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
9582 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
9583 only for limits of less than PTRDIFF_MAX.
9584 * langhooks-def.h (lhd_handle_option): Change function argument
9586 * langhooks.c (lhd_handle_option): Same.
9587 * langhooks.h (handle_option): Same.
9588 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
9590 (var_type, var_type_struct): Same.
9591 (var_set): Handle ByteSize flag.
9592 * optc-gen.awk: Add comments to output to ease debugging. Make
9593 use of HOST_WIDE_INT where appropriate.
9594 * opts-gen-save.awk: Use %lx to format unsigned long.
9595 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
9596 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
9597 arguments. Parse bytes-size suffixes.
9598 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
9599 (enum_value_to_arg): Same.
9600 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
9601 (handle_option): Adjust.
9602 (generate_option): Change function argument to HOST_WIDE_INT.
9603 (cmdline_handle_error): Adjust.
9604 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
9605 (set_option): Change function argument to HOST_WIDE_INT.
9606 (option_enabled): Handle cl_host_wide_int.
9607 (get_option_state): Handle CLVC_SIZE.
9608 (control_warning_option): Same.
9609 * opts.c (common_handle_option): Change function argument to
9610 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
9611 OPT_Wvla_larger_than_.
9612 * opts.h (enum cl_var_type): Add an enumerator.
9613 * stor-layout.c (layout_decl): Print a more meaningful warning.
9614 * toplev.c (output_stack_usage): Adjust.
9616 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
9618 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
9619 call to inline_expand_builtin_string_cmp.
9620 (expand_builtin_strcmp): Likewise.
9621 (expand_builtin_strncmp): Likewise.
9622 (inline_string_cmp): Delete the last parameter, change char_type_node
9623 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
9625 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
9626 the inlining expansion on target where the type of the call has same or
9627 narrower precision than unsigned char.
9629 2018-07-20 David Malcolm <dmalcolm@redhat.com>
9631 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
9632 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
9633 * common.opt (fsave-optimization-record): New option.
9634 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
9635 * doc/invoke.texi (-fsave-optimization-record): New option.
9636 * dumpfile.c: Include "optinfo-emit-json.h".
9637 (struct kv_pair): Move to coretypes.h.
9638 (optgroup_options): Make non-static.
9639 (dump_context::end_scope): Call
9640 optimization_records_maybe_pop_dump_scope.
9641 * dumpfile.h (optgroup_options): New decl.
9642 * json.cc: New file.
9644 * optinfo-emit-json.cc: New file.
9645 * optinfo-emit-json.h: New file.
9646 * optinfo.cc: Include "optinfo-emit-json.h".
9647 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
9648 (optinfo_enabled_p): Check optimization_records_enabled_p.
9649 (optinfo_wants_inlining_info_p): Likewise.
9650 * optinfo.h: Update comment.
9651 * profile-count.c (profile_quality_as_string): New function.
9652 * profile-count.h (profile_quality_as_string): New decl.
9653 (profile_count::quality): New accessor.
9654 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
9655 and optinfo_emit_json_cc_tests.
9656 * selftest.h (selftest::json_cc_tests): New decl.
9657 (selftest::optinfo_emit_json_cc_tests): New decl.
9658 * toplev.c: Include "optinfo-emit-json.h".
9659 (compile_file): Call optimization_records_finish.
9660 (do_compile): Call optimization_records_start.
9661 * tree-ssa-live.c: Include optinfo.h.
9662 (remove_unused_scope_block_p): Retain inlining information if
9663 optinfo_wants_inlining_info_p returns true.
9665 2018-07-20 Richard Biener <rguenther@suse.de>
9668 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
9669 to cover -flto-partition=none.
9671 2018-07-20 Martin Liska <mliska@suse.cz>
9673 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
9674 (get_decl_source_range): Remove unused function.
9676 2018-07-20 Richard Biener <rguenther@suse.de>
9678 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
9679 (struct vn_phi_s): Likewise.
9680 (struct vn_reference_s): Likewise.
9681 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
9682 for searching the slot of an entry known to be in the hash itself.
9683 (vn_phi_hasher::equal): Likewise.
9684 (vn_reference_hasher::equal): Likewise.
9685 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
9687 (optimistic_info, current_info): Remove, keeping only valid_info.
9688 (vn_reference_lookup_1): Remove fallback lookup.
9689 (vn_reference_lookup_2): Likewise.
9690 (vn_nary_op_lookup_1): Likewise.
9691 (vn_phi_lookup): Likewise.
9692 (vn_nary_build_or_lookup_1): Make sure to not chain the built
9694 (vn_reference_insert): Adjust, chain the inserted hash element
9695 at last_inserted_ref.
9696 (vn_reference_insert_pieces): Likewise.
9697 (visit_reference_op_call): Likewise.
9698 (vn_nary_op_insert_into): Chain the inserted hash element at
9700 (vn_nary_op_insert_pieces): Adjust.
9701 (vn_nary_op_insert): Likewise.
9702 (vn_nary_op_insert_stmt): Likewise.
9703 (vn_phi_insert): Adjust, chain the inserted hash element at
9705 (process_scc): Remove clearing and copying the optimistic
9706 table. Instead remove elements inserted during an optimistic
9707 iteration from the single table we maintain.
9708 (init_scc_vn): Adjust.
9709 (free_scc_vn): Likewise.
9710 (sccvn_dom_walker::record_cond): Likewise.
9711 (sccvn_dom_walker::after_dom_children): Likewise.
9713 2018-07-19 Martin Sebor <msebor@redhat.com>
9715 PR tree-optimization/84047
9716 PR tree-optimization/83776
9717 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
9718 (check_array_bounds): Call it.
9720 2018-07-19 Martin Sebor <msebor@redhat.com>
9722 * align.h (align_flags): Use member initialization.
9724 2018-07-19 David Malcolm <dmalcolm@redhat.com>
9726 * Makefile.in (OBJS): Add optinfo.o.
9727 * coretypes.h (class symtab_node): New forward decl.
9728 (struct cgraph_node): New forward decl.
9729 (class varpool_node): New forward decl.
9730 * dump-context.h: New file.
9731 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
9733 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
9734 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
9735 (set_alt_dump_file): Likewise.
9736 (dump_context::~dump_context): New dtor.
9737 (dump_gimple_stmt): Move implementation to...
9738 (dump_context::dump_gimple_stmt): ...this new member function.
9739 Add the stmt to any pending optinfo, creating one if need be.
9740 (dump_gimple_stmt_loc): Move implementation to...
9741 (dump_context::dump_gimple_stmt_loc): ...this new member function.
9742 Start a new optinfo and add the stmt to it.
9743 (dump_gimple_expr): Move implementation to...
9744 (dump_context::dump_gimple_expr): ...this new member function.
9745 Add the stmt to any pending optinfo, creating one if need be.
9746 (dump_gimple_expr_loc): Move implementation to...
9747 (dump_context::dump_gimple_expr_loc): ...this new member function.
9748 Start a new optinfo and add the stmt to it.
9749 (dump_generic_expr): Move implementation to...
9750 (dump_context::dump_generic_expr): ...this new member function.
9751 Add the tree to any pending optinfo, creating one if need be.
9752 (dump_generic_expr_loc): Move implementation to...
9753 (dump_context::dump_generic_expr_loc): ...this new member
9754 function. Add the tree to any pending optinfo, creating one if
9756 (dump_printf): Move implementation to...
9757 (dump_context::dump_printf_va): ...this new member function. Add
9758 the text to any pending optinfo, creating one if need be.
9759 (dump_printf_loc): Move implementation to...
9760 (dump_context::dump_printf_loc_va): ...this new member function.
9761 Start a new optinfo and add the stmt to it.
9762 (dump_dec): Move implementation to...
9763 (dump_context::dump_dec): ...this new member function. Add the
9764 value to any pending optinfo, creating one if need be.
9765 (dump_context::dump_symtab_node): New member function.
9766 (dump_context::get_scope_depth): New member function.
9767 (dump_context::begin_scope): New member function.
9768 (dump_context::end_scope): New member function.
9769 (dump_context::ensure_pending_optinfo): New member function.
9770 (dump_context::begin_next_optinfo): New member function.
9771 (dump_context::end_any_optinfo): New member function.
9772 (dump_context::s_current): New global.
9773 (dump_context::s_default): New global.
9774 (dump_scope_depth): Delete global.
9775 (dumpfile_ensure_any_optinfo_are_flushed): New function.
9776 (dump_symtab_node): New function.
9777 (get_dump_scope_depth): Reimplement in terms of dump_context.
9778 (dump_begin_scope): Likewise.
9779 (dump_end_scope): Likewise.
9780 (selftest::temp_dump_context::temp_dump_context): New ctor.
9781 (selftest::temp_dump_context::~temp_dump_context): New dtor.
9782 (selftest::verify_item): New function.
9783 (ASSERT_IS_TEXT): New macro.
9784 (ASSERT_IS_TREE): New macro.
9785 (ASSERT_IS_GIMPLE): New macro.
9786 (selftest::test_capture_of_dump_calls): New test.
9787 (selftest::dumpfile_c_tests): Call it.
9788 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
9789 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
9790 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
9791 descriptive comment.
9792 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
9793 (dump_node, dump_bb): Move these unrelated decls.
9794 (class dump_manager): Add leading comment.
9795 * optinfo.cc: New file.
9796 * optinfo.h: New file.
9798 2018-07-19 Michael Collison <michael.collison@arm.com>
9799 Richard Henderson <rth@redhat.com>
9801 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
9802 (subti): Handle op1 zero.
9803 (subvti4, usub4ti4): New.
9804 (*sub<GPI>3_compare1_imm): New.
9805 (sub<GPI>3_carryinCV): New.
9806 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
9807 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
9809 2018-07-19 Michael Collison <michael.collison@arm.com>
9810 Richard Henderson <rth@redhat.com>
9812 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
9813 (addti3): Create simpler code if low part is already known to be 0.
9814 (addvti4, uaddvti4): New.
9815 (*add<GPI>3_compareC_cconly_imm): New.
9816 (*add<GPI>3_compareC_cconly): New.
9817 (*add<GPI>3_compareC_imm): New.
9818 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
9819 handle constants within this pattern..
9820 (*add<GPI>3_compareV_cconly_imm): New.
9821 (*add<GPI>3_compareV_cconly): New.
9822 (*add<GPI>3_compareV_imm): New.
9823 (add<GPI>3_compareV): New.
9824 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
9825 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
9826 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
9827 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
9828 with 'comparison' operator.
9829 (*add<GPI>3_compareV_cconly_imm): Ditto.
9830 (*add<GPI>3_compareV_cconly): Ditto.
9831 (*add<GPI>3_compareV_imm): Ditto.
9832 (add<GPI>3_compareV): Ditto.
9833 (add<mode>3_carryinC): Ditto.
9834 (*add<mode>3_carryinC_zero): Ditto.
9835 (*add<mode>3_carryinC): Ditto.
9836 (add<mode>3_carryinV): Ditto.
9837 (*add<mode>3_carryinV_zero): Ditto.
9838 (*add<mode>3_carryinV): Ditto.
9840 2018-07-19 Michael Collison <michael.collison@arm.com>
9841 Richard Henderson <rth@redhat.com>
9843 * config/aarch64/aarch64-modes.def (CC_V): New.
9844 * config/aarch64/aarch64-protos.h
9845 (aarch64_addti_scratch_regs): Declare
9846 (aarch64_subvti_scratch_regs): Declare.
9847 (aarch64_expand_subvti): Declare.
9848 (aarch64_gen_unlikely_cbranch): Declare
9849 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
9850 for signed overflow using CC_Vmode.
9851 (aarch64_get_condition_code_1): Handle CC_Vmode.
9852 (aarch64_gen_unlikely_cbranch): New function.
9853 (aarch64_addti_scratch_regs): New function.
9854 (aarch64_subvti_scratch_regs): New function.
9855 (aarch64_expand_subvti): New function.
9857 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9859 * config/aarch64/aarch64-option-extensions.def: New entry for profile
9861 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
9862 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
9865 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9868 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
9869 address check not strict.
9871 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
9873 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
9875 (store_pair_lanes<mode>): Likewise.
9876 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
9877 enum value 'ADDR_QUERY_LDP_STP_N'.
9878 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
9879 (aarch64_print_address_internal): Add declaration.
9880 (aarch64_print_ldpstp_address): Remove.
9881 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
9882 (aarch64_print_operand): Change printing of 'y'.
9883 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
9884 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
9885 'true' rather than '1'.
9886 * config/aarch64/constraints.md (Uml): Likewise.
9887 (Uml): Rename to Umn.
9890 2018-07-19 Richard Biener <rguenther@suse.de>
9892 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
9894 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
9895 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
9896 (vn_reference_hasher): Likewise.
9897 (struct vn_tables_s): Remove obstack and alloc-pool members.
9898 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
9899 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
9900 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
9901 (vn_reference_insert_pieces): Likewise.
9902 (alloc_vn_nary_op_noinit): Adjust.
9903 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
9904 (vn_phi_eq): Adjust.
9905 (shared_lookup_phiargs): Remove.
9906 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
9907 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
9908 (visit_reference_op_call): Likewise.
9909 (copy_nary, copy_phi, copy_reference): Remove.
9910 (process_scc): Rewind the obstack when iterating. Do not
9911 copy the elements to valid_info but just move them from one
9912 hashtable to the other.
9913 (allocate_vn_table): Adjust.
9914 (free_vn_table): Likewise.
9915 (init_scc_vn): Likewise.
9916 (free_scc_vn): Likewise.
9918 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
9921 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
9922 indirect_return as function type attribute.
9923 (ix86_attribute_table): Change indirect_return to function
9925 * doc/extend.texi: Update indirect_return attribute.
9927 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
9929 * wide-int.h (widest2_int): New.
9930 * gimple-fold.c (arith_overflowed_p): Use it.
9931 * tree.h (widest2_int_cst): New.
9932 * tree-vrp.c (wide_int_binop_overflow): Rename from
9933 vrp_int_const_binop.
9934 Rewrite to work on trees.
9935 (extract_range_from_multiplicative_op_1): Abstract code to...
9936 (wide_int_range_min_max): ...here.
9937 (wide_int_range_cross_product): ...and here.
9938 (extract_range_from_binary_expr_1): Abstract overflow code to...
9939 (wide_int_range_mult_wrapping): ...here.
9940 * tree-vrp.h (wide_int_range_cross_product): New.
9941 (wide_int_range_mult_wrapping): New.
9943 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
9944 Julia Koval <julia.koval@intel.com>
9946 * config/i386/x86-tune-costs.h (skylake_memcpy,
9947 skylake_memset): Replace rep_prefix with unrolling for size 512.
9949 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
9952 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
9953 comparision with EQ_EXPR in last stmt.
9955 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
9957 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
9958 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
9959 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
9960 previously known as "PowerPC AltiVec Built-in Functions". Move
9961 some material to new subsubsections "PowerPC AltiVec Built-in
9962 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
9964 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
9965 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
9966 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
9967 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
9969 2018-07-18 Richard Biener <rguenther@suse.de>
9971 PR tree-optimization/86557
9972 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
9975 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
9977 * config/s390/s390.c (s390_function_profiler): Generate CFI.
9979 2018-07-17 Jeff Law <law@redhat.com>
9981 * config/arm/arm.c (get_label_padding): Update for recent
9982 changes to label_to_alignment.
9984 PR tree-optimization/86010
9985 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
9987 * config/mips/mips.c (vr4130_align_insns): Update for recent
9988 changes to label_to_alignment.
9990 * config/frv/frv.c (frv_label_align): Update for recent changes
9991 to label_to_alignment.
9993 * config/nios2/nios2.c (nios2_label_align): Update for recent
9994 changes which dropped ALIGN_LABELS_LOG.
9996 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
9998 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
9999 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
10001 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
10003 * config/arc/arc.c (arc_label_align): Use align_labels instead of
10004 deprecated align_labels_log.
10006 2018-07-17 Richard Biener <rguenther@suse.de>
10009 * dwarf2out.c (init_sections_and_labels): Always generate
10010 a debug_line_str_section for early LTO debug.
10011 (dwarf2out_finish): Reset debug_line_str_hash output early.
10012 Bump counter for extra dwarf5 .debug_loc labels to not conflict
10014 (dwarf2out_early_finish): Output debug_line_str.
10016 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
10018 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
10019 index register on z196 or later.
10021 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
10023 * config/s390/s390.c (s390_default_align): Set default function
10025 (s390_override_options_after_change): Call s390_default align.
10026 (s390_option_override_internal): Call s390_default_align.
10027 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
10029 2018-07-17 Jakub Jelinek <jakub@redhat.com>
10031 PR middle-end/86542
10032 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
10033 to _looptemp_ clauses, other than the first two.
10035 2018-07-17 Martin Liska <mliska@suse.cz>
10037 * opts.c: Do not enable OPT_falign_* for -Os.
10039 2018-07-17 Martin Liska <mliska@suse.cz>
10041 * align.h (MAX_CODE_ALIGN): New.
10042 (MAX_CODE_ALIGN_VALUE): New.
10043 * common/config/i386/i386-common.c (ix86_handle_option):
10044 (MAX_CODE_ALIGN): Moved to align.h.
10045 * final.c (MAX_CODE_ALIGN): Likewise.
10046 * opts.c (parse_and_check_align_values):
10047 (MAX_CODE_ALIGN): Likewise.
10048 (MAX_CODE_ALIGN_VALUE): Likewise.
10050 2018-07-17 Martin Liska <mliska@suse.cz>
10052 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
10053 in order to fulfil coding style.
10054 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
10055 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10056 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
10057 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
10058 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
10059 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
10060 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
10061 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
10062 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
10064 2018-07-17 Martin Liska <mliska@suse.cz>
10066 * align.h: New file.
10067 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
10069 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
10070 align_flags of label_to_alignment.
10071 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
10073 * config/m68k/m68k.c: Do not use removed align_labels_value and
10075 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
10076 (LOOP_ALIGN): Likewise.
10077 (LABEL_ALIGN): Likewise.
10078 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
10079 Remove not used macro.
10080 (rs6000_loop_align): Change return type to align_flags.
10081 (rs6000_loop_align_max_skip): Remove.
10082 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
10083 Change return type to align_flags.
10084 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
10085 Remove not used macro.
10086 (rs6000_loop_align): Change return type to align_flags.
10087 (rs6000_loop_align_max_skip): Remove.
10088 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
10089 * config/rx/rx-protos.h (rx_align_for_label): Make it
10091 * config/rx/rx.c (rx_align_for_label): Change return type
10093 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
10095 into align_flags class.
10096 (LABEL_ALIGN): Likewise.
10097 (LOOP_ALIGN): Likewise.
10098 * config/s390/s390.c (s390_label_align): Use align_flags
10100 (s390_asm_output_function_label): Likewise.
10101 * config/sh/sh.c (sh_override_options_after_change):
10102 Use align_flags class directly without macros.
10103 (find_barrier): Likewise.
10104 (barrier_align): Likewise.
10105 (sh_loop_align): Likewise.
10106 * config/spu/spu.c (spu_option_override):
10107 Use align_flags_tuple::get_value instead of removed macros.
10108 (spu_sched_init): Likewise.
10109 * config/spu/spu.h (GTY): Likewise.
10110 * config/visium/visium.c (visium_option_override):
10111 Set "8" as default secondary alignment.
10112 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
10113 in order to guarantee secondary alignment of 8.
10114 * coretypes.h: Include align.h header file.
10115 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
10116 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
10117 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
10118 * doc/tm.texi.in: Likewise.
10119 * final.c (struct label_alignment): Remove not used structure.
10120 (LABEL_ALIGN): Change type to align_flags.
10121 (LOOP_ALIGN): Likewise.
10122 (JUMP_ALIGN): Likewise.
10123 (default_loop_align_max_skip): Remove.
10124 (default_label_align_max_skip): Likewise.
10125 (default_jump_align_max_skip): Likewise.
10126 (default_label_align_after_barrier_max_skip):
10127 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
10128 (LABEL_TO_MAX_SKIP): Remove.
10129 (label_to_alignment): Return align_flags type instead of integer.
10130 (label_to_max_skip): Remove.
10131 (align_fuzz): Use align_flags type.
10132 (compute_alignments): Use align_flags type and use align_flags::max
10133 to combine multiple alignments.
10134 (grow_label_align): Grow vec instead of C array.
10135 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
10136 (shorten_branches): Use align_flags type and use align_flags::max
10137 to combine multiple alignments.
10138 (final_scan_insn_1): Remove usage of secondary alignment that comes
10139 from label alignment, but instead use proper secondary alignment
10140 which is computed in grow_label_align.
10141 * flags.h (struct align_flags_tuple): Move to align.h.
10142 (struct align_flags): Likewise.
10143 (state_align_loops): Rename to align_loops.
10144 (state_align_jumps): Rename to align_jumps.
10145 (state_align_labels): Rename to align_labels.
10146 (state_align_functions): Rename to align_functions.
10147 (align_loops_log): Remove.
10148 (align_jumps_log): Remove.
10149 (align_labels_log): Remove.
10150 (align_functions_log): Remove.
10151 (align_loops_max_skip): Remove.
10152 (align_jumps_max_skip): Remove.
10153 (align_labels_max_skip): Remove.
10154 (align_functions_max_skip): Remove.
10155 (align_loops_value): Remove.
10156 (align_jumps_value): Remove.
10157 (align_labels_value): Remove.
10158 (align_functions_value): Remove.
10159 * output.h (label_to_alignment): Change return type to align_flags.
10160 (label_to_max_skip): Remove.
10161 * target.def: Remove loop_align_max_skip, label_align_max_skip,
10162 jump_align_max_skip macros.
10163 * targhooks.h (default_loop_align_max_skip): Remove.
10164 (default_label_align_max_skip): Likewise.
10165 (default_jump_align_max_skip): Likewise.
10166 (default_label_align_after_barrier_max_skip): Remove.
10167 * toplev.c (read_log_maxskip): Use ::normalize function.
10168 (parse_N_M): Remove not used argument and also call ::normalize.
10169 (parse_alignment_opts): Do not pass unused arguments.
10170 * varasm.c (assemble_start_function): Use directly align_functions
10171 instead of removed macros.
10172 * system.h: Do not poison removed macros.
10174 2018-07-17 Jakub Jelinek <jakub@redhat.com>
10176 PR middle-end/86539
10177 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
10178 and cond temporaries don't have reference type if iterator has
10179 pointer type. For init use &for_pre_body instead of pre_p if
10180 for_pre_body is non-empty.
10182 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
10184 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
10185 double-double modes to SFmode directly directly.
10186 (trunc<mode>sf2_fprs): Delete.
10188 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
10190 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
10191 for conversions between IFmode and the decimal floating point modes.
10192 (init_float128_ieee): Use the correct names for conversions between
10193 KFmode and the decimal floating point modes.
10195 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
10197 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
10198 for the conversions between TDmode and IFmode.
10199 (init_float128_ieee): Use more correct names for the conversions
10200 between TDmode and KFmode.
10202 2018-07-16 Jakub Jelinek <jakub@redhat.com>
10204 PR tree-optimization/86526
10205 * builtins.c (expand_builtin_memcmp): Formatting fixes.
10206 (inline_expand_builtin_string_cmp): Likewise.
10207 (inline_string_cmp): Likewise. Use c_readstr instead of
10208 builtin_memcpy_read_str. Add unit_mode temporary.
10210 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
10212 PR middle-end/86528
10213 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
10214 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
10216 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
10218 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
10219 Alphabetize prototypes of built-in functions, separating out
10220 built-in functions that are listed in this section but should be
10221 described elsewhere.
10223 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
10226 * expmed.c (emit_store_flag): Do not emit setcc followed by a
10227 conditional move when trapping comparison was split to a
10228 non-trapping one (and vice versa).
10230 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
10232 * config/s390/s390.c (s390_function_profiler): Generate nops
10233 instead of profiler call sequences.
10234 * config/s390/s390.opt: Add the new option.
10236 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
10238 * config/s390/s390.c (s390_function_profiler): Generate
10239 __mcount_loc section.
10240 * config/s390/s390.opt: Add the new option.
10242 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
10244 * common.opt: Add the new warning.
10245 * config/s390/s390.c (s390_function_profiler): Emit "brasl
10246 %r0,__fentry__" when -mfentry is specified.
10247 (s390_option_override_internal): Disallow -mfentry for 31-bit
10249 * config/s390/s390.opt: Add the new option.
10251 2018-07-16 Richard Biener <rguenther@suse.de>
10254 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
10255 for function-local FUNCTION_DECL and RESULT_DECL immediately.
10257 2018-07-16 Martin Liska <mliska@suse.cz>
10260 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
10263 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
10265 * config/arc/arcHS.md: Update ARCHS scheduling rules.
10267 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
10269 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
10271 * config/arc/arc-cpus.def (hs4x): New cpu.
10273 * config/arc/arc-tables.opt: Regenerate.
10274 * config/arc/arc.c (arc_sched_issue_rate): New function.
10275 (TARGET_SCHED_ISSUE_RATE): Define.
10276 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
10277 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
10279 (attr tune): Add ARCHS4x tune values.
10280 (attr tune_dspmpy): Define.
10281 (*tst): Correct instruction type.
10282 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
10283 * config/arc/arcHS4x.md: New file.
10284 * config/arc/fpu.md: Update instruction type attributes.
10285 * config/arc/t-multilib: Regenerate.
10287 2018-07-16 Tom de Vries <tdevries@suse.de>
10290 * var-tracking.c (vt_initialize): Fix pre_dec handling.
10292 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
10294 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
10297 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
10299 PR tree-optimization/86514
10300 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
10301 conversion to a boolean type from a type with greater precision.
10303 2018-07-16 Tom de Vries <tdevries@suse.de>
10305 * var-tracking.c (vt_initialize): Print adjusted insn slim if
10306 dump_flags request TDF_SLIM.
10308 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
10310 * fold-const.c (int_const_binop_1): Abstract...
10311 (wide_int_binop): ...wide int code here.
10312 (poly_int_binop): ...poly int code here.
10313 Abstract the rest of int_const_binop_1 into int_const_binop.
10314 * fold-const.h (wide_int_binop): New.
10315 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
10316 Remove useless PLUS/MINUS_EXPR case.
10317 (zero_nonzero_bits_from_vr): Move wide int code...
10318 (zero_nonzero_bits_from_bounds): ...here.
10319 (extract_range_from_binary_expr_1): Move mask optimization code...
10320 (range_easy_mask_min_max): ...here.
10321 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
10322 (range_easy_mask_min_max): New.
10324 2018-07-15 Jeff Law <law@redhat.com>
10327 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
10330 2018-07-14 Jim Wilson <jimw@sifive.com>
10332 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
10334 2018-07-14 Paul Koning <ni1d@arrl.net>
10336 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
10338 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
10340 * lto-streamer-out.c (copy_function_or_variable): Dump info about
10343 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
10344 Steve Munroe <munroesj52@gmail.com>
10346 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
10347 (_mm_andnot_si128): Likewise.
10348 (_mm_or_si128): Likewise.
10349 (_mm_xor_si128): Likewise.
10351 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
10353 PR middle-end/78809
10354 * builtins.c (expand_builtin_memcmp): Inline the calls first
10355 when result_eq is false.
10356 (expand_builtin_strcmp): Inline the calls first.
10357 (expand_builtin_strncmp): Likewise.
10358 (inline_string_cmp): New routine. Expand a string compare
10359 call by using a sequence of char comparison.
10360 (inline_expand_builtin_string_cmp): New routine. Inline expansion
10361 a call to str(n)cmp/memcmp.
10362 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
10364 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
10366 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
10368 * config/arm/driver-arm.c: Include arm-native.h.
10369 (host_detect_local_cpu): Use auto-generated data tables.
10370 (vendors, arm_cpu_table): Delete. Move part information to ...
10371 * config/arm/arm-cpus.in: ... here.
10372 * config/arm/parsecpu.awk (gen_native): New function.
10373 (vendor, part): New CPU fields.
10374 (END): Add support for building the native CPU detection tables.
10375 * config/arm/t-arm (arm-native.h): Add build rule.
10376 (driver-arm.o): Add dependency on arm-native.h.
10378 2018-07-13 Richard Biener <rguenther@suse.de>
10380 PR middle-end/85974
10381 * match.pd (addr1 - addr2): Allow either of the operand to
10384 2018-07-13 Tom de Vries <tdevries@suse.de>
10386 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
10389 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
10391 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
10392 arrays instead of numbered variables.
10394 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
10396 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
10397 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
10399 2018-07-13 Richard Biener <rguenther@suse.de>
10402 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
10403 instead of get_context_die.
10405 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
10406 Richard Biener <rguenther@suse.de>
10408 PR middle-end/86489
10409 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
10410 that the loop latch destination where phi is defined.
10412 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
10414 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
10415 (riscv_expand_epilogue): Add assertion to check interrupt mode.
10416 (riscv_set_current_function): Extract getting interrupt type to new
10418 (riscv_get_interrupt_type): New function.
10419 (riscv_merge_decl_attributes): New function, checking interrupt type is
10421 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
10423 2018-07-12 Paul Koning <ni1d@arrl.net>
10425 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
10428 2018-07-12 Paul Koning <ni1d@arrl.net>
10430 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
10431 zero reference, add doloop_end instead.
10432 * doc/md.texi (decrement_and_branch_until_zero): Remove.
10433 (Looping patterns): Remove decrement_and_branch_until_zero. Add
10434 detail for doloop_end.
10436 2018-07-12 Martin Sebor <msebor@redhat.com>
10439 * attribs.c (decl_attributes): Reject conflicting attributes before
10440 calling attribute handlers.
10442 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10444 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
10446 (gcc::dump_manager::get_dump_file_name): likewise.
10447 (dump_begin): Likewise.
10448 * dumpfile.h (dump_begin): Update prototype.
10449 (gcc::dump_manager::get_dump_file_name,
10450 gcc::dump_manager::get_dump_file_name): Update prototype.
10452 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10454 * internal-fn.h (vectorizable_internal_fn_p): New function.
10455 * tree-vect-slp.c (compatible_calls_p): Likewise.
10456 (vect_build_slp_tree_1): Remove nops argument. Handle calls
10457 to internal functions.
10458 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
10460 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10462 * fold-const.h (inverse_conditions_p): Declare.
10463 * fold-const.c (inverse_conditions_p): New function.
10464 * match.pd: Use inverse_conditions_p. Add folds of view_converts
10465 that test the inverse condition of a conditional internal function.
10466 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
10467 * internal-fn.c (internal_fn_mask_index): Handle conditional
10468 internal functions.
10469 (vectorized_internal_fn_supported_p): New function.
10470 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
10471 (any_pred_load_store): Replace with...
10472 (need_to_predicate): ...this new variable.
10473 (redundant_ssa_names): New variable.
10474 (ifcvt_can_use_mask_load_store): Move initial checks to...
10475 (ifcvt_can_predicate): ...this new function. Handle tree codes
10476 for which a conditional internal function exists.
10477 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
10478 instead of ifcvt_can_use_mask_load_store. Update after variable
10480 (predicate_load_or_store): New function, split out from
10481 predicate_mem_writes.
10482 (check_redundant_cond_expr): New function.
10483 (value_available_p): Likewise.
10484 (predicate_rhs_code): Likewise.
10485 (predicate_mem_writes): Rename to...
10486 (predicate_statements): ...this. Use predicate_load_or_store
10487 and predicate_rhs_code.
10488 (combine_blocks, tree_if_conversion): Update after above name changes.
10489 (ifcvt_local_dce): Handle redundant_ssa_names.
10490 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
10491 general conditional functions.
10492 * tree-vect-stmts.c (vectorizable_call): Likewise.
10494 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10495 Alan Hayward <alan.hayward@arm.com>
10496 David Sherwood <david.sherwood@arm.com>
10498 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
10499 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
10500 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
10501 plus and minus and convert them into IFN_COND_FMA-based sequences.
10502 (convert_mult_to_fma): Handle conditional plus and minus.
10504 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10506 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
10507 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
10508 (cond_fnms_optab): New optabs.
10509 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
10510 internal functions.
10511 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
10512 * internal-fn.h (get_conditional_internal_fn): Declare.
10513 (get_unconditional_internal_fn): Likewise.
10514 * internal-fn.c (cond_ternary_direct): New macro.
10515 (expand_cond_ternary_optab_fn): Likewise.
10516 (direct_cond_ternary_optab_supported_p): Likewise.
10517 (FOR_EACH_COND_FN_PAIR): Likewise.
10518 (get_conditional_internal_fn): New function.
10519 (get_unconditional_internal_fn): Likewise.
10520 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
10521 (gimple_match_op::gimple_match_op): Add a new overload for 5
10523 (gimple_match_op::set_op): Likewise.
10524 (gimple_resimplify5): Declare.
10525 * genmatch.c (decision_tree::gen): Generate simplifications for
10527 * gimple-match-head.c (gimple_simplify): Define an overload for
10528 5 operands. Handle calls with 5 arguments in the top-level overload.
10529 (convert_conditional_op): Handle conversions from unconditional
10530 internal functions to conditional ones.
10531 (gimple_resimplify5): New function.
10532 (build_call_internal): Pass a fifth operand.
10533 (maybe_push_res_to_seq): Likewise.
10534 (try_conditional_simplification): Try converting conditional
10535 internal functions to unconditional internal functions.
10536 Handle 3-operand unconditional forms.
10537 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
10538 Define ternary equivalents of the current rules for binary conditional
10539 internal functions.
10540 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
10541 ternary operations.
10542 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
10543 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
10544 (optab): Handle them.
10545 (SVE_COND_FP_TERNARY): New int iterator.
10546 (sve_fmla_op, sve_fmad_op): New int attributes.
10547 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
10548 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
10549 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
10551 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
10553 * target.def (preferred_else_value): New target hook.
10554 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
10555 * doc/tm.texi: Regenerate.
10556 * targhooks.h (default_preferred_else_value): Declare.
10557 * targhooks.c (default_preferred_else_value): New function.
10558 * internal-fn.h (conditional_internal_fn_code): Declare.
10559 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
10560 (get_conditional_internal_fn): Use it.
10561 (conditional_internal_fn_code): New function.
10562 * gimple-match.h (gimple_match_cond): New struct.
10563 (gimple_match_op): Add a cond member function.
10564 (gimple_match_op::gimple_match_op): Update all forms to take a
10566 * genmatch.c (expr::gen_transform): Use the same condition as res_op
10567 for the suboperation, but don't specify a particular else_value.
10568 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
10569 (visit_nary_op, visit_reference_op_load): Pass
10570 gimple_match_cond::UNCOND to the gimple_match_op constructor.
10571 * gimple-match-head.c: Include tree-eh.h
10572 (convert_conditional_op): New function.
10573 (maybe_resimplify_conditional_op): Likewise.
10574 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
10575 (gimple_resimplify2): Likewise.
10576 (gimple_resimplify3): Likewise.
10577 (gimple_resimplify4): Likewise.
10578 (maybe_push_res_to_seq): Return null for conditional operations.
10579 (try_conditional_simplification): New function.
10580 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
10582 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
10584 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
10586 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
10588 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
10590 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
10592 (hash_tree): Do not hash DECL_FCONTEXT
10593 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
10594 Do not stream DECL_FCONTEXT.
10595 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
10596 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
10598 2018-07-12 Richard Biener <rguenther@suse.de>
10601 * dwarf2out.c (gen_block_die): Only output blocks when they have
10602 at least one !DECL_IGNORED_P variable.
10604 2018-07-12 Richard Biener <rguenther@suse.de>
10607 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
10608 Remove -mieee-fp handling.
10610 2018-07-12 Richard Biener <rguenther@suse.de>
10612 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
10613 left-over from last patch.
10615 2018-07-12 Jakub Jelinek <jakub@redhat.com>
10617 PR tree-optimization/86492
10618 * gimple-ssa-store-merging.c
10619 (imm_store_chain_info::coalesce_immediate_stores): Call
10620 check_no_overlap even for the merge_overlapping case. Formatting fix.
10622 2018-07-12 Richard Biener <rguenther@suse.de>
10624 PR middle-end/86479
10625 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
10626 move possibly trapping operations into the conditional.
10628 2018-07-12 Richard Biener <rguenther@suse.de>
10630 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
10631 (vn_lookup_simplify_result): Remove recursion limit applied
10633 (vn_nary_build_or_lookup_1): Adjust.
10634 (try_to_simplify): Likewise.
10635 * gimple-match-head.c (gimple_resimplify1): Instead apply one
10637 (gimple_resimplify2): Likewise.
10638 (gimple_resimplify3): Likewise.
10639 (gimple_resimplify4): Likewise.
10641 2018-07-11 Jakub Jelinek <jakub@redhat.com>
10643 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
10644 Use __mmask64 type instead of __mmask8 for __M argument.
10645 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
10646 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
10648 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
10649 __mmask16 for __M argument.
10650 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
10651 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
10652 to __mmask16 instead of __mmask8.
10653 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
10654 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
10655 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
10656 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
10657 instead of __mmask16 for __U argument.
10658 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
10659 __mmask16 instead of __mmask8 for __U argument.
10660 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
10662 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
10664 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
10666 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
10668 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
10670 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
10672 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
10674 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
10676 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
10678 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
10680 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
10681 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
10682 return type as well as __M argument type and all casts from __mmask8
10684 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
10685 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
10686 return type as well as __M argument type and all casts from __mmask8
10688 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
10689 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
10690 return type as well as __M argument type and all casts from __mmask8
10692 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
10693 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
10694 return type as well as __M argument type and all casts from __mmask8
10696 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
10697 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
10700 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
10702 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
10703 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
10706 2018-07-11 Paul Koning <ni1d@arrl.net>
10708 * doc/md.texi (define_subst): Document how multiple occurrences of
10709 the same argument in the replacement pattern are handled.
10711 2018-07-11 Paul Koning <ni1d@arrl.net>
10713 * doc/extend.texi (Common Variable Attributes): Move "mode" into
10714 alphabetical order.
10715 (Common Type Attributes): Add "mode" attribute.
10717 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
10719 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
10720 stream DECL_ORIGINAL_TYPE.
10721 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
10722 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
10723 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
10724 Do not walk original type.
10725 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
10727 (write_ts_decl_non_common_tree_pointers): Do not stream
10729 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
10730 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
10732 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10734 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
10735 thread twice from the same starting edge.
10737 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
10739 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
10740 * gimple.c (gimple_call_nonnull_result_p): ...here...
10741 (gimple_call_nonnull_arg): ...and here.
10742 * gimple.h (gimple_call_nonnull_result_p): New.
10743 (gimple_call_nonnull_arg): New.
10745 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
10747 * config/arm/arm-cpus.in: Move information from fpu field of each
10748 cpu definition to the isa field.
10749 * config/arm/parsecpu.awk (fpu): Delete match rule.
10750 (gen_comm_data): Don't add bits from the CPU's FPU entry.
10752 2018-07-11 Richard Biener <rguenther@suse.de>
10755 * dwarf2out.c (init_sections_and_labels): Use
10756 output_asm_line_debug_info consistently.
10757 (dwarf2out_early_finish): Likewise.
10758 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
10761 2018-07-11 Richard Biener <rguenther@suse.de>
10763 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
10764 Rework father_bb setting in a way to avoid propagating constants
10765 multiple times on a loop body.
10767 2018-07-10 Mark Wielaard <mark@klomp.org>
10770 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
10772 2018-07-10 Richard Biener <rguenther@suse.de>
10774 * hash-map.h (hash_map::iterator::operator*): Return
10775 references to key and value.
10777 2018-07-10 Jakub Jelinek <jakub@redhat.com>
10780 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
10781 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
10782 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
10783 (gimplify_omp_for): For composite loops, move outer
10784 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
10785 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
10786 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
10787 TREE_LIST for both the original class iterator and the "last" helper
10788 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
10789 loop, remember has_decl_expr from outer composite loops for the
10790 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
10792 2018-07-09 Martin Sebor <msebor@redhat.com>
10794 PR middle-end/77357
10795 PR middle-end/86428
10796 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
10797 accessing implicitly initialized array elements.
10798 * expr.c (string_constant): Handle string initializers of
10799 character arrays within aggregates.
10800 * gimple-fold.c (fold_array_ctor_reference): Add argument.
10801 Store element offset. As a special case, handle zero size.
10802 (fold_nonarray_ctor_reference): Same.
10803 (fold_ctor_reference): Add argument. Store subobject offset.
10804 * gimple-fold.h (fold_ctor_reference): Add argument.
10806 2018-07-09 Paul Koning <ni1d@arrl.net>
10808 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
10809 (pdp11_insn_cost): New function.
10810 (pdp11_md_asm_adjust): New function.
10811 (TARGET_INVALID_WITHIN_DOLOOP): Define.
10812 (pdp11_rtx_costs): Update to match machine better.
10813 (output_addr_const_pdp11): Correct format mismatch warnings.
10814 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
10815 * config/pdp11/pdp11.md: General change to add base_cost and/or
10816 length attributes for use by new pdp11_insn_cost function.
10817 (MIN_BRANCH): Correct definition.
10819 (doloop_end): Use standard pattern name for looping pattern.
10820 (doloop_end_nocc): New.
10821 (movsf): Add another constraint alternative.
10822 (zero_extendqihi2): Add constraint alternatives for not in place
10824 (zero_extendhisi2): Remove.
10825 (shift patterns): Add CC handling variants.
10829 (define_peephole2): New peephole to recognize mov that sets CC for
10832 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10834 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
10835 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
10838 2018-07-09 Jakub Jelinek <jakub@redhat.com>
10841 * real.c (real_nextafter): Return true if result is denormal.
10843 2018-07-09 Martin Liska <mliska@suse.cz>
10845 * common.opt: Add back wrongly removed attribute.
10847 2018-07-09 Richard Biener <rguenther@suse.de>
10850 * dwarf2out.c (gen_block_die): For an early generated DIE
10851 always output high/low PC attributes.
10853 2018-07-09 Tom de Vries <tdevries@suse.de>
10855 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
10856 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
10857 onto VAR_DECL with abstract origin.
10859 2018-07-07 Jim Wilson <jimw@sifive.com>
10861 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
10863 2018-07-07 Tom de Vries <tdevries@suse.de>
10865 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
10867 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
10869 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
10871 (combine_bound): Use wide-int overflow calculation instead of
10873 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
10875 * fold-const.c (int_const_binop_2): Same.
10876 (extract_muldiv_1): Same.
10877 (fold_div_compare): Same.
10878 (fold_abs_const): Same.
10880 * poly-int.h (add): Same.
10884 * predict.c (predict_iv_comparison): Same.
10885 * profile-count.c (slow_safe_scale_64bit): Same.
10886 * simplify-rtx.c (simplify_const_binary_operation): Same.
10887 * tree-chrec.c (tree_fold_binomial): Same.
10888 * tree-data-ref.c (split_constant_offset_1): Same.
10889 * tree-if-conv.c (idx_within_array_bound): Same.
10890 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
10891 * tree-ssa-phiopt.c (minmax_replacement): Same.
10892 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
10893 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
10894 * vr-values.c (vr_values::adjust_range_with_scev): Same.
10895 * wide-int.cc (wi::add_large): Same.
10896 (wi::mul_internal): Same.
10897 (wi::sub_large): Same.
10898 (wi::divmod_internal): Same.
10899 * wide-int.h: Change overflow type to overflow_type for neg, add,
10900 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
10901 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
10902 mul_internal, divmod_internal.
10903 (overflow_type): New enum.
10904 (accumulate_overflow): New.
10906 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
10908 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
10909 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
10911 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
10913 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
10914 argument is checked for zero before entering loop, avoid checking again.
10916 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
10918 * gimplify.h (generic_expr_could_trap_p): Set as global function.
10919 * gimplify.h (generic_expr_could_trap_p): Likwise.
10920 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
10922 2018-07-06 Jakub Jelinek <jakub@redhat.com>
10924 PR tree-optimization/86401
10925 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
10926 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
10927 (fold_bit_and_mask): ... here. New helper function for match.pd.
10928 * fold-const.h (fold_bit_and_mask): Declare.
10929 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
10931 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
10934 * target.def (translate_mode_attribute): New hook.
10935 * targhooks.h (default_translate_mode_attribute): Declare.
10936 * targhooks.c (default_translate_mode_attribute): New function.
10937 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
10938 * doc/tm.texi: Regenerate.
10939 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
10940 (rs6000_translate_mode_attribute): New function.
10942 2018-07-06 Paul Koning <ni1d@arrl.net>
10944 * doc/md.texi (define_split): Document DONE and FAIL.
10945 (define_peephole2): Ditto.
10947 2018-07-05 Jeff Law <law@redhat.com>
10949 PR tree-optimization/86010
10950 * tree-ssa-dse.c (compute_trims): More aggressively trim at
10951 both the head and tail of mem* and str* calls.
10953 2018-07-05 Jim Wilson <jimw@sifive.com>
10955 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
10957 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
10959 * config/aarch64/aarch64-simd.md: correct flags text for
10960 MIN_EXPR replacement.
10962 2018-07-05 James Clarke <jrtc27@jrtc27.com>
10964 * configure: Regenerated.
10966 2018-07-05 Carl Love <cel@us.ibm.com>
10968 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
10969 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
10970 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
10971 VSX_BUILTIN_DOUBLEL_V4SF.
10973 2018-07-05 Martin Sebor <msebor@redhat.com>
10976 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
10977 than its domain to compute its the upper bound of a char array.
10979 2018-07-05 Nathan Sidwell <nathan@acm.org>
10981 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
10982 * doc/cpp.texi: Update comment.
10983 * doc/tm.texi: Rebuilt.
10984 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
10985 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
10986 * doc/extend.texi (Backwards Compatibility): Clarify it is system
10987 headers affected by extern "C".
10988 * system.h: Poison NO_IMPLICIT_EXTERN_C.
10989 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
10990 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
10991 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
10992 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
10993 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
10994 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
10995 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
10996 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
10997 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
10998 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
10999 config/sparc/sp64-elf.h, config/spu/spu.h,
11000 config/stormy16/stormy16.h, config/v850/v850.h,
11001 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
11002 define NO_IMPLICIT_EXTERN_C.
11003 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
11005 2018-07-05 Tamar Christina <tamar.christina@arm.com>
11008 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
11009 instead of GET_MODE_SIZE when comparing Units.
11011 2018-07-05 Tamar Christina <tamar.christina@arm.com>
11014 * rtlanal.c (set_noop_p): Constrain on mode change,
11015 include hard-reg-set.h
11017 2018-07-05 Tamar Christina <tamar.christina@arm.com>
11019 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
11021 2018-07-05 Jakub Jelinek <jakub@redhat.com>
11024 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
11027 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
11029 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
11031 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
11034 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
11036 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
11038 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
11040 PR middle-end/86380
11041 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
11043 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
11045 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
11046 neg_*_op* variables.
11048 2018-07-04 Martin Liska <mliska@suse.cz>
11050 * tree-switch-conversion.c: Define
11051 max_ratio_for_speed and max_ratio_for_size constants.
11053 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
11054 Martin Liska <mliska@suse.cz>
11056 PR middle-end/66240
11059 * common.opt: Rename align options with 'str_' prefix.
11060 * common/config/i386/i386-common.c (set_malign_value): New
11062 (ix86_handle_option): Use it to set -falign-* options/
11063 * config/aarch64/aarch64-protos.h (struct tune_params): Change
11064 type from int to string.
11065 * config/aarch64/aarch64.c: Update default values from int
11067 * config/alpha/alpha.c (alpha_override_options_after_change):
11069 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
11070 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11071 max skip conditionally.
11072 * config/i386/freebsd.h (SUBALIGN_LOG): New.
11073 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11074 max skip conditionally.
11075 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11076 max skip conditionally.
11077 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
11078 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11079 max skip conditionally.
11080 * config/i386/i386.c (struct ptt): Change type from int to
11082 (ix86_default_align): Set default values.
11083 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
11084 max skip conditionally.
11085 * config/i386/iamcu.h (SUBALIGN_LOG): New.
11086 (ASM_OUTPUT_MAX_SKIP_ALIGN):
11087 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
11088 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11089 max skip conditionally.
11090 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
11091 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
11092 * config/i386/x86-64.h (SUBALIGN_LOG): New.
11093 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
11094 max skip conditionally.
11095 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
11096 * config/ia64/ia64.c (ia64_option_override): Set default values
11097 for alignment options.
11098 * config/m68k/m68k.c: Handle new str_align_* options.
11099 * config/mips/mips.c (mips_set_compression_mode): Change
11101 (mips_option_override): Set default values for options.
11102 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
11104 * config/rs6000/rs6000.c (rs6000_option_override_internal):
11106 * config/rx/rx.c (rx_option_override): Likewise.
11107 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
11108 (LABEL_ALIGN): Use align_labels_log.
11109 (LOOP_ALIGN): Use align_loops_align.
11110 * config/s390/s390.c (s390_asm_output_function_label): Use new
11112 * config/sh/sh.c (sh_override_options_after_change):
11113 Change type of constants.
11114 * config/spu/spu.c (spu_sched_init): Likewise.
11115 * config/sparc/sparc.c (sparc_option_override): Set default
11116 values for options.
11117 * config/visium/visium.c (visium_option_override): Likewise.
11118 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
11119 emit p2align format with last argument if it's not needed.
11120 * doc/invoke.texi: Document extended format of -falign-*.
11121 * final.c: Use align_labels alignment.
11122 * flags.h (struct target_flag_state): Change type to use
11124 (struct align_flags_tuple): New.
11125 (struct align_flags): Likewise.
11126 (align_loops_log): Redefine macro to use new types.
11127 (align_loops_max_skip): Redefine macro to use new types.
11128 (align_jumps_log): Redefine macro to use new types.
11129 (align_jumps_max_skip): Redefine macro to use new types.
11130 (align_labels_log): Redefine macro to use new types.
11131 (align_labels_max_skip): Redefine macro to use new types.
11132 (align_functions_log): Redefine macro to use new types.
11133 (align_loops): Redefine macro to use new types.
11134 (align_jumps): Redefine macro to use new types.
11135 (align_labels): Redefine macro to use new types.
11136 (align_functions): Redefine macro to use new types.
11137 (align_functions_max_skip): Redefine macro to use new types.
11138 (align_loops_value): New macro.
11139 (align_jumps_value): New macro.
11140 (align_labels_value): New macro.
11141 (align_functions_value): New macro.
11142 * function.c (invoke_set_current_function_hook): Propagate
11143 alignment values from flags to global variables default in
11145 * ipa-icf.c (sem_function::equals_wpa): Use
11146 cl_optimization_option_eq instead of memcmp.
11147 * lto-streamer.h (cl_optimization_stream_out): Support streaming
11149 (cl_optimization_stream_in): Likewise.
11150 * optc-save-gen.awk: Support strings in cl_optimization.
11151 * opth-gen.awk: Likewise.
11152 * opts.c (finish_options): Remove error checking of invalid
11154 (MAX_CODE_ALIGN): Remove.
11155 (MAX_CODE_ALIGN_VALUE): Likewise.
11156 (parse_and_check_align_values): New function.
11157 (check_alignment_argument): Likewise.
11158 (common_handle_option): Use check_alignment_argument.
11159 * opts.h (parse_and_check_align_values): Declare.
11160 * toplev.c (init_alignments): Remove.
11161 (read_log_maxskip): New.
11162 (parse_N_M): Likewise.
11163 (parse_alignment_opts): Likewise.
11164 (backend_init_target): Remove usage of init_alignments.
11165 * toplev.h (parse_alignment_opts): Declare.
11166 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
11168 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
11169 * tree.c (cl_option_hasher::equal): New.
11170 * varasm.c: Use new global macros.
11172 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
11174 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
11175 Use a simpler align directive also if MAXSKIP = ALIGN-1.
11176 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11177 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11178 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11179 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
11180 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
11181 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
11182 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
11183 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11184 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11185 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11186 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11188 2018-07-04 Martin Liska <mliska@suse.cz>
11189 Jonathan Wakely <jwakely@redhat.com>
11191 * coverage.c: Use correct type.
11192 * doc/invoke.texi: Language correction.
11194 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
11197 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
11198 ENDBRANCH for non-tail call which may return via indirect branch.
11199 * doc/extend.texi: Document indirect_return attribute.
11201 2018-07-03 Martin Sebor <msebor@redhat.com>
11203 PR tree-optimization/86274
11204 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
11206 (format_floating): Correct handling of infinities and NaNs.
11208 2018-07-03 Martin Sebor <msebor@redhat.com>
11210 * print-tree.c (print_real_cst): New function.
11211 (print_node_brief): Call it.
11212 (print_node): Ditto.
11214 2018-07-03 Jeff Law <law@redhat.com>
11216 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
11217 into a single pattern.
11219 * config/h8300/h8300.md (ors code_iterator): New.
11220 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
11221 a single pattern and single splitter.
11222 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
11223 (iorqi3_1, xorqi3_1): Likewise.
11224 (iorqi3, xorqi3 expanders): Similarly.
11226 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
11227 (movmd_internal) into a single pattern using the P mode iterator.
11228 (movmd splitters): Similarly.
11229 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
11230 (movsd splitters): Similarly.
11232 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
11233 ADDB, ADDW and ADDL into a single ADD attribute which selects the
11234 right table based on the size of the operand.
11235 * config/h8300/h8300.md (length_table): Corresponding changes. All
11236 references to "addb", "addw" and "addl" changed to "add".
11237 (btst patterns): Merge two variants into a single pattern.
11238 (tstqi, tsthi): Likewise.
11239 (addhi3_incdec, addsi3_incdec): Likewise.
11240 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
11241 (mulhi3, mulsi3): Likewise.
11242 (udivhi3, udivsi3): Likewise.
11243 (divhi3, divsi3): Likewise.
11244 (andorqi3, andorhi3, andorsi3): Likewise.
11246 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
11249 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
11250 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
11252 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11254 PR tree-optimization/85694
11255 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
11256 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
11258 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
11259 (<u>avg<mode>3_ceil): New patterns.
11261 2018-07-03 David Malcolm <dmalcolm@redhat.com>
11263 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
11264 scan-tree-dump directive.
11265 * gcc.dg/vect/slp-perm-2.c: Likewise.
11266 * gcc.dg/vect/slp-perm-3.c: Likewise.
11267 * gcc.dg/vect/slp-perm-5.c: Likewise.
11268 * gcc.dg/vect/slp-perm-6.c: Likewise.
11269 * gcc.dg/vect/slp-perm-7.c: Likewise.
11270 * gcc.dg/vect/slp-perm-8.c: Likewise.
11272 2018-07-03 Marek Polacek <polacek@redhat.com>
11274 PR middle-end/86202
11275 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
11277 2018-07-03 Richard Biener <rguenther@suse.de>
11280 * tree-ssa-structalias.c (find_func_clobbers): Properly
11281 handle indirect calls.
11283 2018-07-03 Jeff Law <law@redhat.com>
11285 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
11286 (shifts): New code iterator.
11287 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
11288 expander. Fix HImode handling on H8/SX.
11289 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
11290 (subqi3, subhi3, subsi3 expanders): Likewise.
11291 (andqi3, andhi3, andsi3 expanders): Likewise.
11292 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
11293 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
11294 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
11295 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
11296 (zero_extendqihi2, zero_extendqisi2): Likewise.
11297 (extendqihi2, extendqisi2): Likewise.
11298 (rotlqi3, rotlhi3, rotlsi3): Likewise.
11299 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
11300 (rotlqi3_1, rotlhi3_1): Likewise.
11301 (logicalhi3_sn, logicalsi3_sn): Likewise.
11302 (logicalhi3, logicalsi3): Likewise.
11304 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11306 * tree-vect-patterns.c (vect_recog_rotate_pattern)
11307 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
11308 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
11309 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
11310 type to append_pattern_def_seq instead of creating a stmt_vec_info
11312 (build_mask_conversion): Likewise. Remove vinfo argument.
11313 (vect_add_conversion_to_patterm): Likewise, renaming to...
11314 (vect_add_conversion_to_pattern): ...this.
11315 (vect_recog_mask_conversion_pattern): Update call to
11316 build_mask_conversion. Pass the vector type to
11317 append_pattern_def_seq here too.
11318 (vect_recog_gather_scatter_pattern): Update call to
11319 vect_add_conversion_to_pattern.
11321 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11323 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
11324 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11325 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
11326 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
11327 STMT_VINFO_PATTERN_DEF_SEQ to null here.
11328 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
11329 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
11330 append_pattern_def_seq instead of new_pattern_def_seq.
11331 (vect_recog_divmod_pattern): Do both of the above.
11332 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
11335 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11337 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
11338 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
11339 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
11340 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
11341 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
11342 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
11343 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
11344 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
11345 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
11346 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
11347 parameter with a single stmt_vec_info.
11348 (vect_recog_func_ptr): Likewise.
11349 (vect_recog_gather_scatter_pattern): Likewise, folding in...
11350 (vect_try_gather_scatter_pattern): ...this.
11351 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
11352 the stmt_vec_info of the statement to be matched. Don't clear
11353 STMT_VINFO_RELATED_STMT.
11354 (vect_pattern_recog): Update call accordingly.
11356 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11358 PR tree-optimization/85694
11359 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
11360 (uavgM3_ceil): Document new optabs.
11361 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
11362 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
11364 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
11365 (savg_ceil_optab): New optabs.
11366 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
11367 (vect_vect_recog_func_ptrs): Add it.
11368 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
11369 constant directly from the associated lhs.
11371 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11373 * tree-vect-patterns.c (vect_split_statement): New function.
11374 (vect_convert_input): Use it to try to split an existing cast.
11376 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11378 * poly-int.h (print_hex): New function.
11379 * dumpfile.h (dump_dec, dump_hex): Declare.
11380 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
11381 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
11382 min_input_precision, operation_precision and operation_sign.
11383 * tree-vect-patterns.c (vect_get_range_info): New function.
11384 (vect_same_loop_or_bb_p, vect_single_imm_use)
11385 (vect_operation_fits_smaller_type): Delete.
11386 (vect_look_through_possible_promotion): Add an optional
11387 single_use_p parameter.
11388 (vect_recog_over_widening_pattern): Rewrite to use new
11389 stmt_vec_info infomration. Handle one operation at a time.
11390 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
11391 (vect_truncatable_operation_p, vect_set_operation_type)
11392 (vect_set_min_input_precision): New functions.
11393 (vect_determine_min_output_precision_1): Likewise.
11394 (vect_determine_min_output_precision): Likewise.
11395 (vect_determine_precisions_from_range): Likewise.
11396 (vect_determine_precisions_from_users): Likewise.
11397 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
11398 (vect_vect_recog_func_ptrs): Put over_widening first.
11400 (vect_pattern_recog): Call vect_determine_precisions.
11402 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
11404 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
11405 statements that have been replaced by further pattern statements.
11406 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
11408 2018-07-03 Richard Biener <rguenther@suse.de>
11410 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
11411 always set *dt. Dump vectype in vectype overload.
11412 * dumpfile.h (dump_gimple_expr): New function.
11413 (dump_gimple_expr_loc): Likewise.
11414 * dumpfile.c (dump_gimple_expr): New function.
11415 (dump_gimple_expr_loc): Likewise.
11417 2018-07-02 Jeff Law <law@redhat.com>
11419 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
11420 the H8/300, H8/300H and H8/S variants into a single pattern.
11421 (movhi_h8300, movqi_h8300hs): Similarly.
11422 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
11423 (QHI mode iterator): New.
11425 * config/h8300/h8300.md: Remove trailing whitespace.
11427 2018-07-02 Jim Wilson <jimw@sifive.com>
11429 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
11430 instead of emit_insn for interrupt returns.
11431 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
11432 (riscv_sret, riscv_uret): Likewise.
11434 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11436 * pretty-print.c (selftest::test_pp_format): Move save and restore
11437 of quotes to class auto_fix_quotes, and add an instance.
11438 * selftest.c: Include "intl.h".
11439 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
11440 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
11441 * selftest.h (selftest::auto_fix_quotes): New class.
11443 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
11445 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
11446 (aarch64_sve_prepare_conditional_op): Remove.
11447 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
11448 Allow aarch64_simd_reg_or_zero as select operand; remove
11449 the aarch64_sve_prepare_conditional_op call.
11450 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
11451 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11452 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
11453 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
11454 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
11455 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
11456 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
11457 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
11458 and a splitters to match all of the *_any patterns.
11459 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
11461 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
11462 (SVE_COND_FP_BINARY_REV): Remove.
11463 (sve_int_op_rev, sve_fp_op_rev): New.
11464 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
11465 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
11466 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
11467 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
11468 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
11469 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
11470 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
11471 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
11472 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
11474 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
11475 Remove match_dup 1 from the inner unspec.
11476 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
11478 * config/aarch64/aarch64.md (movprfx): New attr.
11479 (length): Default movprfx to 8.
11480 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
11481 (*madd<SVE_I>, *msub<SVE_I): Likewise.
11482 (*<su>mul<SVE_I>3_highpart): Likewise.
11483 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
11484 (*v<ASHIFT><SVE_I>3): Likewise.
11485 (*<su><MAXMIN><SVE_I>3): Likewise.
11486 (*<su><MAXMIN><SVE_F>3): Likewise.
11487 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
11488 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
11489 (*div<SVE_F>4): Likewise.
11491 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
11493 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
11496 2018-07-02 Richard Biener <rguenther@suse.de>
11498 PR tree-optimization/86363
11499 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
11500 memset argument refers to a non-variable address.
11502 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11504 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
11505 {PLUS,MINUS}_EXPR code to...
11506 (adjust_symbolic_bound): ...here,
11507 (combine_bound): ...here,
11508 (set_value_range_with_overflow): ...and here.
11510 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
11512 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
11514 (extract_range_from_abs_expr): ...here.
11516 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
11518 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
11519 -fno-omit-frame-pointer when not optimizing.
11521 2018-07-02 Martin Liska <mliska@suse.cz>
11524 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
11525 (propagate_nothrow): Likewise.
11527 2018-07-02 Martin Liska <mliska@suse.cz>
11530 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
11532 2018-07-02 David Malcolm <dmalcolm@redhat.com>
11534 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
11535 function in r262149, changing "loc" param from source_location to
11536 const dump_location_t &.
11537 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
11538 declaration, as above.
11540 2018-07-01 Paul Koning <ni1d@arrl.net>
11542 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
11543 -munit-asm, -mgnu-asm, -mdec-asm.
11544 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
11545 (pdp11_output_labelref): New.
11546 (pdp11_output_def): New.
11547 (pdp11_output_addr_vec_elt): New.
11548 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
11549 %# and %@ format codes.
11550 (pdp11_option_override): New.
11551 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
11552 (pdp11_output_ident): New.
11553 (pdp11_asm_named_section): New.
11554 (pdp11_asm_init_sections): New.
11555 (pdp11_file_start): New.
11556 (pdp11_file_end): New.
11557 (output_ascii): Use .ascii/.asciz for -mdec-asm.
11558 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
11559 %o, like %c but octal.
11560 (pdp11_option_override): New.
11561 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
11563 (DATA_SECTION_ASM_OP): Ditto.
11564 (READONLY_DATA_SECTION_ASM_OP): New.
11565 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
11566 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
11567 (ASM_OUTPUT_LABELREF): Ditto.
11568 (ASM_OUTPUT_DEF): Ditto.
11569 (ASM_OUTPUT_EXTERNAL): New.
11570 (ASM_OUTPUT_SOURCE_FILENAME): New.
11571 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
11572 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
11573 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
11574 %# and %@ format codes.
11575 * config/pdp11/pdp11.opt (mgnu-asm): New.
11576 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
11577 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
11578 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
11580 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
11582 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
11583 dereferencing path[] beyond its length.
11585 (debug_all_paths): New.
11586 (rewire_first_differing_edge): New.
11587 (adjust_paths_after_duplication): New.
11588 (duplicate_thread_path): Call adjust_paths_after_duplication.
11590 (thread_through_all_blocks): Add new argument to
11591 duplicate_thread_path.
11593 2018-06-30 Jim Wilson <jimw@sifive.com>
11595 * config/riscv/predicates.md (p2m1_shift_operand): New.
11596 (high_mask_shift_operand): New.
11597 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
11598 pattern using p2m1_shift_operand.
11599 (lshsi3_zero_extend_3+2): New combiner pattern using
11600 high_mask_shift_operand.
11602 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11604 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
11606 (vect_recog_rotate_pattern): ...here.
11607 (vect_convert_input): Try to insert casts of invariants in the
11609 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
11610 preheader to be empty.
11612 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11614 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
11615 vector type. If given, install it in the new statement's
11616 STMT_VINFO_VECTYPE.
11617 (vect_element_precision): New function.
11618 (vect_unpromoted_value): New struct.
11619 (vect_unpromoted_value::vect_unpromoted_value): New function.
11620 (vect_unpromoted_value::set_op): Likewise.
11621 (vect_look_through_possible_promotion): Likewise.
11622 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
11623 (vect_widened_op_tree, vect_convert_input): Likewise.
11624 (vect_convert_inputs, vect_convert_output): Likewise.
11625 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
11626 to handle the optional cast of the multiplication result and
11627 vect_widened_op_tree to detect the widened multiplication itself.
11628 Do not require the input and output of promotion casts to have
11629 the same sign, but base the signedness of the operation on the
11630 input rather than the result. If the pattern includes two
11631 promotions, check that those promotions have the same sign.
11632 Do not restrict the MULT_EXPR handling to a double-width result;
11633 handle quadruple-width results and wider. Use vect_convert_inputs
11634 to convert the inputs to the common type.
11635 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
11636 to handle the optional cast of the ABS result. Also allow a sign
11637 change or a sign extension between the ABS and MINUS.
11638 Use vect_widened_op_tree to detect the widened subtraction and use
11639 vect_convert_inputs to convert the inputs to the common type.
11640 (vect_handle_widen_op_by_const): Delete.
11641 (vect_recog_widen_op_pattern): New function.
11642 (vect_recog_widen_mult_pattern): Use it.
11643 (vect_recog_widen_shift_pattern): Likewise.
11644 (vect_recog_widen_sum_pattern): Use
11645 vect_look_through_possible_promotion to handle the promoted
11648 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11650 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
11651 the containing gimple_seq *.
11652 * gimple-iterator.h (gsi_for_stmt): Declare it.
11653 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
11654 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
11655 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
11656 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
11657 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
11659 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
11661 (vect_mark_pattern_stmts): ...here. Handle cases in which the
11662 statement being replaced is part of an existing pattern
11663 definition sequence, inserting the new pattern statements before
11665 (vect_pattern_recog_1): Don't return a bool. If the statement
11666 is already part of a pattern, instead apply pattern matching
11667 to the pattern definition statements. Don't clear the
11668 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
11669 (vect_pattern_recog): Don't break after the first match;
11670 continue processing the pattern definition statements instead.
11671 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
11673 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11675 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
11676 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
11677 (vect_recog_widen_sum_pattern): Use it.
11679 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11681 * tree-vect-loop.c (vectorizable_reduction): Assert that the
11682 phi is not a pattern statement and has not been replaced by
11683 a pattern statement.
11684 * tree-vect-patterns.c (type_conversion_p): Don't check
11685 STMT_VINFO_IN_PATTERN_P.
11686 (vect_recog_vector_vector_shift_pattern): Likewise.
11687 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
11688 the pattern statement rather than the original statement; check
11689 directly for a WIDEN_MULT_EXPR here.
11690 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
11691 vect_is_simple_use to return the pattern statement rather
11692 than the original statement; use is_pattern_stmt_p to check
11693 for such a pattern statement.
11694 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
11695 to return the pattern statement rather than the original statement;
11696 don't do the same transformation here.
11697 (vect_is_simple_use): If the defining statement has been replaced
11698 by a pattern statement, return the pattern statement instead.
11699 Remove the corresponding (local) transformation from the vectype
11702 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
11704 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
11705 end and default to null.
11706 * tree-vect-loop.c (vect_create_epilog_for_reduction)
11707 (vectorizable_reduction): Update calls accordingly, dropping the
11708 gimple ** argument if the passed-back statement isn't needed.
11709 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
11710 (vect_recog_rotate_pattern): Likewise.
11711 (vect_recog_mask_conversion_pattern): Likewise.
11712 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
11713 (vect_mask_constant_operand_p): Likewise.
11714 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
11715 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
11716 (get_group_load_store_type, get_load_store_type): Likewise.
11717 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
11718 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
11719 (vectorizable_conversion, vectorizable_assignment): Likewise.
11720 (vectorizable_shift, vectorizable_operation): Likewise.
11721 (vectorizable_store, vect_is_simple_cond): Likewise.
11722 (vectorizable_condition, vectorizable_comparison): Likewise.
11723 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
11724 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
11725 and move it to the end. Cope with null def_stmt_outs.
11727 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
11729 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
11731 2018-06-29 Jeff Law <law@redhat.com>
11733 * config/v850/v850.c (v850_legitimate_address_p): Handle large
11734 displacements for TARGET_V850E2V3 and newer.
11735 (TARGET_LRA_P): Remove. Defaults to LRA now.
11736 * config/v850/v850.md (sign23byte_load): Remove.
11737 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
11738 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
11740 2018-06-29 Martin Liska <mliska@suse.cz>
11743 * coverage.c (coverage_init): Mangle full path name.
11744 * doc/invoke.texi: Document the change.
11745 * gcov-io.c (mangle_path): New.
11746 * gcov-io.h (mangle_path): Likewise.
11747 * gcov.c (mangle_name): Use mangle_path for path mangling.
11749 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11751 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
11752 if starting source register is not even.
11754 2018-06-29 Martin Liska <mliska@suse.cz>
11756 PR tree-optimization/86263
11757 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
11758 Make edge redirection.
11760 2018-06-29 David Malcolm <dmalcolm@redhat.com>
11762 * dumpfile.c (dump_loc): Add indentation based on scope depth.
11763 (dump_scope_depth): New variable.
11764 (get_dump_scope_depth): New function.
11765 (dump_begin_scope): New function.
11766 (dump_end_scope): New function.
11767 * dumpfile.h (get_dump_scope_depth): New declaration.
11768 (dump_begin_scope): New declaration.
11769 (dump_end_scope): New declaration.
11770 (class auto_dump_scope): New class.
11771 (AUTO_DUMP_SCOPE): New macro.
11772 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
11775 2018-06-29 Richard Biener <rguenther@suse.de>
11777 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
11778 compute_all_dependences succeeds.
11779 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
11780 exceed --param loop-max-datarefs-for-datadeps.
11782 2018-06-29 Jakub Jelinek <jakub@redhat.com>
11784 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
11786 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
11789 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
11790 alternative 0 in preferred_for_speed attribute.
11792 2018-06-28 Paul Koning <ni1d@arrl.net>
11794 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
11795 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
11796 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
11797 * config/pdp11/pdp11.md: Correct "length" attribute calculation
11798 for shift insn patterns.
11800 2018-06-28 David Malcolm <dmalcolm@redhat.com>
11802 * cgraph.c (cgraph_node::get_body): Replace assignments to
11803 "dump_file" with calls to set_dump_file.
11804 * dumpfile.c (alt_dump_file): Make static, and group with...
11805 (alt_flags): ...this definition.
11806 (dumps_are_enabled): New variable.
11807 (refresh_dumps_are_enabled): New function.
11808 (set_dump_file): New function.
11809 (set_alt_dump_file): New function.
11810 (gcc::dump_manager::dump_start): Replace assignments to
11811 "dump_file" and "alt_dump_file" with calls to set_dump_file and
11813 (gcc::dump_manager::dump_finish): Likewise.
11814 * dumpfile.h (alt_dump_file): Delete decl.
11815 (dumps_are_enabled): New variable decl.
11816 (set_dump_file): New function decl.
11817 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
11819 * tree-nested.c (lower_nested_functions): Replace assignments to
11820 "dump_file" with calls to set_dump_file.
11822 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
11824 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
11825 goto_locus of each outgoing edge of each basic block.
11827 2018-06-28 Richard Biener <rguenther@suse.de>
11829 * dwarf2out.c (decl_scope_table): Remove.
11830 (push_decl_scope): Likewise.
11831 (pop_decl_scope): Likewise.
11832 (gen_type_die_for_member): Do not call push/pop_decl_scope.
11833 (gen_struct_or_union_type_die): Likewise.
11834 (gen_tagged_type_die): Likewise.
11835 (dwarf2out_init): Do not initialize decl_scope_table.
11836 (dwarf2out_c_finalize): Do not free it.
11838 2018-06-28 Richard Biener <rguenther@suse.de>
11840 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
11841 deciding whether to not re-use a DIE.
11843 2018-06-28 Richard Biener <rguenther@suse.de>
11845 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
11846 DW_AT_abstract_origin attribute.
11848 2018-06-28 Martin Liska <mliska@suse.cz>
11850 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
11851 Use newly introduced constants.
11852 * tree-switch-conversion.h (struct jump_table_cluster):
11853 Define max_ratio_for_size and max_ratio_for_speed.
11855 2018-06-28 Martin Liska <mliska@suse.cz>
11857 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
11858 Add new checking assert to catch invalid state.
11859 (jump_table_cluster::can_be_handled): Handle single case
11861 (jump_table_cluster::is_beneficial): Bail out for such case.
11862 (bit_test_cluster::find_bit_tests):
11863 Add new checking assert to catch invalid state.
11864 (bit_test_cluster::can_be_handled): Handle single case
11866 (bit_test_cluster::is_beneficial): Bail out for such case.
11867 (switch_decision_tree::analyze_switch_statement):
11870 2018-06-28 Martin Liska <mliska@suse.cz>
11872 * common.opt: Introduce -completion option.
11873 * gcc.c (driver_handle_option): Handle it.
11874 (driver::main): Print completions if completion
11876 * opt-suggestions.c (option_proposer::get_completions):
11878 (option_proposer::suggest_completion): Likewise.
11879 (option_proposer::find_param_completions): Likewise.
11880 (verify_autocompletions): Likewise.
11881 (test_completion_valid_options): Likewise.
11882 (test_completion_valid_params): Likewise.
11883 (in_completion_p): Likewise.
11884 (empty_completion_p): Likewise.
11885 (test_completion_partial_match): Likewise.
11886 (test_completion_garbage): Likewise.
11887 (opt_proposer_c_tests): Likewise.
11888 * opt-suggestions.h: Declare new functions.
11889 * opts.c (common_handle_option): Handle OPT__completion_.
11890 * selftest-run-tests.c (selftest::run_tests): Add
11891 opt_proposer_c_tests.
11892 * selftest.c (assert_str_startswith): New.
11893 * selftest.h (assert_str_startswith): Likewise.
11894 (opt_proposer_c_tests): New.
11895 (ASSERT_STR_STARTSWITH): Likewise.
11897 2018-06-28 Martin Liska <mliska@suse.cz>
11899 * Makefile.in: Add opt-suggestions.o.
11900 * gcc-main.c: Include opt-suggestions.h.
11901 * gcc.c (driver::driver): Likewise.
11902 (driver::~driver): Remove m_option_suggestions.
11903 (driver::build_option_suggestions): Moved to option_proposer.
11904 (driver::suggest_option): Likewise.
11905 (driver::handle_unrecognized_options): Use option_proposer.
11906 * gcc.h (class driver): Add new memver m_option_proposer.
11907 * opt-suggestions.c: New file.
11908 * opt-suggestions.h: New file.
11910 2018-06-28 Martin Liska <mliska@suse.cz>
11912 * vec.h (class auto_string_vec): New (moved from auto_argvec).
11913 (auto_string_vec::~auto_string_vec): Likewise.
11915 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
11917 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
11918 prevent_decl_creation_for_types fields up and add reset_location field.
11919 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
11920 statement if id->reset_location is true.
11921 (copy_edges_for_bb): Do not set goto_locus on the new edges if
11922 id->reset_location is true.
11923 (copy_phis_for_bb): Force input_location on the arguments if
11924 id->reset_location is true.
11925 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
11926 is set on the function to be inlined.
11928 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
11930 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
11932 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
11934 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
11935 registers for Pmode.
11936 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
11937 hard registers for the clobbered pseudo.
11939 2018-06-27 Paul Koning <ni1d@arrl.net>
11941 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
11942 mutually exclusive options.
11943 * config/pdp11/constraints.md (h): New constraint.
11944 (O): Update definition to match shift code generation.
11945 (D): New constraint.
11946 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
11948 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
11950 (output_jump): Change arguments.
11951 (pdp11_fixed_cc_regs): New function.
11952 (pdp11_cc_mode): Ditto.
11953 (pdp11_expand_shift): Ditto.
11954 (pdp11_assemble_shift): Ditto.
11955 (pdp11_small_shift): Ditto.
11956 (pdp11_branch_cost): Remove.
11957 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
11959 (pdp11_register_move_cost): Update for CC registers.
11960 (pdp11_rtx_costs): Add case for LSHIFTRT.
11961 (pdp11_output_jump): Add CCNZ mode conditional branches.
11962 (notice_update_cc_on_set): Remove.
11963 (pdp11_cc_mode): New function.
11964 (simple_memory_operand): Correct pre/post decrement case.
11965 (no_side_effect_operand): New function.
11966 (pdp11_regno_reg_class): Add CC_REGS class.
11967 (pdp11_fixed_cc_regs): New function.
11968 (pdp11_small_shift): New function.
11969 (pdp11_expand_shift): New function to expand shift insns.
11970 (pdp11_assemble_shift): New function to output shifts.
11971 (pdp11_branch_cost): Remove.
11972 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
11973 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
11974 (WCHAR_TYPE): Ditto.
11975 (PTRDIFF_TYPE): Ditto.
11976 (ADJUST_INSN_LENGTH): New macro.
11977 (FIXED_REGISTERS): Add CC registers.
11978 (CALL_USED_REGISTERS): Ditto.
11979 (reg_class): Ditto.
11980 (REG_CLASS_NAMES): Ditto.
11981 (REG_CLASS_CONTENTS): Ditto.
11982 (SELECT_CC_MODE): Use new function.
11983 (TARGET_FLAGS_REGNUM): New macro.
11984 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
11985 (cc0_reg_rtx): Remove.
11986 (CC_STATUS_MDEP): Remove.
11987 (CC_STATUS_MDEFP_INIT): Remove.
11988 (CC_IN_FPU): Remove.
11989 (NOTICE_UPDATE_CC): Remove.
11990 (REGISTER_NAMES): Add CC registers.
11991 (BRANCH_COST): Change to constant 1.
11992 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
11994 * config/pdp11/pdp11.opt (mbcopy): Remove.
11995 (mbcopy-builtin): Remove.
11996 (mbranch-cheap): Remove.
11997 (mbranch-expensive): Remove.
11998 * config/pdp11/predicates.md (expand_shift_operand): Update to
11999 match shift code generation.
12000 (ccnz_operator): New predicate.
12001 * doc/invoke.texi (PDP-11 Options): Remove deleted options
12002 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
12003 Remove non-existent option -mabshi, -mno-abshi. Document mutually
12005 * doc/md.texi (PDP-11): Document new D and h constraints. Update
12006 description of O constraint.
12008 2018-06-27 Jeff Law <law@redhat.com>
12009 Austin Law <austinklaw@gmail.com>
12011 * config/v850/v850.md (addsi3_set_flags): New pattern.
12012 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
12013 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
12014 (zero_extendhisi2_v850_set_flags): Likewise.
12015 (zero_extendqisi2_v850_set_flags): Likewise.
12016 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
12017 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
12018 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
12020 * config/v850/v850-protos.h (notice_update_cc): Remove.
12021 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
12022 (v850_print_operand): Handle 'D' and "d".
12023 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
12024 Add handling of arithmetic/logical operations compared against zero.
12025 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
12026 Do not look at v850_compare_op, instead get mode from last argument.
12027 (v850_gen_compare): Remove
12028 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
12029 after reload for prologue insns.
12030 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
12032 (construct_save_jarl): Likewise.
12033 (TARGET_FLAGS_REGNUM): Define.
12034 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
12035 (NOTICE_UPDATE_CC): Remove.
12036 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
12037 than cc0. Conditionalize on reload_completed.
12038 (cmpsi_insn, setfcc_insn): Likewise.
12039 (tst1 splitter): Turn into define_and_split which sets the flags
12041 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
12042 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
12043 (cstoresf4, cstoredf4): Clobber the flags.
12044 (cmpsi, cmpsf, cmpdf): Remove expanders.
12045 (setf_insn): Remove pattern.
12046 (addsi3): Turn into define_and_split which clobbers the flags after
12047 reload and a suitable pattern (addsi3_clobber_flags) for use after
12049 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
12050 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
12051 (ashrsi3, ashrsi3_v850e2): Likewise.
12052 (bins): Clobber the flags.
12053 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
12054 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
12055 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
12056 (call_value_internal_short, call_value_internal_long): Likewise.
12057 (callt_save_interrupt, callt_return_interrupt): Likewise.
12058 (save_interrupt, return_interrupt): Likewise.
12059 (callt_save_all_interrupt, save_all_interrupt): Likewise.
12060 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
12061 (restore_all_interrupt, _restore_all_interrupt): Likewise.
12062 (All FP comparisons): Only allow after reload has completed.
12064 (divh, divhu): Tweak output template.
12065 (branch_z_normal, branch_z_invert): Remove
12066 (branch_nz_normal, branch_nz_invert): Likewise.
12067 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
12069 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
12070 * config/v850/v850.c (notice_update_cc): Remove.
12071 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
12072 (CC_NO_CARRY): Likewise.
12073 (NOTICE_UPDATE_CC): Define to nothing.
12074 * config/v850/v850.md: Remove block comment on cc0 handling
12075 Remove "cc" attribute from all patterns. Remove cc_status handling
12076 from all patterns. Minor formatting fixes.
12078 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12080 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
12081 (cortex-a76.cortex-a55): Likewise.
12082 * config/aarch64/aarch64-tune.md: Regenerate.
12083 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
12084 cortex-a76.cortex-a55.
12086 2018-06-27 Jeff Law <law@redhat.com>
12088 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
12089 (MULTILIB_DIRNAMES): Similarly.
12091 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
12093 * gimple.h (gimple_return_retbnd): Delete.
12094 (gimple_return_set_retbnd): Likewise.
12095 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
12096 gimple_return_set_retbnd.
12097 * gimple-pretty-print.c (dump_gimple_return): Remove call to
12098 gimple_return_retbnd and adjust.
12099 * tree-inline.h (struct copy_body_data): Remove retbnd field.
12100 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
12101 Explicitly return NULL in a couple more cases. Move assertion
12102 on debug statements and remove unreachable code.
12103 (reset_debug_binding): Do not test id->retbnd.
12104 (expand_call_inline): Do not set it.
12106 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
12108 * configure.ac: Add --disable-gcov option.
12109 * configure: Regenerate.
12110 * Makefile.in: Honour @enable_gcov@.
12111 * doc/install.texi: Document --disable-gcov.
12113 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12115 * config/arm/arm-cpus.in (cortex-a76): New entry.
12116 (cortex-a76.cortex-a55): Likewise.
12117 * config/arm/arm-tables.opt: Regenerate.
12118 * config/arm/arm-tune.md: Likewise.
12119 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
12120 * doc/invoke.texi (ARM Options): Document cortex-a76 and
12121 cortex-a76.cortex-a55.
12123 2018-06-27 Tamar Christina <tamar.christina@arm.com>
12126 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
12128 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
12130 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
12132 (EPILOGUE_USES): Likewise.
12134 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
12136 * tree-inline.c (remap_location): New function extracted from...
12137 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
12138 (copy_phis_for_bb): ...here. Call remap_location.
12139 (copy_cfg_body): Adjust call to copy_edges_for_bb.
12141 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
12143 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
12144 unaligned vsx for 16B memset.
12146 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
12149 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
12150 ieee128_float_type_node to long_double_type_node unless
12151 TARGET_LONG_DOUBLE_128 is set.
12153 2018-06-26 David Malcolm <dmalcolm@redhat.com>
12155 * cfgloop.c (get_loop_location): Convert return type from
12156 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
12157 by implicit construction from rtx_insn *, and using
12158 dump_user_location_t::from_function_decl for the fallback case.
12159 * cfgloop.h (get_loop_location): Convert return type from
12160 location_t to dump_user_location_t.
12161 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
12162 dump_printf_loc to pass in a dump_location_t rather than a
12163 location_t, via the gimple stmt.
12164 * coverage.c (get_coverage_counts): Update calls to
12165 dump_printf_loc to pass in dump_location_t rather than a
12167 * doc/optinfo.texi (Dump types): Convert example of
12168 dump_printf_loc from taking "locus" to taking "insn". Update
12169 description of the "_loc" calls to cover dump_location_t.
12170 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
12172 (dump_user_location_t::dump_user_location_t): New constructors,
12173 from gimple *stmt and rtx_insn *.
12174 (dump_user_location_t::from_function_decl): New function.
12175 (dump_loc): Make static.
12176 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
12177 const dump_location_t &.
12178 (dump_generic_expr_loc): Delete.
12179 (dump_printf_loc): Convert param "loc" from location_t to
12180 const dump_location_t &.
12181 (selftest::test_impl_location): New function.
12182 (selftest::dumpfile_c_tests): New function.
12183 * dumpfile.h: Include "profile-count.h".
12184 (class dump_user_location_t): New class.
12185 (struct dump_impl_location_t): New struct.
12186 (class dump_location_t): New class.
12187 (dump_printf_loc): Convert 2nd param from source_location to
12188 const dump_location_t &.
12189 (dump_generic_expr_loc): Delete.
12190 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
12191 const dump_location_t &.
12192 * gimple-fold.c (fold_gimple_assign): Update call to
12193 dump_printf_loc to pass in a dump_location_t rather than a
12194 location_t, via the gimple stmt.
12195 (gimple_fold_call): Likewise.
12196 * gimple-loop-interchange.cc
12197 (loop_cand::analyze_iloop_reduction_var): Update for change to
12198 check_reduction_path.
12199 (tree_loop_interchange::interchange): Update for change to
12200 find_loop_location.
12201 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
12202 change in return-type of find_loop_location.
12203 (graphite_regenerate_ast_isl): Likewise.
12204 * graphite-optimize-isl.c (optimize_isl): Likewise.
12205 * graphite.c (graphite_transform_loops): Likewise.
12206 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
12207 pass in a dump_location_t rather than a location_t, via the
12209 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
12210 * ipa.c (walk_polymorphic_call_targets): Likewise.
12211 * loop-unroll.c (report_unroll): Convert "locus" param from
12212 location_t to dump_location_t.
12213 (decide_unrolling): Update for change to get_loop_location's
12215 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
12216 location_t to dump_user_location_t.
12217 (grid_find_single_omp_among_assignments_1): Updates calls to
12218 dump_printf_loc to pass in a dump_location_t rather than a
12219 location_t, via the gimple stmt.
12220 (grid_parallel_clauses_gridifiable): Convert "tloc" from
12221 location_t to dump_location_t. Updates calls to dump_printf_loc
12222 to pass in a dump_location_t rather than a location_t, via the
12224 (grid_inner_loop_gridifiable_p): Likewise.
12225 (grid_dist_follows_simple_pattern): Likewise.
12226 (grid_gfor_follows_tiling_pattern): Likewise.
12227 (grid_target_follows_gridifiable_pattern): Likewise.
12228 (grid_attempt_target_gridification): Convert initialization
12229 of local "grid" from memset to zero-initialization; FIXME: does
12230 this require C++11? Update call to dump_printf_loc to pass in a
12231 optinfo_location rather than a location_t, via the gimple stmt.
12232 * profile.c (read_profile_edge_counts): Updates call to
12233 dump_printf_loc to pass in a dump_location_t rather than a
12235 (compute_branch_probabilities): Likewise.
12236 * selftest-run-tests.c (selftest::run_tests): Call
12238 * selftest.h (dumpfile_c_tests): New decl.
12239 * tree-loop-distribution.c (pass_loop_distribution::execute):
12240 Update for change in return type of find_loop_location.
12241 * tree-parloops.c (parallelize_loops): Likewise.
12242 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
12243 "locus" from location_t to dump_user_location_t.
12244 (canonicalize_loop_induction_variables): Likewise.
12245 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
12246 for change in return type of find_loop_location.
12247 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
12248 to dump_printf_loc to pass in a dump_location_t rather than a
12249 location_t, via the stmt.
12250 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
12252 * tree-vect-loop-manip.c (find_loop_location): Convert return
12253 type from source_location to dump_user_location_t.
12254 (vect_do_peeling): Update for above change.
12255 (vect_loop_versioning): Update for change in type of
12257 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
12258 from location_t to dump_user_location_t.
12259 (vect_estimate_min_profitable_iters): Update for change in type
12261 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
12262 location_t to dump_location_t.
12263 (vect_slp_bb): Update for change in type of vect_location.
12264 * tree-vectorizer.c (vect_location): Convert from source_location
12265 to dump_user_location_t.
12266 (try_vectorize_loop_1): Update for change in vect_location's type.
12267 (vectorize_loops): Likewise.
12268 (increase_alignment): Likewise.
12269 * tree-vectorizer.h (vect_location): Convert from source_location
12270 to dump_user_location_t.
12271 (find_loop_location): Convert return type from source_location to
12272 dump_user_location_t.
12273 (check_reduction_path): Convert 1st param from location_t to
12274 dump_user_location_t.
12275 * value-prof.c (check_counter): Update call to dump_printf_loc to
12276 pass in a dump_user_location_t rather than a location_t; update
12277 call to error_at for change in type of "locus".
12278 (check_ic_target): Update call to dump_printf_loc to
12279 pass in a dump_user_location_t rather than a location_t, via the
12282 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
12284 * config/s390/s390.h (enum processor_flags): Do not use
12285 default tune parameter when -march was specified.
12287 2018-06-26 Jakub Jelinek <jakub@redhat.com>
12290 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
12291 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
12294 2018-06-26 Richard Biener <rguenther@suse.de>
12296 PR tree-optimization/86287
12298 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
12299 (vect_analyze_loop): Initialize n_stmts.
12301 2018-06-26 Richard Biener <rguenther@suse.de>
12303 PR middle-end/86271
12304 * fold-const.c (fold_convertible_p): Pointer extension
12307 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
12310 * dwarf2out.c (loc_list_has_views): Adjust comments.
12311 (dw_loc_list): Split single cross-partition range with
12314 2018-06-25 Jeff Law <law@redhat.com>
12316 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
12317 on -mbig-switch by default.
12319 * config/v850/predicates.md (const_float_1_operand): Fix match_code
12321 (const_float_0_operand): Remove unused predicate.
12322 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
12323 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
12324 (recipsf2): New expander. Original pattern now called
12326 (recipdf2, recipdf2_insn): Similarly.
12327 (rsqrtsf2, rsqrtsf2_insn): Similarly
12328 (rsqrtdf2, rsqrtdf2_insn): Similarly
12330 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
12332 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
12333 Simplify logic for FreeBSD (twice).
12335 2018-06-25 Martin Sebor <msebor@redhat.com>
12337 PR tree-optimization/86204
12338 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
12339 a strnlen result if it's less than the length of the string.
12341 2018-06-25 Martin Sebor <msebor@redhat.com>
12343 PR tree-optimization/85700
12344 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
12345 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
12346 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
12348 2018-06-25 Martin Sebor <msebor@redhat.com>
12350 * doc/extend.texi (Zero-length arrays): Update and clarify.
12352 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
12354 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
12355 added IEEE/IBM long double multilib support on PowerPC little
12356 endian Linux systems.
12357 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
12358 (MULTILIB_DEFAULTS): Likewise.
12359 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12361 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
12362 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
12363 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
12365 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
12367 PR middle-end/86311
12368 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
12369 (REORDER_45): Likewise.
12371 2018-06-25 Jeff Law <law@redhat.com>
12373 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
12374 dividend to 32 bits. Adjust length.
12375 (udivmodhi4): Cleanup output template. Fix length.
12377 2018-06-25 Carl Love <cel@us.ibm.com>
12379 * config/rs6000/vsx.md: Change word selector to prefered location.
12381 2018-06-25 Richard Biener <rguenther@suse.de>
12383 PR tree-optimization/86304
12384 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
12385 epilogue-if-converted loops as well.
12387 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
12389 * lto-section-out.c (lto_begin_section): Do not print section
12390 name for noaddr and unnumbered dumps.
12392 2018-06-25 Richard Biener <rguenther@suse.de>
12394 * tree-vectorizer.h (struct vec_info_shared): New structure
12395 with parts split out from struct vec_info and loop_nest from
12396 struct _loop_vec_info.
12397 (struct vec_info): Adjust accordingly.
12398 (struct _loop_vec_info): Likewise.
12399 (LOOP_VINFO_LOOP_NEST): Adjust.
12400 (LOOP_VINFO_DATAREFS): Likewise.
12401 (LOOP_VINFO_DDRS): Likewise.
12402 (struct _bb_vec_info): Likewise.
12403 (BB_VINFO_DATAREFS): Likewise.
12404 (BB_VINFO_DDRS): Likewise.
12405 (struct _stmt_vec_info): Add dr_aux member.
12406 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
12407 (DR_MISALIGNMENT_UNINITIALIZED): New.
12408 (set_dr_misalignment): Adjust.
12409 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
12410 (vect_analyze_loop): Adjust prototype.
12411 (vect_analyze_loop_form): Likewise.
12412 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
12413 Compute dependences lazily.
12414 (vect_record_base_alignments): Use shared datarefs/ddrs.
12415 (vect_verify_datarefs_alignment): Likewise.
12416 (vect_analyze_data_refs_alignment): Likewise.
12417 (vect_analyze_data_ref_accesses): Likewise.
12418 (vect_analyze_data_refs): Likewise.
12419 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
12420 constructor parameter for shared part.
12421 (vect_analyze_loop_form): Pass in shared part and adjust.
12422 (vect_analyze_loop_2): Pass in storage for the number of
12423 stmts. Move loop nest finding to the caller. Compute
12425 (vect_analyze_loop): Pass in shared part.
12426 (vect_transform_loop): Verify shared datarefs are unchanged.
12427 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
12428 constructor parameter for shared part.
12429 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
12430 (vect_slp_bb): Verify shared datarefs are unchanged before
12432 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
12434 (new_stmt_vec_info): Initialize DR_AUX misalignment to
12435 DR_MISALIGNMENT_UNINITIALIZED.
12436 * tree-vectorizer.c (vec_info::vec_info): Add constructor
12437 parameter for shared part.
12438 (vec_info::~vec_info): Adjust.
12439 (vec_info_shared::vec_info_shared): New.
12440 (vec_info_shared::~vec_info_shared): Likewise.
12441 (vec_info_shared::save_datarefs): Likewise.
12442 (vec_info_shared::check_datarefs): Likewise.
12443 (try_vectorize_loop_1): Construct shared part live for analyses
12444 of a single loop for multiple vector sizes.
12445 * tree-parloops.c (gather_scalar_reductions): Adjust.
12447 2018-06-25 Richard Biener <rguenther@suse.de>
12449 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
12450 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
12451 (vect_analyze_data_refs): Remove similar code from here and
12452 simplify accordingly.
12454 2018-06-25 Richard Biener <rguenther@suse.de>
12456 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
12457 for reverse storage order accesses rather than asserting
12458 they cannot happen here.
12460 2018-06-25 Tom de Vries <tdevries@suse.de>
12463 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
12464 Use data16 instead of .byte for insn prefix.
12466 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
12469 * parser.c (make_char_string_pack): Pass this literal chars
12470 through cpp_interpret_string.
12471 (cp_parser_userdef_numeric_literal): Check the result of
12472 make_char_string_pack.
12474 2018-06-24 Maya Rashish <coypu@sdf.org>
12476 * ginclude/stddef.h: Simplify conditions around avoiding
12477 re-definition of __size_t.
12479 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12481 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
12482 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
12484 2018-06-22 Maya Rashish <coypu@sdf.org>
12486 * doc/invoke.texi (mno-fancy-math-387): Update for changes
12487 made to OpenBSD and NetBSD through the years.
12489 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
12491 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12492 behavior of vec_pack (vector double, vector double) to match
12493 behavior of vec_float2 (vector double, vector double).
12495 2018-06-22 Olivier Hainque <hainque@adacore.com>
12497 * gimplify.c (gimplify_function_tree): Prevent creation
12498 of a trampoline for the address of the current function
12499 passed to entry/exit instrumentation hooks.
12501 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
12504 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
12507 2018-06-22 Martin Liska <mliska@suse.cz>
12509 PR tree-optimization/86263
12510 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12511 Bail out if is_enabled is false.
12512 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
12514 (jump_table_cluster::is_enabled): New function.
12516 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12518 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
12519 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
12520 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
12521 (lto_input_ts_binfo_tree_pointers): Likewise.
12522 * tree-streamer-out.c (streamer_write_tree_bitfields,
12523 write_ts_binfo_tree_pointers): Likewise.
12524 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
12526 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
12528 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
12530 2018-06-22 Martin Liska <mliska@suse.cz>
12532 * symbol-summary.h (get): Make it pure and inline move
12533 functionality from ::get function.
12534 (get): Remove and inline into ::get and ::get_create.
12535 (get_create): Move code from ::get function.
12537 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12540 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
12541 -x assembler-with-cpp.
12543 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12545 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
12546 _FILE_OFFSET_BITS=64 for C++.
12548 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
12550 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
12551 conversion insn that shows up when pr85657-3.c is compiled using
12552 IEEE 128-bit long double.
12553 (neg<mode>2_internal): Use the correct mode to check whether the
12554 mode is IBM extended.
12555 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
12556 multiply and divide external functions from being created more
12559 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12561 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
12563 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
12564 the edge can be forwarded.
12565 (cfg_layout_merge_blocks): Likewise.
12567 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12569 * except.c (finish_eh_generation): Commit edge insertions only after
12570 the EH edges have been redirected from post-landing to landing pads.
12572 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12574 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
12575 create_tmp_var_for to create the FRAME decl.
12576 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
12578 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
12580 * tree-inline.c (copy_edges_for_bb): Minor tweak.
12581 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
12582 debug statement when resetting its value.
12583 (expand_call_inline): Copy the locus of the call onto the assignment
12584 of the return value, if any. Use local variable in more cases.
12586 2018-06-21 Martin Liska <mliska@suse.cz>
12588 * ipa-pure-const.c (propagate_nothrow): Use
12589 funct_state_summaries->get.
12590 (dump_malloc_lattice): Likewise.
12591 (propagate_malloc): Likewise.
12593 2018-06-21 Richard Biener <rguenther@suse.de>
12595 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
12596 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
12597 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
12599 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
12600 BLOCK_ABSTRACT_ORIGIN unconditionally.
12602 2018-06-21 David Malcolm <dmalcolm@redhat.com>
12604 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
12606 * ipa-reference.c (ipa_reference_c_finalize): Delete
12607 ipa_ref_opt_sum_summaries and set it to NULL.
12609 2018-06-21 Tom de Vries <tdevries@suse.de>
12611 PR tree-optimization/85859
12612 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
12613 test with comment from bb_no_side_effects_p.
12615 2018-06-21 Richard Biener <rguenther@suse.de>
12617 PR tree-optimization/86232
12618 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
12619 max for constant niter.
12621 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12623 * config/aarch64/aarch64-simd.md
12624 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
12626 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
12628 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
12629 Make opernads of the unspec commutative.
12631 2018-06-21 Richard Biener <rguenther@suse.de>
12633 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
12634 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12635 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
12636 (vect_analyze_data_ref_dependence): Re-order checks to deal with
12638 (vect_record_base_alignments): Do not record base alignment
12639 for gathers or scatters.
12640 (vect_compute_data_ref_alignment): Drop return value that is always
12641 true. Bail out early for gathers or scatters.
12642 (vect_enhance_data_refs_alignment): Bail out early for gathers
12644 (vect_find_same_alignment_drs): Likewise.
12645 (vect_analyze_data_refs_alignment): Remove dead code.
12646 (vect_slp_analyze_and_verify_node_alignment): Likewise.
12647 (vect_analyze_data_refs): For possible gathers or scatters do
12648 not create an alternate DR, just check their possible validity
12649 and mark them. Adjust DECL_NONALIASED handling to not rely
12650 on DR_BASE_ADDRESS.
12651 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
12652 update inits of gathers or scatters.
12653 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
12654 Also copy gather/scatter flag to pattern vinfo.
12656 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
12658 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
12659 behavior of vec_packsu (vector unsigned long long, vector unsigned
12660 long long) to match behavior of vec_packs with same signature.
12662 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
12663 Thomas Schwinge <thomas@codesourcery.com>
12664 Cesar Philippidis <cesar@codesourcery.com>
12666 * gimplify.c (gimplify_scan_omp_clauses): Add support for
12667 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12668 (gimplify_adjust_omp_clauses): Likewise.
12669 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
12670 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
12671 (gimplify_omp_target_update): Update handling of acc update and
12673 * omp-low.c (install_var_field): Remove unused parameter
12674 base_pointers_restrict.
12675 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
12676 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
12678 (omp_target_base_pointers_restrict_p): Delete.
12679 (scan_omp_target): Update call to scan_sharing_clauses.
12680 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
12682 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
12683 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
12684 (convert_local_omp_clauses): Likewise.
12685 * tree-pretty-print.c (dump_omp_clause): Likewise.
12686 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
12688 (omp_clause_code_name): Likewise.
12690 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12693 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
12696 PR tree-optimization/86231
12697 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
12698 anti-range don't overwrite *vr0min before using it to compute *vr0max.
12700 2018-06-20 Tom de Vries <tdevries@suse.de>
12702 PR tree-optimization/86097
12703 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
12704 iv type if signedness of iv type is not the same as that of *nit.
12706 2018-06-20 Jakub Jelinek <jakub@redhat.com>
12708 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
12709 EDGE_EH edges, verify they are all EDGE_EH.
12711 2018-06-20 Maya Rashish <coypu@sdf.org>
12713 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
12715 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12717 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
12718 * config/aarch64/aarch64.c (xgene1_tunings): Add
12719 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
12720 (aarch64_mode_valid_for_sched_fusion_p):
12721 Allow 16-byte modes.
12722 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
12723 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
12725 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
12727 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
12728 * config/aarch64/iterators.md (VQ2): New mode iterator.
12730 2018-06-20 Martin Liska <mliska@suse.cz>
12732 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
12733 Change default ratio from 10 to 8.
12735 2018-06-20 Martin Liska <mliska@suse.cz>
12737 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
12739 (bit_test_cluster::find_bit_tests): Likewise.
12740 (switch_decision_tree::analyze_switch_statement): Find clusters.
12741 * tree-switch-conversion.h (struct jump_table_cluster): Document
12744 2018-06-20 Martin Liska <mliska@suse.cz>
12746 * tree-switch-conversion.c (switch_conversion::collect):
12747 Record m_uniq property.
12748 (switch_conversion::expand): Bail out for special conditions.
12749 (group_cluster::~group_cluster): New.
12750 (group_cluster::group_cluster): Likewise.
12751 (group_cluster::dump): Likewise.
12752 (jump_table_cluster::emit): New.
12753 (switch_decision_tree::fix_phi_operands_for_edges): New.
12754 (struct case_node): Remove struct.
12755 (jump_table_cluster::can_be_handled): New.
12756 (case_values_threshold): Moved to header.
12757 (reset_out_edges_aux): Likewise.
12758 (jump_table_cluster::is_beneficial): New.
12759 (bit_test_cluster::can_be_handled): Likewise.
12760 (add_case_node): Remove.
12761 (bit_test_cluster::is_beneficial): New.
12762 (case_bit_test::cmp): New.
12763 (bit_test_cluster::emit): New.
12764 (expand_switch_as_decision_tree_p): Remove.
12765 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
12766 (fix_phi_operands_for_edge): Likewise.
12767 (switch_decision_tree::analyze_switch_statement): New.
12768 (compute_cases_per_edge): Move ...
12769 (switch_decision_tree::compute_cases_per_edge): ... here.
12770 (try_switch_expansion): Likewise.
12771 (switch_decision_tree::try_switch_expansion): Likewise.
12772 (record_phi_operand_mapping): Likewise.
12773 (switch_decision_tree::record_phi_operand_mapping): Likewise.
12774 (emit_case_decision_tree): Likewise.
12775 (switch_decision_tree::emit): Likewise.
12776 (balance_case_nodes): Likewise.
12777 (switch_decision_tree::balance_case_nodes): Likewise.
12778 (dump_case_nodes): Likewise.
12779 (switch_decision_tree::dump_case_nodes): Likewise.
12780 (emit_jump): Likewise.
12781 (switch_decision_tree::emit_jump): Likewise.
12782 (emit_cmp_and_jump_insns): Likewise.
12783 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
12784 (emit_case_nodes): Likewise.
12785 (switch_decision_tree::emit_case_nodes): Likewise.
12786 (conditional_probability): Remove.
12787 * tree-switch-conversion.h (enum cluster_type): New.
12789 (struct cluster): Likewise.
12790 (cluster::cluster): Likewise.
12791 (struct simple_cluster): Likewise.
12792 (simple_cluster::simple_cluster): Likewise.
12793 (struct group_cluster): Likewise.
12794 (struct jump_table_cluster): Likewise.
12795 (struct bit_test_cluster): Likewise.
12796 (struct min_cluster_item): Likewise.
12797 (struct case_tree_node): Likewise.
12798 (case_tree_node::case_tree_node): Likewise.
12799 (jump_table_cluster::case_values_threshold): Likewise.
12800 (struct case_bit_test): Likewise.
12801 (struct switch_decision_tree): Likewise.
12802 (struct switch_conversion): Likewise.
12803 (switch_decision_tree::reset_out_edges_aux): Likewise.
12805 2018-06-20 Martin Liska <mliska@suse.cz>
12807 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
12808 (hoist_edge_and_branch_if_true): Likewise.
12809 (expand_switch_using_bit_tests_p): Likewise.
12810 (struct case_bit_test): Likewise.
12811 (case_bit_test_cmp): Likewise.
12812 (emit_case_bit_tests): Likewise.
12813 (switch_conversion::switch_conversion): New class.
12814 (struct switch_conv_info): Remove old struct.
12815 (collect_switch_conv_info): More to ...
12816 (switch_conversion::collect): ... this.
12817 (check_range): Likewise.
12818 (switch_conversion::check_range): Likewise.
12819 (check_all_empty_except_final): Likewise.
12820 (switch_conversion::check_all_empty_except_final): Likewise.
12821 (check_final_bb): Likewise.
12822 (switch_conversion::check_final_bb): Likewise.
12823 (create_temp_arrays): Likewise.
12824 (switch_conversion::create_temp_arrays): Likewise.
12825 (free_temp_arrays): Likewise.
12826 (gather_default_values): Likewise.
12827 (switch_conversion::gather_default_values): Likewise.
12828 (build_constructors): Likewise.
12829 (switch_conversion::build_constructors): Likewise.
12830 (constructor_contains_same_values_p): Likewise.
12831 (switch_conversion::contains_same_values_p): Likewise.
12832 (array_value_type): Likewise.
12833 (switch_conversion::array_value_type): Likewise.
12834 (build_one_array): Likewise.
12835 (switch_conversion::build_one_array): Likewise.
12836 (build_arrays): Likewise.
12837 (switch_conversion::build_arrays): Likewise.
12838 (gen_def_assigns): Likewise.
12839 (switch_conversion::gen_def_assigns): Likewise.
12840 (prune_bbs): Likewise.
12841 (switch_conversion::prune_bbs): Likewise.
12842 (fix_phi_nodes): Likewise.
12843 (switch_conversion::fix_phi_nodes): Likewise.
12844 (gen_inbound_check): Likewise.
12845 (switch_conversion::gen_inbound_check): Likewise.
12846 (process_switch): Use the newly created class.
12847 (switch_conversion::expand): New.
12848 (switch_conversion::~switch_conversion): New.
12849 * tree-switch-conversion.h: New file.
12851 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12853 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
12854 tree-vect-patterns.c.
12855 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
12856 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
12857 (vect_recog_sad_pattern): Likewise.
12858 (vect_recog_widen_sum_pattern): Likewise.
12859 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
12860 (vect_recog_widen_shift_pattern): Remove the type_in argument.
12861 (vect_recog_rotate_pattern): Likewise.
12862 (vect_recog_mult_pattern): Likewise.
12863 (vect_recog_vector_vector_shift_pattern): Likewise.
12864 (vect_recog_divmod_pattern): Likewise.
12865 (vect_recog_mixed_size_cond_pattern): Likewise.
12866 (vect_recog_bool_pattern): Likewise.
12867 (vect_recog_mask_conversion_pattern): Likewise.
12868 (vect_try_gather_scatter_pattern): Likewise.
12869 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
12870 (vect_recog_over_widening_pattern): Likewise.
12871 (vect_recog_gather_scatter_pattern): Likewise.
12872 (vect_recog_func_ptr): Move from tree-vectorizer.h
12873 (vect_vect_recog_func_ptrs): Move further down the file.
12874 (vect_recog_func): Likewise. Remove the third argument.
12875 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
12876 (vect_pattern_recog_1): Expect the pattern function to do any
12877 necessary target tests. Also expect it to provide a vector type.
12878 Remove the type_in handling.
12880 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12882 * tree-vect-patterns.c (vect_pattern_detected): New function.
12883 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
12884 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
12885 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
12886 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
12887 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
12888 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
12889 (vect_recog_mask_conversion_pattern)
12890 (vect_try_gather_scatter_pattern): Likewise.
12892 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12894 * tree-vect-patterns.c (vect_get_internal_def): New function.
12895 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
12896 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
12897 (search_type_for_mask_1): Use it.
12899 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12901 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
12902 redundant WIDEN_SUM_EXPR handling.
12903 (vect_recog_sad_pattern): Likewise.
12905 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12907 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
12908 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
12909 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
12910 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
12911 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
12913 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12915 * tree-vect-stmts.c (vectorizable_call): Make sure that we
12916 use the stmt_vec_info of the original bb statement for the
12917 new zero assignment, even if the call is part of a pattern.
12919 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12921 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
12922 that the sequence is attached to the original statement rather
12923 than the pattern statement.
12924 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
12925 PATTERN_DEF_SEQ from the original statement rather than
12926 the main pattern statement.
12927 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
12928 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
12929 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
12931 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
12933 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
12934 definition statements before the early exit for statements that aren't
12936 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
12938 (vect_transform_loop): ...here. Process pattern definition
12939 statements without first checking whether the main pattern
12940 statement is live or relevant.
12942 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
12944 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
12945 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
12947 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
12949 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
12950 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
12951 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
12952 (expand_block_compare): Change select_block_compare_mode call.
12953 (expand_strncmp_align_check): Use new functions, fix comment.
12954 (emit_final_str_compare_gpr): New function.
12955 (expand_strn_compare): Refactor and clean up code.
12956 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
12958 2018-06-19 Tony Reix <tony.reix@atos.com>
12959 Damien Bergamini <damien.bergamini@atos.com>
12960 David Edelsohn <dje.gcc@gmail.com>
12962 * collect2.c (static_obj): New variable.
12963 (static_libs): New variable.
12964 (is_in_list): Uncomment declaration.
12965 (main): Track AIX libraries linked statically.
12966 (is_in_list): Uncomment definition.
12967 (scan_prog_file): Don't add AIX shared libraries initializer
12968 to constructor list if linking statically.
12970 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
12972 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
12974 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
12976 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
12978 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
12981 2018-06-19 Martin Liska <mliska@suse.cz>
12983 * config/i386/i386.c (ix86_can_inline_p): Do not use
12984 ipa_fn_summaries::get_create.
12985 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
12987 (devirtualization_time_bonus): Likewise.
12988 (ipcp_propagate_stage): Likewise.
12989 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
12990 (edge_set_predicate): Likewise.
12991 (evaluate_conditions_for_known_args): Likewise.
12992 (evaluate_properties_for_edge): Likewise.
12993 (ipa_call_summary::reset): Tranform to ...
12994 (ipa_call_summary::~ipa_call_summary): ... this.
12995 (ipa_fn_summary::reset): Transform to ...
12996 (ipa_fn_summary::~ipa_fn_summary): ... this.
12997 (ipa_fn_summary_t::remove): Rename to ...
12998 (ipa_fn_summary_t::remove_callees): ... this.
12999 (ipa_fn_summary_t::duplicate): Use placement new
13000 instead of memory copy.
13001 (ipa_call_summary_t::duplicate): Likewise.
13002 (ipa_call_summary_t::remove): Remove.
13003 (dump_ipa_call_summary): Change get_create to get.
13004 (ipa_dump_fn_summary): Dump only when summary exists.
13005 (analyze_function_body): Use symbol_summary::get instead
13007 (compute_fn_summary): Likewise.
13008 (estimate_edge_devirt_benefit): Likewise.
13009 (estimate_edge_size_and_time): Likewise.
13010 (inline_update_callee_summaries): Likewise.
13011 (remap_edge_change_prob): Likewise.
13012 (remap_edge_summaries): Likewise.
13013 (ipa_merge_fn_summary_after_inlining): Likewise.
13014 (write_ipa_call_summary): Likewise.
13015 (ipa_fn_summary_write): Likewise.
13016 (ipa_free_fn_summary): Likewise.
13017 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
13018 (struct ipa_call_summary): Likewise.
13019 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
13021 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
13022 (estimate_size_after_inlining): Likewise.
13023 (estimate_growth): Likewise.
13024 (growth_likely_positive): Likewise.
13025 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
13026 (inline_call): Likewise.
13027 * ipa-inline.c (caller_growth_limits): Likewise.
13028 (can_inline_edge_p): Likewise.
13029 (can_inline_edge_by_limits_p): Likewise.
13030 (compute_uninlined_call_time): Likewise.
13031 (compute_inlined_call_time): Likewise.
13032 (want_inline_small_function_p): Likewise.
13033 (edge_badness): Likewise.
13034 (update_caller_keys): Likewise.
13035 (update_callee_keys): Likewise.
13036 (inline_small_functions): Likewise.
13037 (inline_to_all_callers_1): Likewise.
13038 (dump_overall_stats): Likewise.
13039 (early_inline_small_functions): Likewise.
13040 (early_inliner): Likewise.
13041 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
13042 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13043 * ipa-pure-const.c (malloc_candidate_p): Likewise.
13044 * ipa-split.c (execute_split_functions): Likewise.
13045 * symbol-summary.h: Likewise.
13046 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
13048 2018-06-19 Richard Biener <rguenther@suse.de>
13050 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
13051 (vectorize_loops): ... here. Fix dbgcnt handling.
13052 (try_vectorize_loop): Wrap try_vectorize_loop_1.
13054 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
13057 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
13058 ieee128 argument takes up only one (vector) register, not two (floating
13061 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
13063 * gimplify.c (gimplify_init_constructor): Really never clear for an
13064 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
13066 2018-06-19 Richard Biener <rguenther@suse.de>
13068 PR tree-optimization/86179
13069 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
13070 after failed recognition.
13072 2018-06-18 Martin Sebor <msebor@redhat.com>
13074 PR middle-end/85602
13075 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
13076 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
13077 Handle integer subtraction.
13078 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
13079 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
13081 2018-06-18 David Malcolm <dmalcolm@redhat.com>
13083 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
13084 param from rtx to rtx_insn *.
13085 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
13087 (frv_ifcvt_modify_insn): Likwise.
13088 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
13089 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
13090 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
13091 as_a <rtx_insn *> cast to local "unprotected_region" once
13092 it's been established that it's not NULL or pc_rtx.
13093 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
13094 param "sethi" from rtx to rtx_insn *.
13095 (nds32_group_float_insns): Likewise for param "insn".
13096 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
13098 (vax_output_int_subtract): Likewise.
13099 * config/vax/vax.c (vax_output_int_add): Likewise for param
13101 (vax_output_int_subtract): Likewise.
13102 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
13103 (emit_pattern_after): Likewise for param "after".
13104 (emit_insn_after): Likewise.
13105 (emit_jump_insn_after): Likewise.
13106 (emit_call_insn_after): Likewise.
13107 (emit_debug_insn_after): Likewise.
13108 (emit_pattern_before): Likewise for param "before".
13109 (emit_insn_before): Likewise.
13110 (emit_jump_insn_before): Likewise.
13111 * final.c (get_insn_template): Likewise for param "insn", removing
13113 * output.h (get_insn_template): Likewise for 2nd param.
13114 * rtl.h (emit_insn_before): Likewise.
13115 (emit_jump_insn_before): Likewise.
13116 (emit_debug_insn_before_noloc): Likewise.
13117 (emit_insn_after): Likewise.
13118 (emit_jump_insn_after): Likewise.
13119 (emit_call_insn_after): Likewise.
13120 (emit_debug_insn_after): Likewise.
13121 (set_insn_deleted): Likewise for param.
13123 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
13126 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
13127 floating point modes, so that IFmode is numerically greater than
13128 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
13129 to declare the ordering. This prevents IFmode from being
13130 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
13131 machine. Include rs6000-modes.h to share the fractional values
13132 between genmodes* and the rest of the compiler.
13133 (IFmode): Likewise.
13134 (KFmode): Likewise.
13135 (TFmode): Likewise.
13136 * config/rs6000/rs6000-modes.h: New file.
13137 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
13138 meaning of rs6000_long_double_size so that 126..128 selects an
13139 appropriate 128-bit floating point type.
13140 (rs6000_option_override_internal): Likewise.
13141 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
13142 (TARGET_LONG_DOUBLE_128): Change the meaning of
13143 rs6000_long_double_size so that 126..128 selects an appropriate
13144 128-bit floating point type.
13145 (LONG_DOUBLE_TYPE_SIZE): Update comment.
13146 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
13147 source and destination to match the standard usage.
13148 (truncifkf2): Likewise.
13149 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
13150 ISA 2.07 to use an explicit clobber, instead of passing in a
13152 (copysign<mode>3_soft): Likewise.
13154 2018-06-18 David Malcolm <dmalcolm@redhat.com>
13156 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
13157 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
13158 (vect_slp_analyze_instance_dependence): Likewise.
13159 (vect_enhance_data_refs_alignment): Likewise.
13160 (vect_analyze_data_refs_alignment): Likewise.
13161 (vect_slp_analyze_and_verify_instance_alignment
13162 (vect_analyze_data_ref_accesses): Likewise.
13163 (vect_prune_runtime_alias_test_list): Likewise.
13164 (vect_analyze_data_refs): Likewise.
13165 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
13166 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
13167 (vect_analyze_scalar_cycles_1): Likewise.
13168 (vect_get_loop_niters): Likewise.
13169 (vect_analyze_loop_form_1): Likewise.
13170 (vect_update_vf_for_slp): Likewise.
13171 (vect_analyze_loop_operations): Likewise.
13172 (vect_analyze_loop): Likewise.
13173 (vectorizable_induction): Likewise.
13174 (vect_transform_loop): Likewise.
13175 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
13176 * tree-vect-slp.c (vect_analyze_slp): Likewise.
13177 (vect_make_slp_decision): Likewise.
13178 (vect_detect_hybrid_slp): Likewise.
13179 (vect_slp_analyze_operations): Likewise.
13180 (vect_slp_bb): Likewise.
13181 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
13182 (vectorizable_bswap): Likewise.
13183 (vectorizable_call): Likewise.
13184 (vectorizable_simd_clone_call): Likewise.
13185 (vectorizable_conversion): Likewise.
13186 (vectorizable_assignment): Likewise.
13187 (vectorizable_shift): Likewise.
13188 (vectorizable_operation): Likewise.
13189 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
13191 2018-06-18 Martin Sebor <msebor@redhat.com>
13193 PR tree-optimization/81384
13194 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
13195 * builtins.c (expand_builtin_strnlen): New function.
13196 (expand_builtin): Call it.
13197 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
13198 * builtins.def (BUILT_IN_STRNLEN): New.
13199 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
13200 Warn for bounds in excess of maximum object size.
13201 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
13202 single-value ranges. Handle strnlen.
13203 (handle_builtin_strlen): Handle strnlen.
13204 (strlen_check_and_optimize_stmt): Same.
13205 * doc/extend.texi (Other Builtins): Document strnlen.
13207 2018-06-18 Maya Rashish <coypu@sdf.org>
13209 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
13210 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
13211 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
13213 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
13215 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
13217 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13219 * tree.c (escaped_string::escape): Replace cast to char * by
13220 const_cast<char *> (unescaped).
13222 2018-06-18 Nick Clifton <nickc@redhat.com>
13225 * tree.c (escaped_string): New class. Converts an unescaped
13226 string into its escaped equivalent.
13227 (warn_deprecated_use): Use the new class to convert the
13228 deprecation message, if present.
13229 (test_escaped_strings): New self test.
13230 (test_c_tests): Add test_escaped_strings.
13231 * doc/extend.texi (deprecated): Add a note that the
13232 deprecation message is affected by the -fmessage-length
13233 option, and that control characters will be escaped.
13234 (#pragma GCC error): Document this pragma.
13235 (#pragma GCC warning): Likewise.
13236 * doc/invoke.texi (-fmessage-length): Document this option's
13237 effect on the #warning and #error preprocessor directives and
13238 the deprecated attribute.
13240 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
13242 * tree.c (decl_value_expr_lookup): Revert latest change.
13243 (decl_value_expr_insert): Likewise.
13245 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
13247 * gimplify.c (nonlocal_vlas): Delete.
13248 (nonlocal_vla_vars): Likewise.
13249 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
13251 (gimplify_body): Do not create and destroy nonlocal_vlas.
13252 * tree-nested.c: Include diagnostic.h.
13253 (use_pointer_in_frame): Tweak.
13254 (lookup_field_for_decl): Add assertion and declare the transformation.
13255 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
13256 internal error when the reference is in a wrong context. Do not
13257 create a debug decl by default.
13258 (note_nonlocal_block_vlas): Delete.
13259 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
13260 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
13261 create a debug decl by default.
13262 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
13263 call is in a wrong context.
13264 (fixup_vla_decls): New function.
13265 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
13266 debug variables were created.
13267 * tree.c (decl_value_expr_lookup): Add checking assertion.
13268 (decl_value_expr_insert): Likewise.
13270 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
13272 PR middle-end/82479
13273 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
13274 * tree-scalar-evolution.c (interpret_expr): Likewise.
13275 (expression_expensive_p): Likewise.
13276 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
13277 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
13278 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
13279 (ssa_defined_by_minus_one_stmt_p): New.
13281 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
13283 PR middle-end/64946
13284 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
13285 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
13286 * dojump.c (do_jump): Likewise.
13287 * expr.c (expand_expr_real_2): Check operand type's sign.
13288 * fold-const.c (const_unop): Handle ABSU_EXPR.
13289 (fold_abs_const): Likewise.
13290 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
13291 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
13292 (strip_sign_op_1): Likesise.
13293 * match.pd: Add new pattern to generate ABSU_EXPR.
13294 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
13295 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
13296 * tree-eh.c (operation_could_trap_helper_p): Likewise.
13297 * tree-inline.c (estimate_operator_cost): Likewise.
13298 * tree-pretty-print.c (dump_generic_node): Likewise.
13299 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
13300 * tree.def (ABSU_EXPR): New.
13302 2018-06-16 Jakub Jelinek <jakub@redhat.com>
13304 PR middle-end/86095
13305 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
13306 documented as preserved for backward compatibility only.
13307 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
13309 PR rtl-optimization/86108
13310 * bb-reorder.c (create_forwarder_block): Renamed to ...
13311 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
13312 jump from new landing pad to the second part.
13313 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
13316 2018-06-15 Jakub Jelinek <jakub@redhat.com>
13318 PR middle-end/85878
13319 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
13320 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
13321 Only call store_expr for halves if the mode is the same.
13323 PR middle-end/86123
13324 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
13325 Fix up comment formatting.
13327 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
13329 * typed-splay-tree.h (typed_splay_tree::remove): New function.
13330 (typed_splay_tree::closure,
13331 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
13332 (typed_splay_tree::typed_splay_tree,
13333 typed_splay_tree::operator =): Declared private.
13334 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
13335 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
13336 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
13337 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
13338 typed_splay_tree::splay_tree_splay,
13339 typed_splay_tree::splay_tree_foreach_helper,
13340 typed_splay_tree::splay_tree_insert,
13341 typed_splay_tree::splay_tree_remove,
13342 typed_splay_tree::splay_tree_lookup,
13343 typed_splay_tree::splay_tree_predecessor,
13344 typed_splay_tree::splay_tree_successor,
13345 typed_splay_tree::splay_tree_min,
13346 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
13347 (typed_splay_tree::root, typed_splay_tree::comp,
13348 typed_splay_tree::delete_key,
13349 typed_splay_tree::delete_value): New data members.
13350 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
13351 typed_splay_tree::remove.
13353 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
13355 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
13356 -mginv and -mno-ginv to the assembler.
13357 * config/mips/mips.opt (-mcrc): New option.
13358 (-mginv): Likewise.
13359 * doc/invoke.text (-mcrc): Document.
13360 (-mginv): Likewise.
13362 2018-06-15 Nick Clifton <nickc@redhat.com>
13365 * tree.c (escaped_string): New class. Converts an unescaped
13366 string into its escaped equivalent.
13367 (warn_deprecated_use): Use the new class to convert the
13368 deprecation message, if present.
13369 (test_escaped_strings): New self test.
13370 (test_c_tests): Add test_escaped_strings.
13371 * doc/extend.texi (deprecated): Add a note that the
13372 deprecation message is affected by the -fmessage-length
13373 option, and that control characters will be escaped.
13374 (#pragma GCC error): Document this pragma.
13375 (#pragma GCC warning): Likewise.
13376 * doc/invoke.texi (-fmessage-length): Document this option's
13377 effect on the #warning and #error preprocessor directives and
13378 the deprecated attribute.
13380 2018-06-15 Richard Biener <rguenther@suse.de>
13382 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
13383 here, also noting vector size used.
13384 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
13385 size used in MSG_OPTIMIZED_LOCATIONS dump.
13386 (pass_slp_vectorize::execute): Adjust.
13388 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
13391 * config/arc/arc.c (arc_return_address_register): Fix
13394 2018-06-15 Richard Biener <rguenther@suse.de>
13396 PR middle-end/86159
13397 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
13398 leave useless conversion stripping to force_gimple_operand_gsi.
13399 (gimplify_build2): Likewise.
13400 (gimplify_build1): Likewise.
13402 2018-06-15 Richard Biener <rguenther@suse.de>
13404 PR middle-end/86076
13405 * tree-cfg.c (move_stmt_op): unshare invariant addresses
13406 before adjusting their block.
13408 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
13410 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
13411 multilibs for *-*-rtems*.
13412 * config/riscv/t-rtems: New file.
13414 2018-06-14 Jakub Jelinek <jakub@redhat.com>
13416 PR middle-end/86122
13417 * match.pd ((A +- CST1) +- CST2): Punt if last resort
13418 unsigned_type_for returns NULL.
13421 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
13422 subregs of multi-word pseudos unless the float mode has word size.
13424 2018-06-14 Richard Biener <rguenther@suse.de>
13426 PR middle-end/86139
13427 * tree-vect-generic.c (build_word_mode_vector_type): Remove
13428 duplicate and harmful type_hash_canon.
13429 * tree.c (type_hash_canon): Assert we didn't find ourselves.
13431 2018-06-14 Richard Biener <rguenther@suse.de>
13434 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
13437 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
13439 * config/rtems.h (STDINT_LONG32): Define.
13441 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
13442 Prachi Godbole <prachi.godbole@imgtec.com>
13444 * config/mips/mips-cpus.def: Define P6600.
13445 * config/mips/mips-tables.opt: Regenerate.
13446 * config/mips/mips.c (mips_ucbranch_type): New enum.
13447 (mips_rtx_cost_data): Add support for P6600.
13448 (mips_issue_rate): Likewise.
13449 (mips_multipass_dfa_lookahead): Likewise.
13450 (mips_avoid_hazard): Likewise.
13451 (mips_reorg_process_insns): Likewise.
13452 (mips_classify_branch_p6600): New function.
13453 * config/mips/mips.h (TUNE_P6600): New define.
13454 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
13455 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
13456 * config/mips/mips.md: Include p6600.md.
13457 (processor): Add p6600.
13458 * config/mips/p6600.md: New file.
13459 * doc/invoke.texi: Add p6600 to supported architectures.
13461 2018-06-13 Martin Sebor <msebor@redhat.com>
13463 PR tree-optimization/86114
13464 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
13466 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
13468 2018-06-13 Richard Biener <rguenther@suse.de>
13470 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
13471 Properly set vector type of the intermediate stmt.
13472 * tree-vect-stmts.c (vectorizable_operation): The destination
13473 var always has vectype_out type.
13475 2018-06-13 Jeff Law <law@redhat.com>
13477 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
13478 integer 0 for argument to print_rtl_with_bb.
13479 (rl78_reorg): Likewise.
13481 2018-06-13 David Malcolm <dmalcolm@redhat.com>
13483 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
13484 from rtx to rtx_insn *.
13485 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
13487 (add_sched_insns_for_speculation): Likewise for local "target",
13488 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13489 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
13490 from rtx_insn ** to rtx_code_label **.
13491 (reorg_emit_nops): Likewise.
13492 (c6x_reorg): Likewise for local "call_labels".
13493 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
13495 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
13496 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
13497 the loops over LABEL_REFS.
13498 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
13500 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
13501 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
13502 (split_branches): Strengthen local "olabel" from rtx to
13503 rtx_insn *, adding a safe_as_a cast.
13504 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
13506 (add_insn_after): Likewise for first two params.
13507 (add_insn_before): Likewise.
13508 (remove_insn): Likewise for param.
13509 (emit_pattern_before_noloc): Likewise for second and third params.
13510 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
13511 (emit_call_insn_before_noloc): Likewise.
13512 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
13514 (emit_barrier_before): Likewise.
13515 (emit_label_before): Strengthen "label" param from "rtx" to
13516 "rtx_code_label *". Strengthen "before" param from "rtx" to
13518 (emit_insn_after_1): Strengthen "after" param from "rtx" to
13520 (emit_pattern_after_noloc): Likewise.
13521 (emit_insn_after_noloc): Likewise.
13522 (emit_jump_insn_after_noloc): Likewise.
13523 (emit_call_insn_after_noloc): Likewise.
13524 (emit_debug_insn_after_noloc): Likewise.
13525 (emit_barrier_after): Likewise.
13526 (emit_label_after): Likewise for both params.
13527 (emit_pattern_after_setloc): Likewise for "after" param. Convert
13528 "loc" param from "int" to "location_t".
13529 (emit_insn_after_setloc): Likewise.
13530 (emit_jump_insn_after_setloc): Likewise.
13531 (emit_call_insn_after_setloc): Likewise.
13532 (emit_debug_insn_after_setloc): Likewise.
13533 (emit_pattern_before_setloc): Likewise for "before" param. Convert
13534 "loc" param from "int" to "location_t".
13535 (emit_pattern_before): Convert NULL_RTX to NULL.
13536 (emit_insn_before_setloc): Convert "loc" param from "int" to
13538 (emit_jump_insn_before_setloc): Likewise.
13539 (emit_call_insn_before_setloc): Likewise.
13540 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
13541 rtx_insn *. Convert "loc" param from "int" to "location_t".
13542 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
13543 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
13544 Convert 3rd param from "int" to "location_t".
13545 (emit_barrier_before, emit_barrier_after, next_real_insn):
13546 Strengthen param from rtx to rtx_insn *.
13547 (emit_label_before): Strengthen 1st param from "rtx" to
13548 "rtx_code_label *". Strengthen 2nd param from "rtx" to
13550 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
13551 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
13552 Strengthen 2nd param from "rtx" to "rtx_insn *".
13553 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
13554 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
13555 Likewise. Convert 3rd param from "int" to "location_t".
13556 (emit_label_after): Strengthen 1st param from "rtx" to
13557 "rtx_code_label *".
13558 (next_real_insn, remove_insn): Strengthen param from "rtx" to
13560 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
13561 from "rtx" to "rtx_insn *".
13563 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
13565 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
13566 bodies streamed in with -Q.
13567 * dumpfile.c (dump_files): Add lto-stream-out dump file.
13568 * dumpfile.h (tree_dump_index): Add lto_stream_out.
13569 * gimple-streamer-out.c: Include gimple-pretty-print.h
13570 (output_bb): Dump stmts streamed.
13571 * lto-section-out.c: Include print-tree.h
13572 (lto_begin_section): Dump sections created.
13573 (lto_output_decl_index): Dump decl encoded.
13574 * lto-streamer-out.c: Include print-tree.h
13575 (create_output_block): Dump output block created.
13576 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
13577 (output_function): Dump function output.
13578 (output_constructor): Dump constructor streamed.
13579 (write_global_stream): Output indexes encoded.
13580 (produce_asm_for_decls): Dump streams encoded.
13581 * lto-streamer.c (streamer_dump_file): New global var.
13582 * lto-streamer.h (streamer_dump_file): Declare.
13583 * passes.c (ipa_write_summaries): Initialize streamer dump.
13584 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
13587 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
13590 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
13591 offsets for register save directives. Emit a second batch of save
13592 directives, if need be, when the function accesses prior frames.
13594 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13596 * config/arc/fpu.md (fmasf4): Force operand to register.
13597 (fnmasf4): Likewise.
13599 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13601 * config/arc/arc-protos.h (arc_pad_return): Remove.
13602 * config/arc/arc.c (machine_function): Remove force_short_suffix
13604 (arc_print_operand): Adjust printing of '&'.
13605 (arc_verify_short): Remove conditional printing of short suffix.
13606 (arc_final_prescan_insn): Remove reference to size_reason.
13607 (pad_return): New function.
13608 (arc_reorg): Call pad_return.
13609 (arc_pad_return): Remove.
13610 (arc_init_machine_status): Remove reference to force_short_suffix.
13611 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
13612 (attr length): When attribute iscompact is true force to 2
13613 regardless; in the case of maybe check if we want to force the
13614 instruction to have 4 bytes length.
13615 (nopv): Change it to generate 4 byte long nop as well.
13616 (blockage): New pattern.
13617 (simple_return): Remove call to arc_pad_return.
13618 (p_return_i): Likewise.
13620 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13622 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
13624 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13626 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
13629 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13631 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
13632 for ARC700 and ARCv2.
13634 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
13637 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
13638 operands[2] instead of operands[1].
13641 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13643 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
13644 case, check whether the outer register overlaps an unallocatable
13645 register, not just whether it fits the required class.
13647 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
13649 * poly-int.h (can_div_trunc_p): Add new overload in which all values
13651 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
13652 (memrefs_conflict_p): Likewise.
13653 (init_alias_analysis): Likewise.
13654 * cfgexpand.c (expand_debug_expr): Likewise.
13655 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
13656 * cse.c (fold_rtx): Likewise.
13657 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
13658 * expr.c (emit_block_move_hints): Likewise.
13659 (clear_storage_hints, push_block, emit_push_insn): Likewise.
13660 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
13661 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
13662 (emit_group_store): Likewise.
13663 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
13664 to read the PRE/POST_MODIFY increment.
13665 * calls.c (store_one_arg): Use strip_offset.
13666 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
13668 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
13670 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
13671 (simplify_binary_operation_1): Extend CONST_INT handling to
13673 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
13674 than a HOST_WIDE_INT.
13675 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
13677 (adjust_mems, add_stores): Update accodingly.
13678 (vt_canonicalize_addr): Track polynomial offsets.
13679 (emit_note_insn_var_location): Likewise.
13680 (vt_add_function_parameter): Likewise.
13681 (vt_initialize): Likewise.
13683 2018-06-12 Jeff Law <law@redhat.com>
13685 * config.gcc (alpha*-*-freebsd*): Remove.
13686 * config/alpha/freebsd.h: Remove.
13688 2018-06-12 David Malcolm <dmalcolm@redhat.com>
13691 * spellcheck-tree.c (levenshtein_distance): Rename to...
13692 (get_edit_distance): ...this, and update for underlying renaming.
13693 * spellcheck-tree.h (levenshtein_distance): Rename to...
13694 (get_edit_distance): ...this.
13695 * spellcheck.c (levenshtein_distance): Rename to...
13696 (get_edit_distance): ...this. Convert from Levenshtein distance
13697 to Damerau-Levenshtein distance by supporting transpositions of
13698 adjacent characters. Rename "v1" to "v_next" and "v0" to
13700 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
13701 (selftest::test_edit_distance_unit_test_oneway): ...this, and
13702 update for underlying renaming.
13703 (selftest::levenshtein_distance_unit_test): Rename to...
13704 (selftest::test_get_edit_distance_unit): ...this, and update for
13705 underlying renaming.
13706 (selftest::test_find_closest_string): Add example from PR 69968
13707 where transposition helps
13708 (selftest::test_metric_conditions): Update for renaming.
13709 (selftest::test_metric_conditions): Likewise.
13710 (selftest::spellcheck_c_tests): Likewise.
13711 * spellcheck.h (levenshtein_distance): Rename both overloads to...
13712 (get_edit_distance): ...this.
13713 (best_match::consider): Update for renaming.
13715 2018-06-12 Martin Sebor <msebor@redhat.com>
13717 PR tree-optimization/85259
13718 * builtins.c (compute_objsize): Handle constant offsets.
13719 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
13720 true iff a warning has been issued.
13721 * gimple.h (gimple_nonartificial_location): New function.
13722 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
13723 gimple_nonartificial_location and handle -Wno-system-headers.
13724 (handle_builtin_stxncpy): Same.
13726 2018-06-12 Martin Sebor <msebor@redhat.com>
13729 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
13731 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13733 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13734 BUILTIN_VEC_XST entries for pointer to double and long long.
13736 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
13739 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13741 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
13744 2018-06-12 Martin Liska <mliska@suse.cz>
13746 * doc/options.texi: Document IntegerRange.
13748 2018-06-12 Martin Liska <mliska@suse.cz>
13750 * config/i386/i386.opt: Make MPX-related options as Deprecated.
13751 * opt-functions.awk: Handle Deprecated flag.
13752 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
13754 (read_cmdline_option): Report warning for a deprecated option.
13755 * opts.h (struct cl_option): Add new field cl_deprecated.
13756 (CL_ERR_DEPRECATED): New.
13758 2018-06-12 Martin Liska <mliska@suse.cz>
13760 * doc/options.texi: Document Deprecated option flag.
13762 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13764 * config/arc/arc-arch.h (arc_extras): New enum.
13765 (arc_cpu_t):Add field extra.
13766 (arc_cpu_types): Consider the extras.
13767 * config/arc/arc-cpus.def: Add extras info.
13768 * config/arc/arc-opts.h (processor_type): Consider extra field.
13769 * config/arc/arc.c (arc_override_options): Handle extra field.
13771 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
13773 * config/arc/arc-arch.h: Update ARC_OPTX macro.
13774 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
13776 * config/arc/arc.c (arc_init): Update pic warning.
13777 (irq_range): Update irq range parsing warnings.
13778 (arc_override_options): Update various warning messages.
13779 (arc_handle_aux_attribute): Likewise.
13781 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
13783 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
13785 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
13787 * doc/sourcebuild.texi: Document usage of line number 0 in verify
13788 compiler messages directives.
13790 2018-06-12 Matthew Fortune <mfortune@gmail.com>
13792 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
13793 * config/mips/mips-tables.opt: Regenerate.
13794 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
13796 * doc/invoke.texi: Document -march=i6500.
13798 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
13800 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
13801 (i6400_gpmul): Add cpu_unit.
13802 (i6400_gpdiv): Likewise.
13803 (i6400_msa_add_d): Update reservations.
13804 (i6400_msa_int_add) Likewise.
13805 (i6400_msa_short_logic3) Likewise.
13806 (i6400_msa_short_logic2) Likewise.
13807 (i6400_msa_short_logic) Likewise.
13808 (i6400_msa_move) Likewise.
13809 (i6400_msa_cmp) Likewise.
13810 (i6400_msa_short_float2) Likewise.
13811 (i6400_msa_div_d) Likewise.
13812 (i6400_msa_long_logic1) Likewise.
13813 (i6400_msa_long_logic2) Likewise.
13814 (i6400_msa_mult) Likewise.
13815 (i6400_msa_long_float2) Likewise.
13816 (i6400_msa_long_float4) Likewise.
13817 (i6400_msa_long_float5) Likewise.
13818 (i6400_msa_long_float8) Likewise.
13819 (i6400_fpu_fadd): Include frint type.
13820 (i6400_fpu_store): New define_insn_reservation.
13821 (i6400_fpu_load): Likewise.
13822 (i6400_fpu_move): Likewise.
13823 (i6400_fpu_fcmp): Likewise.
13824 (i6400_fpu_fmadd): Likewise.
13825 (i6400_int_mult): Include imul3nc type and update reservation.
13826 (i6400_int_div): Include idiv3 type and update reservation.
13827 (i6400_int_load): Update to check type not move_type.
13828 (i6400_int_store): Likewise.
13829 (i6400_int_prefetch): Set zero latency.
13831 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
13833 * gcc.c: Document new %@{...} sequence.
13834 (LINK_COMMAND_SPEC): Use it for the -L switches.
13835 (cpp_unique_options): Use it for the -I switches.
13836 (at_file_argbuf): New global variable.
13837 (in_at_file): Likewise.
13838 (alloc_args): Create at_file_argbuf.
13839 (clear_args): Truncate at_file_argbuf.
13840 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
13841 (open_at_file): New function.
13842 (close_at_file): Likewise.
13843 (create_at_file): Delete.
13844 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
13847 (validate_switches_from_spec): Deal with %@{...} sequence.
13848 (validate_switches): Likewise.
13849 (driver::finalize): Call clear_args.
13851 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
13853 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
13855 2018-06-11 Martin Sebor <msebor@redhat.com>
13857 * doc/invoke.texi (-Wall): List -Wc++17-compat.
13858 (Wno-class-memaccess): Add @opindex.
13859 (Wno-templates, Wno-multiple-inheritance): Same.
13860 (Wno-virtual-inheritance, Wno-namespaces): Same.
13861 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
13862 (Wno-format-overflow, Wno-format-truncation): Same.
13863 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
13864 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
13865 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
13866 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
13867 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
13868 (Wno-misspelled-isr): Same.
13870 2018-06-11 Martin Sebor <msebor@redhat.com>
13872 * PR tree-optimization/86083
13873 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
13875 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
13877 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
13879 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
13882 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
13883 on the correct operand.
13884 (*movdi_internal64): Ditto.
13886 2018-06-11 Martin Liska <mliska@suse.cz>
13888 PR tree-optimization/86089
13889 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
13891 2018-06-11 Julia Koval <julia.koval@intel.com>
13893 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
13894 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
13895 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
13897 2018-06-11 Olivier Hainque <hainque@adacore.com>
13899 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
13900 for Ada with strict dwarf2.
13902 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
13905 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
13908 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
13910 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
13912 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
13914 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
13917 2018-06-08 Martin Liska <mliska@suse.cz>
13919 * tree-cfg.h (debug_function): Fix argument type to match
13922 2018-06-08 Martin Liska <mliska@suse.cz>
13924 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
13925 Remove usage of MPX-related (and removed) fields.
13926 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
13928 2018-06-08 David Malcolm <dmalcolm@redhat.com>
13930 * cfg.c (debug): Use TDF_NONE rather than 0.
13931 * cfghooks.c (debug): Likewise.
13932 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
13933 (struct dump_option_value_info): Convert to...
13934 (struct kv_pair): ...this template type.
13935 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
13937 (optinfo_verbosity_options): Likewise.
13938 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
13940 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
13941 than int for "optgroup_flags" param.
13942 (dump_generic_expr_loc): Use dump_flags_t rather than int for
13944 (dump_dec): Likewise.
13945 (dump_finish): Use TDF_NONE rather than 0.
13946 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
13947 rather than int for "optgroup_flags" param. Use TDF_NONE rather
13948 than 0. Update for change to option_ptr.
13949 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
13950 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
13951 0. Update for changes to optinfo_verbosity_options and
13953 (opt_info_switch_p): Convert optgroup_flags from int to
13955 (dump_basic_block): Use dump_flags_t rather than int
13956 for "dump_kind" param.
13957 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
13958 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
13959 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
13960 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
13961 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
13962 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
13963 TDF_NONE): Convert from macros to...
13964 (enum dump_flag): ...this new enum.
13965 (dump_flags_t): Update to use enum.
13966 (operator|, operator&, operator~, operator|=, operator&=):
13967 Implement for dump_flags_t.
13968 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
13969 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
13970 Convert from macros to...
13971 (enum optgroup_flag): ...this new enum.
13972 (optgroup_flags_t): New typedef.
13973 (operator|, operator|=): Implement for optgroup_flags_t.
13974 (struct dump_file_info): Convert field "alt_flags" to
13975 dump_flags_t. Convert field "optgroup_flags" to
13977 (dump_basic_block): Use dump_flags_t rather than int for param.
13978 (dump_generic_expr_loc): Likewise.
13979 (dump_dec): Likewise.
13980 (dump_register): Convert param "optgroup_flags" to
13982 (opt_info_enable_passes): Likewise.
13983 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
13985 * gimple-pretty-print.c (debug): Likewise.
13986 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
13987 (merged_store_group::apply_stores): Likewise.
13988 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
13989 * gimple.c (verify_gimple_pp): Likewise.
13990 * graphite-poly.c (print_pbb_body): Likewise.
13991 * passes.c (pass_manager::register_one_dump_file): Convert
13992 local "optgroup_flags" to optgroup_flags_t.
13993 * print-tree.c (print_node): Use TDF_NONE rather than 0.
13995 (debug_body): Likewise.
13996 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
13997 to optgroup_flags_t.
13998 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
14000 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
14001 (convert_mult_to_fma): Likewise.
14002 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
14003 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
14004 * tree-vect-data-refs.c (dump_lower_bound): Convert param
14005 "dump_kind" to dump_flags_t.
14007 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
14009 * config/rs6000/rs6000.c (min, max): Delete.
14011 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
14013 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
14014 -mabi=spe and -mabi=no-spe.
14016 2018-06-08 Martin Liska <mliska@suse.cz>
14018 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
14019 where we expect an existing summary.
14021 2018-06-08 Martin Liska <mliska@suse.cz>
14023 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
14024 * ipa-inline.h (estimate_edge_growth): Likewise.
14026 2018-06-08 Martin Liska <mliska@suse.cz>
14028 * cgraph.c (function_version_hasher::hash): Use
14029 cgraph_node::get_uid ().
14030 (function_version_hasher::equal):
14031 * cgraph.h (cgraph_node::get_uid): New method.
14032 * ipa-inline.c (update_caller_keys): Use
14033 cgraph_node::get_uid ().
14034 (update_callee_keys): Likewise.
14035 * ipa-utils.c (searchc): Likewise.
14036 (ipa_reduced_postorder): Likewise.
14037 * lto-cgraph.c (input_node): Likewise.
14038 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
14039 * symbol-summary.h (symtab_insertion): Likewise.
14040 (symtab_removal): Likewise.
14041 (symtab_duplication): Likewise.
14042 * tree-pretty-print.c (dump_function_header): Likewise.
14043 * tree-sra.c (convert_callers_for_node): Likewise.
14045 2018-06-08 Martin Liska <mliska@suse.cz>
14047 * cgraph.c (symbol_table::create_edge): Always assign a new
14049 (symbol_table::free_edge): Do not recycle numbers.
14050 * cgraph.h (cgraph_edge::get): New method.
14051 * symbol-summary.h (symtab_removal): Use it.
14052 (symtab_duplication): Likewise.
14053 (call_summary::hashable_uid): Remove.
14055 2018-06-08 Martin Liska <mliska@suse.cz>
14057 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
14058 (initialize_growth_caches): Remove.
14059 (free_growth_caches): Likewise.
14060 (do_estimate_edge_time): Use edge_growth_cache.
14061 (do_estimate_edge_size): Likewise.
14062 (do_estimate_edge_hints): Likewise.
14063 * ipa-inline.c (reset_edge_caches): Likewise.
14064 (recursive_inlining): Likewise.
14065 (inline_small_functions): Likewise.
14066 * ipa-inline.h (initialize_growth_caches): Remove.
14067 (estimate_edge_size): Likewise.
14068 (estimate_edge_time): Likewise.
14069 (estimate_edge_hints): Likewise.
14070 (reset_edge_growth_cache): Likewise.
14071 * symbol-summary.h (call_summary::remove): New method.
14073 2018-06-08 Martin Liska <mliska@suse.cz>
14075 * ipa-cp.c (class edge_clone_summary): New summary.
14076 (grow_edge_clone_vectors): Remove.
14077 (ipcp_edge_duplication_hook): Remove.
14078 (class edge_clone_summary_t): New call_summary class.
14079 (ipcp_edge_removal_hook): Remove.
14080 (edge_clone_summary_t::duplicate): New function.
14081 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
14082 (create_specialized_node): Likewise.
14083 (ipcp_driver): Initialize edge_clone_summaries and do not
14086 2018-06-08 Martin Liska <mliska@suse.cz>
14088 * symbol-summary.h (get): New function.
14089 (call_summary::m_initialize_when_cloning): New class member.
14091 2018-06-08 Martin Liska <mliska@suse.cz>
14093 * cgraph.c (cgraph_node::remove): Do not recycle uid.
14094 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
14095 (symbol_table::allocate_cgraph_symbol): Do not set uid.
14096 * passes.c (uid_hash_t): Record removed_nodes by their uids.
14097 (remove_cgraph_node_from_order): Use the removed_nodes set.
14098 (do_per_function_toporder): Likwise.
14099 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
14100 instead of summary_uid.
14101 (symtab_removal): Likewise.
14102 (symtab_duplication): Likewise.
14104 2018-06-08 Martin Liska <mliska@suse.cz>
14106 * ipa-cp.c (ipcp_store_bits_results): Use
14107 ipcp_transformation_sum.
14108 (ipcp_store_vr_results): Likewise.
14109 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
14111 (ipcp_transformation_initialize): ... this.
14112 (ipa_set_node_agg_value_chain):
14113 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
14114 (write_ipcp_transformation_info): Likewise.
14115 (read_ipcp_transformation_info): Likewise.
14116 (ipcp_update_bits): Likewise.
14117 (ipcp_update_vr): Likewise.
14118 (ipcp_transform_function): Likewise.
14119 * ipa-prop.h: Rename ipcp_transformation_summary to
14120 ipcp_transformation.
14121 (class ipcp_transformation_t): New function summary.
14122 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
14123 (ipa_get_agg_replacements_for_node): Likewise.
14125 2018-06-08 Martin Liska <mliska@suse.cz>
14127 * ipa-pure-const.c (struct funct_state_d): Do it class instead
14129 (class funct_state_summary_t): New function_summary class.
14130 (has_function_state): Remove.
14131 (get_function_state): Likewise.
14132 (set_function_state): Likewise.
14133 (add_new_function): Likewise.
14134 (funct_state_summary_t::insert): New function.
14135 (duplicate_node_data): Remove.
14136 (remove_node_data): Remove.
14137 (funct_state_summary_t::duplicate): New function.
14138 (register_hooks): Create new funct_state_summaries.
14139 (pure_const_generate_summary): Use it.
14140 (pure_const_write_summary): Likewise.
14141 (pure_const_read_summary): Likewise.
14142 (propagate_pure_const): Likewise.
14143 (propagate_nothrow): Likewise.
14144 (dump_malloc_lattice): Likewise.
14145 (propagate_malloc): Likewise.
14146 (execute): Do not register hooks, just remove summary
14148 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
14151 2018-06-08 Martin Liska <mliska@suse.cz>
14153 * ipa-reference.c (remove_node_data): Remove.
14154 (duplicate_node_data): Likewise.
14155 (class ipa_ref_var_info_summary_t): New class.
14156 (class ipa_ref_opt_summary_t): Likewise.
14157 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
14158 (get_reference_optimization_summary): Use
14159 ipa_ref_opt_sum_summaries.
14160 (set_reference_vars_info): Remove.
14161 (set_reference_optimization_summary): Likewise.
14162 (ipa_init): Create summaries.
14163 (init_function_info): Use function summary.
14164 (ipa_ref_opt_summary_t::duplicate): New function.
14165 (ipa_ref_opt_summary_t::remove): New function.
14166 (get_read_write_all_from_node): Fix GNU coding style.
14167 (propagate): Use function summary.
14168 (write_node_summary_p): Fix GNU coding style.
14169 (stream_out_bitmap): Likewise.
14170 (ipa_reference_read_optimization_summary): Use function summary.
14171 (ipa_reference_c_finalize): Do not release hooks.
14173 2018-06-08 Martin Liska <mliska@suse.cz>
14175 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
14176 (analyze_function_body): Extract multiple calls of get_create.
14177 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
14178 * ipa-inline.c (recursive_inlining): Use ::get method.
14179 * ipa-inline.h (estimate_edge_growth): Likewise.
14181 2018-06-08 Martin Liska <mliska@suse.cz>
14183 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
14185 (hsa_function_summary::hsa_function_summary): Use the new enum
14187 (hsa_gpu_implementation_p): Use hsa_summaries::get.
14188 * hsa-gen.c (hsa_get_host_function): Likewise.
14189 (get_brig_function_name): Likewise.
14190 * ipa-hsa.c (process_hsa_functions): Likewise.
14191 (ipa_hsa_write_summary): Likewise.
14192 * symbol-summary.h (symtab_duplication): Use ::get function/
14193 (get): New function.
14195 2018-06-08 Martin Liska <mliska@suse.cz>
14197 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
14199 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
14200 (hsa_register_kernel): Likewise.
14201 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
14202 * hsa-gen.c (hsa_get_host_function): Likewise.
14203 (get_brig_function_name): Likewise.
14204 (generate_hsa): Likewise.
14205 (pass_gen_hsail::execute): Likewise.
14206 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
14207 (devirtualization_time_bonus): Likewise.
14208 (ipcp_propagate_stage): Likewise.
14209 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
14210 (edge_set_predicate): Likewise.
14211 (evaluate_conditions_for_known_args): Likewise.
14212 (evaluate_properties_for_edge): Likewise.
14213 (ipa_fn_summary::reset): Likewise.
14214 (ipa_fn_summary_t::duplicate): Likewise.
14215 (dump_ipa_call_summary): Likewise.
14216 (ipa_dump_fn_summary): Likewise.
14217 (analyze_function_body): Likewise.
14218 (compute_fn_summary): Likewise.
14219 (estimate_edge_devirt_benefit): Likewise.
14220 (estimate_edge_size_and_time): Likewise.
14221 (estimate_calls_size_and_time): Likewise.
14222 (estimate_node_size_and_time): Likewise.
14223 (inline_update_callee_summaries): Likewise.
14224 (remap_edge_change_prob): Likewise.
14225 (remap_edge_summaries): Likewise.
14226 (ipa_merge_fn_summary_after_inlining): Likewise.
14227 (ipa_update_overall_fn_summary): Likewise.
14228 (read_ipa_call_summary): Likewise.
14229 (inline_read_section): Likewise.
14230 (write_ipa_call_summary): Likewise.
14231 (ipa_fn_summary_write): Likewise.
14232 (ipa_free_fn_summary): Likewise.
14233 * ipa-hsa.c (process_hsa_functions): Likewise.
14234 (ipa_hsa_write_summary): Likewise.
14235 (ipa_hsa_read_section): Likewise.
14236 * ipa-icf.c (sem_function::merge): Likewise.
14237 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
14238 (do_estimate_edge_time): Likewise.
14239 (estimate_size_after_inlining): Likewise.
14240 (estimate_growth): Likewise.
14241 (growth_likely_positive): Likewise.
14242 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
14243 (inline_call): Likewise.
14244 * ipa-inline.c (caller_growth_limits): Likewise.
14245 (can_inline_edge_p): Likewise.
14246 (can_inline_edge_by_limits_p): Likewise.
14247 (compute_uninlined_call_time): Likewise.
14248 (compute_inlined_call_time): Likewise.
14249 (want_inline_small_function_p): Likewise.
14250 (edge_badness): Likewise.
14251 (update_caller_keys): Likewise.
14252 (update_callee_keys): Likewise.
14253 (recursive_inlining): Likewise.
14254 (inline_small_functions): Likewise.
14255 (inline_to_all_callers_1): Likewise.
14256 (dump_overall_stats): Likewise.
14257 (early_inline_small_functions): Likewise.
14258 (early_inliner): Likewise.
14259 * ipa-inline.h (estimate_edge_growth): Likewise.
14260 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
14261 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14262 * ipa-prop.h (IPA_NODE_REF): Likewise.
14263 (IPA_EDGE_REF): Likewise.
14264 * ipa-pure-const.c (malloc_candidate_p): Likewise.
14265 (propagate_malloc): Likewise.
14266 * ipa-split.c (execute_split_functions): Likewise.
14267 * symbol-summary.h: Rename get to get_create.
14269 (get_create): Likewise.
14270 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
14272 2018-06-08 Martin Liska <mliska@suse.cz>
14274 * symbol-summary.h (release): Move definition out of class
14276 (symtab_removal): Likewise.
14277 (symtab_duplication): Likewise.
14279 2018-06-08 Martin Liska <mliska@suse.cz>
14281 * symbol-summary.h (function_summary): Move constructor
14282 implementation out of class declaration.
14283 (release): Likewise.
14284 (symtab_insertion): Likewise.
14285 (symtab_removal): Likewise.
14286 (symtab_duplication): Likewise.
14289 2018-06-08 Martin Liska <mliska@suse.cz>
14291 * Makefile.in: Remove support for MPX (macros, related functions,
14292 fields in cgraph_node, ...).
14293 * builtin-types.def (BT_BND): Likewise.
14294 (BT_FN_BND_CONST_PTR): Likewise.
14295 (BT_FN_CONST_PTR_BND): Likewise.
14296 (BT_FN_VOID_PTR_BND): Likewise.
14297 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
14298 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
14299 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
14300 (expand_builtin_mempcpy_with_bounds): Likewise.
14301 (expand_builtin_memset_with_bounds): Likewise.
14302 (expand_builtin_memset_args): Likewise.
14303 (std_expand_builtin_va_start): Likewise.
14304 (expand_builtin): Likewise.
14305 (expand_builtin_with_bounds): Likewise.
14306 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
14307 (DEF_LIB_BUILTIN_CHKP): Likewise.
14308 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
14309 (DEF_CHKP_BUILTIN): Likewise.
14310 (BUILT_IN_MEMCPY): Likewise.
14311 (BUILT_IN_MEMMOVE): Likewise.
14312 (BUILT_IN_MEMPCPY): Likewise.
14313 (BUILT_IN_MEMSET): Likewise.
14314 (BUILT_IN_STPCPY): Likewise.
14315 (BUILT_IN_STRCAT): Likewise.
14316 (BUILT_IN_STRCHR): Likewise.
14317 (BUILT_IN_STRCPY): Likewise.
14318 (BUILT_IN_STRLEN): Likewise.
14319 (BUILT_IN_MEMCPY_CHK): Likewise.
14320 (BUILT_IN_MEMMOVE_CHK): Likewise.
14321 (BUILT_IN_MEMPCPY_CHK): Likewise.
14322 (BUILT_IN_MEMSET_CHK): Likewise.
14323 (BUILT_IN_STPCPY_CHK): Likewise.
14324 (BUILT_IN_STRCAT_CHK): Likewise.
14325 (BUILT_IN_STRCPY_CHK): Likewise.
14326 * calls.c (store_bounds): Likewise.
14327 (emit_call_1): Likewise.
14328 (special_function_p): Likewise.
14329 (maybe_warn_nonstring_arg): Likewise.
14330 (initialize_argument_information): Likewise.
14331 (finalize_must_preallocate): Likewise.
14332 (compute_argument_addresses): Likewise.
14333 (expand_call): Likewise.
14334 * cfgexpand.c (expand_call_stmt): Likewise.
14335 (expand_return): Likewise.
14336 (expand_gimple_stmt_1): Likewise.
14337 (pass_expand::execute): Likewise.
14338 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
14339 (cgraph_node::remove): Likewise.
14340 (cgraph_node::dump): Likewise.
14341 (cgraph_node::verify_node): Likewise.
14342 * cgraph.h (chkp_function_instrumented_p): Likewise.
14343 (symtab_node::get_alias_target): Likewise.
14344 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
14345 (cgraph_local_p): Likewise.
14346 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
14347 (cgraph_edge::rebuild_references): Likewise.
14348 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
14349 (walk_polymorphic_call_targets): Likewise.
14350 (cgraph_node::expand_thunk): Likewise.
14351 (symbol_table::output_weakrefs): Likewise.
14352 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
14353 (ix86_handle_option): Likewise.
14354 * config/i386/constraints.md: Likewise.
14355 * config/i386/i386-builtin-types.def (BND): Likewise.
14359 * config/i386/i386-builtin.def (BDESC_END): Likewise.
14360 (BDESC_FIRST): Likewise.
14362 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
14363 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
14364 * config/i386/i386.c (enum reg_class): Likewise.
14365 (ix86_target_string): Likewise.
14366 (ix86_option_override_internal): Likewise.
14367 (ix86_conditional_register_usage): Likewise.
14368 (ix86_valid_target_attribute_inner_p): Likewise.
14369 (ix86_set_indirect_branch_type): Likewise.
14370 (ix86_set_current_function): Likewise.
14371 (ix86_function_arg_regno_p): Likewise.
14372 (init_cumulative_args): Likewise.
14373 (ix86_function_arg_advance): Likewise.
14374 (ix86_function_arg): Likewise.
14375 (ix86_pass_by_reference): Likewise.
14376 (ix86_function_value_regno_p): Likewise.
14377 (ix86_function_value_1): Likewise.
14378 (ix86_function_value_bounds): Likewise.
14379 (ix86_return_in_memory): Likewise.
14380 (ix86_setup_incoming_vararg_bounds): Likewise.
14381 (ix86_va_start): Likewise.
14382 (indirect_thunk_need_prefix): Likewise.
14383 (print_reg): Likewise.
14384 (ix86_print_operand): Likewise.
14385 (ix86_expand_call): Likewise.
14386 (ix86_output_function_return): Likewise.
14387 (reg_encoded_number): Likewise.
14388 (BDESC_VERIFYS): Likewise.
14389 (ix86_init_mpx_builtins): Likewise.
14390 (ix86_init_builtins): Likewise.
14391 (ix86_emit_cmove): Likewise.
14392 (ix86_emit_move_max): Likewise.
14393 (ix86_expand_builtin): Likewise.
14394 (ix86_builtin_mpx_function): Likewise.
14395 (ix86_get_arg_address_for_bt): Likewise.
14396 (ix86_load_bounds): Likewise.
14397 (ix86_store_bounds): Likewise.
14398 (ix86_load_returned_bounds): Likewise.
14399 (ix86_store_returned_bounds): Likewise.
14400 (ix86_class_likely_spilled_p): Likewise.
14401 (ix86_hard_regno_mode_ok): Likewise.
14402 (x86_order_regs_for_local_alloc): Likewise.
14403 (ix86_mitigate_rop): Likewise.
14404 (ix86_bnd_prefixed_insn_p): Likewise.
14405 (ix86_mpx_bound_mode): Likewise.
14406 (ix86_make_bounds_constant): Likewise.
14407 (ix86_initialize_bounds): Likewise.
14408 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
14409 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
14410 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
14411 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
14412 (TARGET_CHKP_BOUND_MODE): Likewise.
14413 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
14414 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
14415 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
14416 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
14417 * config/i386/i386.h (TARGET_MPX): Likewise.
14418 (TARGET_MPX_P): Likewise.
14419 (VALID_BND_REG_MODE): Likewise.
14420 (FIRST_BND_REG): Likewise.
14421 (LAST_BND_REG): Likewise.
14422 (enum reg_class): Likewise.
14423 (BND_REG_P): Likewise.
14424 (BND_REGNO_P): Likewise.
14425 (BNDmode): Likewise.
14426 (ADJUST_INSN_LENGTH): Likewise.
14427 * config/i386/i386.md: Likewise.
14428 * config/i386/i386.opt: Likewise.
14429 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
14430 (defined): Likewise.
14431 (LINK_MPX): Likewise.
14432 (MPX_SPEC): Likewise.
14433 (LIBMPX_SPEC): Likewise.
14434 (LIBMPXWRAPPERS_SPEC): Likewise.
14435 (CHKP_SPEC): Likewise.
14436 * config/i386/predicates.md: Likewise.
14437 * dbxout.c (dbxout_type): Likewise.
14438 * doc/extend.texi: Likewise.
14439 * doc/invoke.texi: Likewise.
14440 * doc/md.texi: Likewise.
14441 * doc/tm.texi: Likewise.
14442 * doc/tm.texi.in: Likewise.
14443 * dwarf2out.c (is_base_type): Likewise.
14444 (gen_formal_types_die): Likewise.
14445 (gen_subprogram_die): Likewise.
14446 (gen_type_die_with_usage): Likewise.
14447 (gen_decl_die): Likewise.
14448 (dwarf2out_late_global_decl): Likewise.
14449 * expr.c (expand_assignment): Likewise.
14450 (emit_storent_insn): Likewise.
14451 (store_expr_with_bounds): Likewise.
14452 (store_expr): Likewise.
14453 (expand_expr_real_1): Likewise.
14454 * expr.h (store_expr_with_bounds): Likewise.
14455 * function.c (use_register_for_decl): Likewise.
14456 (struct bounds_parm_data): Likewise.
14457 (assign_parms_augmented_arg_list): Likewise.
14458 (assign_parm_find_entry_rtl): Likewise.
14459 (assign_parm_is_stack_parm): Likewise.
14460 (assign_parm_load_bounds): Likewise.
14461 (assign_bounds): Likewise.
14462 (assign_parms): Likewise.
14463 (expand_function_start): Likewise.
14464 * gcc.c (CHKP_SPEC): Likewise.
14465 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
14466 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
14467 (wrestrict_dom_walker::check_call): Likewise.
14468 * gimple.c (gimple_build_call_from_tree): Likewise.
14469 * gimple.h (enum gf_mask): Likewise.
14470 (gimple_call_with_bounds_p): Likewise.
14471 (gimple_call_set_with_bounds): Likewise.
14472 * gimplify.c (gimplify_init_constructor): Likewise.
14473 * ipa-cp.c (initialize_node_lattices): Likewise.
14474 (propagate_constants_across_call): Likewise.
14475 (find_more_scalar_values_for_callers_subset): Likewise.
14476 * ipa-hsa.c (process_hsa_functions): Likewise.
14477 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
14478 * ipa-icf.c (sem_function::merge): Likewise.
14479 * ipa-inline.c (early_inliner): Likewise.
14480 * ipa-pure-const.c (warn_function_noreturn): Likewise.
14481 (warn_function_cold): Likewise.
14482 (propagate_pure_const): Likewise.
14483 * ipa-ref.h (enum GTY): Likewise.
14484 * ipa-split.c (find_retbnd): Likewise.
14485 (consider_split): Likewise.
14486 (split_function): Likewise.
14487 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
14488 * ipa.c (walk_polymorphic_call_targets): Likewise.
14489 (symbol_table::remove_unreachable_nodes): Likewise.
14490 (process_references): Likewise.
14491 (cgraph_build_static_cdtor_1): Likewise.
14492 * lto-cgraph.c (lto_output_node): Likewise.
14493 (output_refs): Likewise.
14494 (compute_ltrans_boundary): Likewise.
14495 (input_overwrite_node): Likewise.
14496 (input_node): Likewise.
14497 (input_cgraph_1): Likewise.
14498 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
14499 * passes.c (pass_manager::execute_early_local_passes): Likewise.
14500 (class pass_chkp_instrumentation_passes): Likewise.
14501 (make_pass_chkp_instrumentation_passes): Likewise.
14502 * passes.def: Likewise.
14503 * rtl.h (struct GTY): Likewise.
14504 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
14505 * stor-layout.c (layout_type): Likewise.
14506 * symtab.c: Likewise.
14507 * target.def: Likewise.
14508 * targhooks.c (default_chkp_bound_type): Likewise.
14509 (default_chkp_bound_mode): Likewise.
14510 (default_builtin_chkp_function): Likewise.
14511 (default_chkp_function_value_bounds): Likewise.
14512 (default_chkp_make_bounds_constant): Likewise.
14513 (default_chkp_initialize_bounds): Likewise.
14514 * targhooks.h (default_chkp_bound_type): Likewise.
14515 (default_chkp_bound_mode): Likewise.
14516 (default_builtin_chkp_function): Likewise.
14517 (default_chkp_function_value_bounds): Likewise.
14518 (default_chkp_make_bounds_constant): Likewise.
14519 (default_chkp_initialize_bounds): Likewise.
14520 * toplev.c (compile_file): Likewise.
14521 (process_options): Likewise.
14522 * tree-core.h (DEF_BUILTIN): Likewise.
14523 (DEF_BUILTIN_CHKP): Likewise.
14524 * tree-inline.c (declare_return_variable): Likewise.
14525 (remap_gimple_stmt): Likewise.
14526 (copy_bb): Likewise.
14527 (initialize_inlined_parameters): Likewise.
14528 (expand_call_inline): Likewise.
14529 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
14530 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
14531 (make_pass_ipa_chkp_produce_thunks): Likewise.
14532 (make_pass_chkp): Likewise.
14533 (make_pass_chkp_opt): Likewise.
14534 (make_pass_chkp_instrumentation_passes): Likewise.
14535 * tree-pretty-print.c (dump_generic_node): Likewise.
14536 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
14537 * tree-ssa-dce.c (propagate_necessity): Likewise.
14538 (eliminate_unnecessary_stmts): Likewise.
14539 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
14540 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
14541 * tree-ssa-sccvn.h: Likewise.
14542 * tree-ssa-strlen.c (get_string_length): Likewise.
14543 (valid_builtin_call): Likewise.
14544 (adjust_last_stmt): Likewise.
14545 (handle_builtin_strchr): Likewise.
14546 (handle_builtin_strcpy): Likewise.
14547 (handle_builtin_stxncpy): Likewise.
14548 (handle_builtin_memcpy): Likewise.
14549 (handle_builtin_strcat): Likewise.
14550 (strlen_check_and_optimize_stmt): Likewise.
14551 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
14552 * tree-streamer-in.c: Likewise.
14553 * tree-streamer.c (record_common_node): Likewise.
14554 * tree.c (tree_code_size): Likewise.
14555 (wide_int_to_tree_1): Likewise.
14556 (type_contains_placeholder_1): Likewise.
14557 (build_common_tree_nodes): Likewise.
14558 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
14559 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
14560 (POINTER_BOUNDS_P): Likewise.
14561 (BOUNDED_TYPE_P): Likewise.
14562 (BOUNDED_P): Likewise.
14563 (CALL_WITH_BOUNDS_P): Likewise.
14564 (pointer_bounds_type_node): Likewise.
14565 * value-prof.c (gimple_ic): Likewise.
14566 * var-tracking.c (vt_add_function_parameters): Likewise.
14567 * varasm.c (make_decl_rtl): Likewise.
14568 (assemble_start_function): Likewise.
14569 (output_constant): Likewise.
14570 (maybe_assemble_visibility): Likewise.
14571 * varpool.c (ctor_for_folding): Likewise.
14572 * chkp-builtins.def: Remove.
14573 * ipa-chkp.c: Remove.
14574 * ipa-chkp.h: Remove.
14575 * rtl-chkp.c: Remove.
14576 * rtl-chkp.h: Remove.
14577 * tree-chkp-opt.c: Remove.
14578 * tree-chkp.c: Remove.
14579 * tree-chkp.h: Remove.
14581 2018-06-07 Carl Love <cel@us.ibm.com>
14583 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14584 vextract_fp_from_shortl): Add BE support.
14586 2018-06-07 Paul Koning <ni1d@arrl.net>
14588 * compare-elim.c (try_merge_compare): Don't merge compare if
14589 address contains a side effect.
14590 (try_eliminate_compare): Likewise.
14592 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
14594 * config.gcc: Support "tremont".
14595 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
14596 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14598 * config/i386/i386.c (m_TREMONT): Define.
14599 (processor_target_table): Add "tremont".
14600 (PTA_TREMONT): Define.
14601 (ix86_lea_outperforms): Add TARGET_TREMONT.
14602 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
14603 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
14604 and M_INTEL_GOLDMONT_PLUS.
14605 (fold_builtin_cpu): Add "tremont".
14606 (ix86_add_stmt_cost): Add TARGET_TREMONT.
14607 (ix86_option_override_internal): Add "tremont".
14608 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
14609 (processor_type): Add PROCESSOR_TREMONT.
14610 * config/i386/x86-tune.def: Add m_TREMONT.
14611 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
14613 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14615 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
14616 symbol defined for msp430i* devices to be lower case.
14618 2018-06-07 Richard Biener <rguenther@suse.de>
14620 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
14621 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
14622 Properly wrap signed arithmetic if overflow wraps.
14624 2018-06-07 Jakub Jelinek <jakub@redhat.com>
14626 PR tree-optimization/69615
14627 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
14628 of a cast from a same precision integral SSA_NAME in a bb dominated
14629 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
14630 cast to utype if rhs2 has already a compatible type.
14632 2018-06-07 Richard Biener <rguenther@suse.de>
14634 PR tree-optimization/85935
14635 * graphite-scop-detection.c (find_params_in_bb): Analyze
14636 condition operands with respect to the correct loop. Assert
14637 the analysis doesn't fail.
14639 2018-06-04 Carl Love <cel@us.ibm.com>
14641 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
14642 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
14643 as it is slightly cheaper.
14644 (first_match_or_eos_index_<mode>):
14645 Calculate index using natural element order.
14646 (first_match_index_<mode>):
14647 Calculate index using natural element order.
14648 (first_match_or_eos_index_<mode>):
14649 Calculate index using natural order.
14650 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
14651 for BE and LE modes.
14652 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
14653 P9V_BUILTIN_VCLZLSBB_V16QI.
14654 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
14657 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14659 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
14660 indentation and line wrap for many prototypes. Add missing
14661 @smallexample directives around block of prototypes for vec_xl and
14664 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
14666 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
14667 track if we pass or return IEEE 128-bit floating point.
14668 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
14669 C++ mangling that is compatible with GCC 8.1.
14670 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
14671 (init_cumulative_args): Note if we pass or return IEEE 128-bit
14672 floating point types.
14673 (rs6000_function_arg_advance_1): Likewise.
14674 (rs6000_mangle_type): Optionally generate mangled names that match
14675 what GCC 8.1 generated for IEEE 128-bit floating point types.
14676 (rs6000_globalize_decl_name): If we have an external function that
14677 passes or returns IEEE 128-bit types, generate a weak reference
14678 from the mangled name used in GCC 8.1 to the current mangled
14680 (rs6000_init_builtins): Make __ibm128 use the long double type if
14681 long double is IBM extended double. Make __float128 use the long
14682 double type if long double is IEEE 128-bit.
14685 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
14686 macro for __ibm128 built-in functions.
14687 (PACK_IF): Add __ibm128 pack/unpack functions.
14688 (UNPACK_IF): Likewise.
14689 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
14690 enable long double built-in functions if long double is IEEE
14691 128-bit floating point.
14692 (rs6000_invalid_builtin): Update long double built-in function
14694 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
14695 functions, adjust the built-in function to use the long double
14696 built-in function if __ibm128 and long double are the same type.
14697 * doc/extend.texi (PowerPC builtins): Update documention for
14698 __builtin_{,un}pack_longdouble. Add documentation for
14699 __builtin_{,un}pack_ibm128.
14701 2018-06-06 Jim Wilson <jimw@sifive.com>
14703 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
14704 (struct machine_function): New field interrupt_mode.
14705 (riscv_handle_type_attribute): New function. Add forward declaration.
14706 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
14707 (riscv_expand_epilogue): Check interrupt_mode field.
14708 (riscv_set_current_function): Check interrupt attribute args and
14709 set interrupt_mode field.
14710 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
14711 (riscv_sret, riscv_uret): New.
14712 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
14713 new arguments to interrupt attribute.
14715 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
14718 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
14719 Don't handle -mcpu=power8 if -mpower9-vector is also used.
14721 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
14723 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
14724 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
14725 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
14726 several redundant entries.
14728 2018-06-06 David Malcolm <dmalcolm@redhat.com>
14730 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
14731 type from "rtx" to "rtx_insn *".
14732 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
14733 for local "call_insn", removing cast.
14734 (ix86_expand_call): Likewise, introducing a "call_insn" local.
14736 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
14738 PR tree-optimization/86066
14739 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
14740 for BIT_INSERT_EXPR stores.
14742 2018-06-06 Richard Biener <rguenther@suse.de>
14744 PR tree-optimization/86062
14745 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
14746 component refs ontop
14747 of to be offsetted base.
14749 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
14751 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
14752 to be static and remove check on interrupt attribute name.
14754 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
14756 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
14757 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
14759 2018-06-05 Steve Ellcey <sellcey@cavium.com>
14762 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
14764 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
14765 Remove second argument, change how error is called.
14766 (aarch64_layout_arg): Remove second argument from
14767 aarch64_err_no_fpadvsimd call.
14768 (aarch64_init_cumulative_args): Ditto.
14769 (aarch64_gimplify_va_arg_expr): Ditto.
14770 * config/aarch64/aarch64.md (mov<mode>): Ditto.
14772 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
14774 * config/i386/i386.md (simple_return_indirect_internal): New expander.
14775 (*simple_return_indirect_internal<mode>): Rename from
14776 simple_return_indirect_internal. Use W mode iterator.
14777 (rstorssp): New expander.
14778 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
14779 (clrssbsy): New expander.
14780 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
14782 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14784 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
14786 (cmse_check_pointed_object): Likewise.
14788 2018-06-05 Martin Liska <mliska@suse.cz>
14790 PR gcov-profile/47618
14791 * doc/invoke.texi: Document how -fprofile-dir format
14794 2018-06-05 Richard Biener <rguenther@suse.de>
14796 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
14797 removal pretend DOM info isn't available so we do not update
14798 it and only remove edges, not dominated blocks. Actually free
14799 DOM info in case we removed something. Remove unreachable blocks.
14800 (mfb_keep_latches): Work with either DOM info or marked backedges.
14801 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
14802 first. Mark backedges if DOM info isn't available.
14803 (Re-)compute DOM info after cleanup_control_flow_pre.
14805 2018-06-05 Richard Biener <rguenther@suse.de>
14807 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
14808 (locus_discrim_hasher::hash): Adjust.
14809 (locus_discrim_hasher::equal): Likewise.
14810 (next_discriminator_for_locus): Work on line directly.
14811 (same_line_p): Pass in expanded locus1 as well.
14812 (assign_discriminators): Avoid redundant location expansions.
14814 2018-06-05 Richard Biener <rguenther@suse.de>
14816 PR tree-optimization/86046
14817 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
14818 if required after clearing TREE_ADDRESSABLE.
14820 2018-06-05 Richard Biener <rguenther@suse.de>
14822 PR tree-optimization/86047
14823 * tree-ssa-loop.c (for_each_index): Glob handling of all
14824 decls and constants and really handle all of them.
14826 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14829 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
14830 qualifier_void_pointer and qualifier_const_void_pointer.
14831 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
14832 (arm_init_builtins): Handle the above.
14833 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
14834 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
14835 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
14838 2018-06-05 Martin Liska <mliska@suse.cz>
14840 * auto-profile.c (read_autofdo_file): Do not use
14841 gcov_ctr_summary struct.
14842 (afdo_callsite_hot_enough_for_early_inline): Likewise.
14843 * coverage.c (struct counts_entry): Likewise.
14844 (read_counts_file): Read just single summary entry.
14845 (get_coverage_counts): Use gcov_summary struct.
14846 * coverage.h (get_coverage_counts): Likewise.
14847 * gcov-dump.c (dump_working_sets): Likewise.
14848 (tag_summary): Dump just single summary.
14849 * gcov-io.c (gcov_write_summary): Write just histogram
14851 (gcov_read_summary): Read just single summary.
14852 (compute_working_sets): Use gcov_summary struct.
14853 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
14854 of GCOV_COUNTERS_SUMMABLE.
14855 (GCOV_COUNTERS_SUMMABLE): Remove.
14856 (GCOV_FIRST_VALUE_COUNTER): Replace with
14857 GCOV_COUNTER_V_INTERVAL.
14858 (struct gcov_ctr_summary): Remove.
14859 (struct gcov_summary): Directly use fields of former
14861 (compute_working_sets): Use gcov_summary struct.
14862 * gcov.c (read_count_file): Do not use ctrs fields.
14863 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
14865 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
14867 * profile.c: Likewise.
14868 * profile.h: Likewise.
14870 2018-06-05 Martin Liska <mliska@suse.cz>
14872 PR gcov-profile/84846
14873 * gcov.c (output_lines): Print working directory only
14874 in intermediate format.
14876 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
14878 * config/s390/s390-builtin-types.def: Add void function type.
14879 * config/s390/s390-builtins.def: Use the function type for the
14882 2018-06-04 Jim Wilson <jimw@sifive.com>
14884 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
14886 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
14887 and maybe_eh_return. Change regno to unsigned int. Use new args to
14888 handle EH_RETURN_DATA_REGNO registers properly.
14889 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
14890 (riscv_expand_epilogue): Update comment. Change argument name and
14891 type. Update code to use new name and type. Pass new args to
14892 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
14894 * config/riscv/riscv.md (NORMAL_RETURN): New.
14895 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
14896 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
14897 (eh_return): Call gen_eh_return_internal and emit barrier.
14898 (eh_return_internal): Call riscv_expand_epilogue.
14900 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
14902 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
14903 bit_insertion field and declare can_be_merged_into method.
14904 (merged_store_group::can_be_merged_into): New method.
14905 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
14906 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
14907 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
14909 2018-06-04 Richard Biener <rguenther@suse.de>
14911 PR tree-optimization/85955
14912 * builtins.c (fold_builtin_sincos): Convert pointers to
14913 destination to appropriate type before dereferencing.
14915 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
14917 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
14919 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
14921 * expr.c (expand_expr_real_1): Force the operand into memory if
14922 its TYPE_MODE is BLKmode and if there is no integer mode for
14923 the number of bits being extracted.
14925 2018-06-04 Jakub Jelinek <jakub@redhat.com>
14929 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
14930 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
14932 2018-06-04 Richard Biener <rguenther@suse.de>
14934 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
14935 (cleanup_tree_cfg_noloop): ... single caller. Do
14936 start_recording_case_labels later.
14938 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
14940 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
14941 to _IMMINTRIN_H_INCLUDED.
14942 * config/i386/pconfigintrin.h: Ditto.
14943 * config/i386/waitpkgintrin.h: Ditto.
14944 * config/i386/immintrin.h: Add includes for sgxintrin.h,
14945 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
14946 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
14947 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
14948 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
14949 waitpkgintrin.h and cldemoteintrin.h.
14951 2018-06-04 Richard Biener <rguenther@suse.de>
14953 PR tree-optimization/86038
14954 * tracer.c (find_best_successor): Check probability for
14955 being initialized, bail out if not.
14957 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
14960 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
14961 of bits to ignore when comparing architectures.
14963 2018-06-04 Jakub Jelinek <jakub@redhat.com>
14965 PR tree-optimization/69615
14966 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
14967 maximum or minimum of the type, try to merge it also as if
14968 range1 is + [-, x - 1] or + [x + 1, -].
14971 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
14973 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
14975 PR tree-optimization/86034
14976 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
14977 the unsigned bitfield type in a bit insertion sequence if it does not
14978 have a larger precision than the bitfield size.
14979 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
14981 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
14983 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
14985 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
14987 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
14988 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
14989 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
14990 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
14992 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
14994 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
14995 Disable -fdelete-null-pointer-checks for ELF toolchain.
14997 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
14998 Kito Cheng <kito.cheng@gmail.com>
15000 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
15001 (nds32le-*-*, nds32be-*-*): Integrate checking process.
15002 (nds32*-*-*): Add glibc and uclibc conditions.
15003 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
15004 (TARGET_EXCEPT_UNWIND_INFO): Define.
15005 * config/nds32/elf.h: New file.
15006 * config/nds32/linux.h: New file.
15007 * config/nds32/nds32-elf.opt: New file.
15008 * config/nds32/nds32-linux.opt: New file.
15009 * config/nds32/nds32-fp-as-gp.c
15010 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
15011 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
15013 (nds32_asm_file_end): Ditto.
15014 (nds32_print_operand): Ditto.
15015 (nds32_insert_attributes): Ditto.
15016 (nds32_init_libfuncs): New function.
15017 (TARGET_HAVE_TLS): Define.
15018 (TARGET_INIT_LIBFUNCS): Define.
15019 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
15021 (TARGET_ELF): Apply different mcmodel setting.
15022 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
15023 been migrated into elf.h and linux.h files.
15024 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
15025 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
15026 (mcmodel): The content has been migrated into nds32-elf.opt and
15027 nds32-linux.opt files.
15028 * config/nds32/t-elf: New file.
15029 * config/nds32/t-linux: New file.
15031 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
15032 Shiva Chen <shiva0217@gmail.com>
15034 * config/nds32/constants.md (unspec_volatile_element): Add
15035 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
15036 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
15038 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
15039 (make_pass_nds32_fp_as_gp): Declare.
15040 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
15042 (nds32_asm_function_end_prologue): Remove unused asm output.
15043 (nds32_asm_function_begin_epilogue): Remove unused asm output.
15044 (nds32_asm_file_start): Output necessary fp_as_gp information.
15045 (nds32_option_override): Adjust register usage.
15046 (nds32_expand_prologue): Consider fp_as_gp situation.
15047 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
15048 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
15051 (simple_return): Ditto.
15052 (omit_fp_begin): Output special directive for fp_as_gp.
15053 (omit_fp_end): Output special directive for fp_as_gp.
15054 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
15055 mforbid-fp-as-gp): New options.
15057 2018-06-01 Mark Wielaard <mark@klomp.org>
15059 * dwarf2out.c (dwarf2out_finish): Remove generation of
15060 DW_AT_loclists_base.
15062 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
15064 * gimple-ssa-store-merging.c: Include gimple-fold.h.
15065 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
15066 (struct merged_store_group): Add bit_insertion field.
15067 (dump_char_array): Use standard hexadecimal format.
15068 (merged_store_group::merged_store_group): Set bit_insertion to false.
15069 (merged_store_group::apply_stores): Use optimal buffer size. Deal
15070 with BIT_INSERT_EXPR stores. Move up code updating the mask and
15071 also print the mask in the dump file.
15072 (pass_store_merging::gate): Minor tweak.
15073 (imm_store_chain_info::coalesce_immediate): Fix wrong association
15074 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
15075 stores with INTEGER_CST stores.
15076 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
15077 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
15078 and use it throughout. Generate bit insertion sequences if need be.
15079 (pass_store_merging::process_store): Remove redundant condition.
15080 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
15082 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
15084 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
15085 the 128-bit floating point types. Fix function comment.
15087 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15089 * config/aarch64/aarch64-simd.md
15090 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
15092 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
15095 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
15097 PR tree-optimization/85989
15098 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
15100 (backprop::intersect_uses): Check it when deciding whether this
15101 is a backedge reference.
15102 (backprop::process_block): Add each phi to m_visited_phis
15103 after visiting it, then clear it at the end.
15105 2018-06-01 Richard Biener <rguenther@suse.de>
15107 * tree-vectorizer.h (vect_dr_stmt): New function.
15108 (vect_get_load_cost): Adjust.
15109 (vect_get_store_cost): Likewise.
15110 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
15111 Use vect_dr_stmt instead of DR_SMTT.
15112 (vect_record_base_alignments): Likewise.
15113 (vect_calculate_target_alignment): Likewise.
15114 (vect_compute_data_ref_alignment): Likewise and make static.
15115 (vect_update_misalignment_for_peel): Likewise.
15116 (vect_verify_datarefs_alignment): Likewise.
15117 (vector_alignment_reachable_p): Likewise.
15118 (vect_get_data_access_cost): Likewise. Pass down
15119 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
15120 (vect_get_peeling_costs_all_drs): Likewise.
15121 (vect_peeling_hash_get_lowest_cost): Likewise.
15122 (vect_enhance_data_refs_alignment): Likewise.
15123 (vect_find_same_alignment_drs): Likewise.
15124 (vect_analyze_data_refs_alignment): Likewise.
15125 (vect_analyze_group_access_1): Likewise.
15126 (vect_analyze_group_access): Likewise.
15127 (vect_analyze_data_ref_access): Likewise.
15128 (vect_analyze_data_ref_accesses): Likewise.
15129 (vect_vfa_segment_size): Likewise.
15130 (vect_small_gap_p): Likewise.
15131 (vectorizable_with_step_bound_p): Likewise.
15132 (vect_prune_runtime_alias_test_list): Likewise.
15133 (vect_analyze_data_refs): Likewise.
15134 (vect_supportable_dr_alignment): Likewise.
15135 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
15136 (vect_gen_prolog_loop_niters): Likewise.
15137 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
15138 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
15140 (vect_recog_mask_conversion_pattern): Likewise.
15141 (vect_try_gather_scatter_pattern): Likewise.
15142 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
15143 to vect_get_store_cost.
15144 (vect_get_store_cost): Get stmt_info instead of DR.
15145 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
15146 (vect_get_load_cost): Get stmt_info instead of DR.
15148 2018-06-01 Richard Biener <rguenther@suse.de>
15150 PR middle-end/86017
15151 * gimple-fold.c (var_decl_component_p): Also allow offsetted
15152 vars wrapped in MEM_REFs.
15154 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
15156 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15157 Fix subreg tests so that we only return a choice between
15158 GENERAL_REGS and FP_REGS if the original classes included both.
15160 2018-06-01 Richard Biener <rguenther@suse.de>
15163 * tree-ssa-structalias.c (get_function_part_constraint):
15164 Handle NULL fi->decl.
15165 (find_func_aliases_for_call): Properly handle indirect
15166 fi from direct call.
15167 (find_func_clobbers): Likewise.
15168 (ipa_pta_execute): Likewise.
15169 (create_variable_info_for): For functions that are ifunc_resolver
15170 resolve to a varinfo that contains the result of the resolver call.
15171 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
15174 2018-05-31 Michael Collison <michael.collison@arm.com>
15176 * config/aarch64/aarch64.md:
15177 (*fix_to_zero_extenddfdi2): New pattern.
15178 * gcc.target/aarch64/fix_extend1.c: New testcase.
15180 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
15182 PR middle-end/78809
15183 PR middle-end/83026
15184 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
15185 and BUILT_IN_STRNCMP_EQ.
15186 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
15187 BUILT_IN_STRNCMP_EQ.
15188 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
15189 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
15190 (gimple_fold_builtin): Likewise.
15191 * tree-ssa-strlen.c (compute_string_length): New function.
15192 (determine_min_obsize): New function.
15193 (handle_builtin_string_cmp): New function to handle calls to
15194 string compare functions.
15195 (strlen_optimize_stmt): Add handling to builtin string compare
15197 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15198 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
15199 * tree.c (build_common_builtin_nodes): Add new defines of
15200 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
15202 2018-05-31 Jakub Jelinek <jakub@redhat.com>
15205 * bb-reorder.c (pass_partition_blocks::gate): Return false for
15206 functions with naked attribute.
15208 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
15210 * config/i386/sse.md (avx_vec_concat<mode>):
15211 Substitute concat_tg_mode mode attribute with xtg_mode.
15212 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
15213 (concat_tg_mode): Remove mode attribute.
15215 2018-05-31 Martin Sebor <msebor@redhat.com>
15218 * calls.c (alloc_max_size): Correct a logic error/typo.
15219 Treat excessive arguments as infinite. Warn for invalid arguments.
15220 * doc/invoke.texi (-Walloc-size-larger-than): Update.
15222 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
15225 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
15226 and movx for Haswell.
15228 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
15229 Cesar Philippidis <cesar@codesourcery.com>
15231 PR middle-end/85879
15232 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
15233 when emitting error on private/firstprivate reductions.
15234 * omp-low.c (lower_omp_target): Avoid reference-type processing
15235 on pointers for firstprivate clause.
15237 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
15239 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
15242 * config/aarch64/aarch64-simd.md
15243 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
15244 (aarch64_ld1_x3_<mode>): Likewise
15245 (aarch64_st1x2<VALLDIF:mode>): Likewise
15246 (aarch64_st1_x2_<mode>): Likewise
15247 (aarch64_st1x3<VALLDIF:mode>): Likewise
15248 (aarch64_st1_x3_<mode>): Likewise
15249 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
15250 (vld1_s8_x3): Likewise.
15251 (vld1_u16_x3): Likewise.
15252 (vld1_s16_x3): Likewise.
15253 (vld1_u32_x3): Likewise.
15254 (vld1_s32_x3): Likewise.
15255 (vld1_u64_x3): Likewise.
15256 (vld1_s64_x3): Likewise.
15257 (vld1_f16_x3): Likewise.
15258 (vld1_f32_x3): Likewise.
15259 (vld1_f64_x3): Likewise.
15260 (vld1_p8_x3): Likewise.
15261 (vld1_p16_x3): Likewise.
15262 (vld1_p64_x3): Likewise.
15263 (vld1q_u8_x3): Likewise.
15264 (vld1q_s8_x3): Likewise.
15265 (vld1q_u16_x3): Likewise.
15266 (vld1q_s16_x3): Likewise.
15267 (vld1q_u32_x3): Likewise.
15268 (vld1q_s32_x3): Likewise.
15269 (vld1q_u64_x3): Likewise.
15270 (vld1q_s64_x3): Likewise.
15271 (vld1q_f16_x3): Likewise.
15272 (vld1q_f32_x3): Likewise.
15273 (vld1q_f64_x3): Likewise.
15274 (vld1q_p8_x3): Likewise.
15275 (vld1q_p16_x3): Likewise.
15276 (vld1q_p64_x3): Likewise.
15277 (vst1_s64_x2): Likewise.
15278 (vst1_u64_x2): Likewise.
15279 (vst1_f64_x2): Likewise.
15280 (vst1_s8_x2): Likewise.
15281 (vst1_p8_x2): Likewise.
15282 (vst1_s16_x2): Likewise.
15283 (vst1_p16_x2): Likewise.
15284 (vst1_s32_x2): Likewise.
15285 (vst1_u8_x2): Likewise.
15286 (vst1_u16_x2): Likewise.
15287 (vst1_u32_x2): Likewise.
15288 (vst1_f16_x2): Likewise.
15289 (vst1_f32_x2): Likewise.
15290 (vst1_p64_x2): Likewise.
15291 (vst1q_s8_x2): Likewise.
15292 (vst1q_p8_x2): Likewise.
15293 (vst1q_s16_x2): Likewise.
15294 (vst1q_p16_x2): Likewise.
15295 (vst1q_s32_x2): Likewise.
15296 (vst1q_s64_x2): Likewise.
15297 (vst1q_u8_x2): Likewise.
15298 (vst1q_u16_x2): Likewise.
15299 (vst1q_u32_x2): Likewise.
15300 (vst1q_u64_x2): Likewise.
15301 (vst1q_f16_x2): Likewise.
15302 (vst1q_f32_x2): Likewise.
15303 (vst1q_f64_x2): Likewise.
15304 (vst1q_p64_x2): Likewise.
15305 (vst1_s64_x3): Likewise.
15306 (vst1_u64_x3): Likewise.
15307 (vst1_f64_x3): Likewise.
15308 (vst1_s8_x3): Likewise.
15309 (vst1_p8_x3): Likewise.
15310 (vst1_s16_x3): Likewise.
15311 (vst1_p16_x3): Likewise.
15312 (vst1_s32_x3): Likewise.
15313 (vst1_u8_x3): Likewise.
15314 (vst1_u16_x3): Likewise.
15315 (vst1_u32_x3): Likewise.
15316 (vst1_f16_x3): Likewise.
15317 (vst1_f32_x3): Likewise.
15318 (vst1_p64_x3): Likewise.
15319 (vst1q_s8_x3): Likewise.
15320 (vst1q_p8_x3): Likewise.
15321 (vst1q_s16_x3): Likewise.
15322 (vst1q_p16_x3): Likewise.
15323 (vst1q_s32_x3): Likewise.
15324 (vst1q_s64_x3): Likewise.
15325 (vst1q_u8_x3): Likewise.
15326 (vst1q_u16_x3): Likewise.
15327 (vst1q_u32_x3): Likewise.
15328 (vst1q_u64_x3): Likewise.
15329 (vst1q_f16_x3): Likewise.
15330 (vst1q_f32_x3): Likewise.
15331 (vst1q_f64_x3): Likewise.
15332 (vst1q_p64_x3): Likewise.
15334 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15336 * config/msp430/msp430.c (msp430_output_labelref): Prepend
15337 user_label_prefix to name.
15339 * tree-core.h: Update comment about the format of NAME string
15340 passed to handler in attribute_spec.
15342 * config/msp430/msp430.md: Remove erroneous subreg expression from
15343 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
15344 zero_extend{q,h}isi2.
15346 2018-05-30 Borislav Petkov <bp@suse.de>
15348 * doc/extend.texi: Document some architecture specific
15349 constraints and sort entries.
15351 2018-05-30 Martin Sebor <msebor@redhat.com>
15353 PR middle-end/85369
15354 * builtins.c (expand_builtin_stpcpy_1): New function.
15355 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
15356 only if the former succeeds.
15358 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
15360 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
15363 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15365 * doc/invoke.texi (-flinker-output): Document
15367 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15369 * passes.c (ipa_write_summaries): Only modify statements if body
15371 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
15372 incrementally linking.
15373 (ipa_passes): Likewise.
15374 * lto-cgraph.c (lto_output_node): When incrementally linking do not
15375 pass down resolution info.
15376 * common.opt (flag_incremental_link): Update info.
15377 * gcc.c (plugin specs): Turn flinker-output=* to
15378 -plugin-opt=-linker-output-known
15379 * toplev.c (compile_file): Also cut compilation when doing incremental
15381 * flag-types. (enum lto_partition_model): Add
15382 LTO_LINKER_OUTPUT_NOLTOREL.
15383 (invoke.texi): Add -flinker-output docs.
15384 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
15385 link same way as WPA; do not stream in dead initializers.
15387 * dwarf2out.c (dwarf2out_die_ref_for_decl,
15388 darf2out_register_external_decl): Support incremental link.
15390 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15392 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
15394 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
15396 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
15397 it down to simple_object_copy_lto_debug_sections.
15398 (run_gcc): Determine incremental LTO link time and configure
15399 lto1 into non-wpa mode, disable renaming of debug sections.
15401 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
15403 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
15404 descriptions of various incorrectly documented functions.
15406 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
15409 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
15410 address check not strict.
15412 2018-05-30 Richard Biener <rguenther@suse.de>
15414 PR tree-optimization/85964
15415 * tracer.c (better_p): Drop initialized count check, we only
15416 call the function with initialized counts now.
15417 (find_best_successor): Do find a best edge if one
15418 has uninitialized count.
15419 (find_best_predecessor): Likewise. Do BB frequency check only
15420 if count is initialized.
15422 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
15424 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
15425 (aarch64_ldrstr_offset_compare): New.
15426 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
15427 load/store orderings.
15428 (aarch64_gen_adjusted_ldpstp): Likewise.
15430 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
15432 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
15433 Check for subset of GENERAL_REGS and FP_REGS.
15434 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
15437 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
15439 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
15440 and wi::to_poly_offset. Add the current offset and then check
15441 whether the sum fits, rather than using an unchecked addition of
15442 a checked term. Check for a shwi rather than a uhwi.
15443 * expr.c (get_bit_range): Use tree_to_poly_uint64.
15444 (store_constructor): Use poly_int_tree_p.
15445 (expand_expr_real_1): Likewise.
15446 * function.c (assign_temp): Likewise.
15447 * fold-const.c (const_binop): Use poly_int_tree_p and
15448 wi::to_poly_offset.
15449 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
15451 * ipa-icf-gimple.c (func_checker::compare_operand): Use
15452 to_poly_offset for MEM offsets.
15453 * ipa-icf.c (sem_variable::equals): Likewise.
15454 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
15455 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
15456 wi::to_poly_offset for BIT_FIELD_REF offsets.
15457 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
15458 wi::to_poly_offset.
15459 * var-tracking.c (emit_note_insn_var_location): Use
15460 tree_to_poly_uint64.
15462 2018-05-29 Jim Wilson <jimw@sifive.com>
15464 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
15466 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
15469 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
15470 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
15472 (sse4_1_round<mode>2): Use nonimmediate_operand
15473 for operand 1 predicate.
15475 2018-05-29 Martin Sebor <msebor@redhat.com>
15476 Richard Biener <rguenther@suse.de>
15479 * calls.c (get_size_range): Call determine_value_range instead
15480 of get_value_range..
15481 * tree-vrp.h (determine_value_range): Declared new function.
15482 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
15484 2018-05-29 Richard Biener <rguenther@suse.de>
15486 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
15487 sure to use non-pattern stmts for get_earlier_stmt arguments.
15488 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
15489 called on pattern stmts.
15490 (get_later_stmt): Likewise.
15492 2018-05-29 Martin Liska <mliska@suse.cz>
15494 PR gcov-profile/85759
15495 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
15498 2018-05-29 Jakub Jelinek <jakub@redhat.com>
15500 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
15502 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
15506 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15507 VEC_PACK_FLOAT_EXPR): New tree codes.
15508 * tree-pretty-print.c (op_code_prio): Handle
15509 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15510 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15511 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15512 * tree-inline.c (estimate_operator_cost): Likewise.
15513 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
15514 * fold-const.c (const_binop): Likewise.
15515 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
15516 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
15517 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
15518 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
15519 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
15520 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
15521 * expr.c (expand_expr_real_2): Likewise.
15522 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
15523 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
15524 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
15526 * optabs.c (expand_widen_pattern_expr): For
15527 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
15528 sign from result type rather than operand's type.
15529 (expand_binop_directly): For vec_packu_float_optab and
15530 vec_packs_float_optab allow result type to be different from operand's
15532 * optabs-tree.c (optab_for_tree_code): Handle
15533 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15534 VEC_PACK_FLOAT_EXPR. Formatting fixes.
15535 * tree-vect-generic.c (expand_vector_operations_1): Handle
15536 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
15537 VEC_PACK_FLOAT_EXPR.
15538 * tree-vect-stmts.c (supportable_widening_operation): Handle
15540 (supportable_narrowing_operation): Handle FLOAT_EXPR.
15541 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
15542 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
15543 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
15544 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
15546 (vec_pack<floatprefix>_float_<mode>): New expander.
15547 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
15549 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
15550 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
15551 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
15552 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
15553 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
15555 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
15556 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
15557 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
15558 VEC_PACK_FLOAT_EXPR): Document.
15560 2018-05-29 Richard Biener <rguenther@suse.de>
15562 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
15564 (stmt_vec_info_vec): Make pointer.
15565 (init_stmt_vec_info_vec): Remove.
15566 (free_stmt_vec_info_vec): Likewise.
15567 (set_stmt_vec_info_vec): New function.
15568 (free_stmt_vec_infos): Likewise.
15569 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
15570 (set_vinfo_for_stmt): Likewise.
15571 (get_earlier_stmt): Likewise.
15572 (get_later_stmt): Likewise.
15573 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
15574 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
15575 (vec_info::~vec_info): Free stmt_vec_infos.
15576 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
15577 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
15578 (pass_slp_vectorize::execute): Likewise.
15579 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
15580 (free_stmt_vec_info_vec): Likewise.
15581 (set_stmt_vec_info_vec): New function.
15582 (free_stmt_vec_infos): Likewise.
15583 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
15584 the global stmt_vec_info_vec.
15585 * tree-parloops.c (gather_scalar_reductions): Use
15586 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
15589 2018-05-29 Richard Biener <rguenther@suse.de>
15591 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
15593 2018-05-29 Martin Liska <mliska@suse.cz>
15594 David Malcolm <dmalcolm@redhat.com>
15596 * vec.c (test_reverse): New.
15597 (vec_c_tests): Add new test.
15598 * vec.h (vl_ptr>::reverse): New function.
15600 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
15602 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
15604 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
15607 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15609 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
15611 2018-05-28 Richard Biener <rguenther@suse.de>
15613 PR tree-optimization/85933
15614 * tree-vect-data-refs.c (vect_record_base_alignments): Only
15615 look at stmts marked as vectorizable.
15617 2018-05-28 Richard Biener <rguenther@suse.de>
15619 PR tree-optimization/85934
15620 * tree-vect-generic.c (expand_vector_operations_1): Hoist
15621 vector boolean check before scalar optimization.
15623 2018-05-28 Jakub Jelinek <jakub@redhat.com>
15625 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
15628 2018-05-28 Mark Wielaard <mark@klomp.org>
15630 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
15631 if it is an expression containing a minus sign.
15633 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
15635 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
15637 2018-05-27 Paul Koning <ni1d@arrl.net>
15639 * config/pdp11/pdp11.md (truncsihi2): Remove.
15641 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
15642 Chung-Ju Wu <jasonwucj@gmail.com>
15644 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
15646 (unaligned_store_dw): Ditto.
15647 * config/nds32/nds32-memory-manipulation.c
15648 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
15649 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
15650 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
15651 (emit_setmem_word_loop): Rename to ...
15652 (emit_setmem_doubleword_loop): ... this.
15653 (nds32_gen_dup_4_byte_to_word_value): New function.
15654 (nds32_gen_dup_8_byte_to_double_word_value): New function.
15655 (nds32_expand_setmem_loop): Refine implementation.
15656 (nds32_expand_setmem_loop_v3m): Ditto.
15657 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
15660 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15662 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
15664 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
15666 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
15667 (nds32_init_machine_status): Initialize machine->attr_naked_p and
15668 machine->attr_no_prologue_p.
15669 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
15670 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
15671 (nds32_expand_epilogue): Consider attr_naked_p.
15672 (nds32_expand_epilogue_v3pop): Likewise.
15673 (nds32_can_use_return_insn): Likewise.
15674 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
15675 attr_no_prologue_p fields.
15676 * config/nds32/nds32.opt (mret-in-naked-func): New option.
15678 2018-05-27 Jakub Jelinek <jakub@redhat.com>
15681 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
15683 * config/i386/sse.md
15684 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
15686 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
15688 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
15690 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
15692 (*<floatsuffix>floatv2div2sf2): Rename to ...
15693 (*float<floatunssuffix>v2div2sf2): ... this.
15694 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
15695 (float<floatunssuffix>v2div2sf2_mask): ... this.
15696 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
15697 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
15698 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
15700 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
15702 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15704 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
15706 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15708 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
15710 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
15711 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
15712 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
15713 gen_ufix_truncv8dfv8si2.
15714 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
15715 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
15716 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
15717 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
15718 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
15719 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
15720 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
15721 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
15723 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
15727 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
15729 2018-05-25 Jim Wilson <jimw@sifive.com>
15731 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
15732 * config/riscv/riscv.c (struct machine_function): Add
15733 interrupt_handler_p and attribute_checked_p fields.
15734 (riscv_attribute_table): Add interrupt.
15735 (riscv_interrupt_type_p): New.
15736 (riscv_save_reg_p): Save extra regs for interrupt handler.
15737 (riscv_use_save_libcall): Return false for interrupt handler.
15738 (riscv_first_stack_step): Add forward declaration.
15739 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
15740 for interrupt handler with large frame. Use it for saved reg list.
15741 (riscv_expand_prologue): Move flag_stack_usage_info support to
15742 eliminate duplication.
15743 (riscv_expand_epilogue): Generate mret for interrupt handler.
15744 (riscv_epilogue_uses): New.
15745 (riscv_can_use_return_insn): Return false for interrupt handler.
15746 (riscv_function_ok_for_sibcall): Likewise.
15747 (riscv_set_current_function): Add interrupt handler support.
15748 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
15749 * config/riscv/riscv.md (UNSPECV_MRET): New.
15751 (riscv_frflags, riscv_fsflags): Use tab after opcode.
15753 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
15755 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
15757 PR tree-optimization/85712
15758 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
15759 this candidate has already been replaced in-situ by a copy.
15761 2018-05-25 Jason Merrill <jason@redhat.com>
15763 PR c++/80485 - inline function non-zero address.
15764 * symtab.c (nonzero_address): Check DECL_COMDAT.
15766 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
15769 * config/alpha/alpha.md (ashlsi3): New insn pattern.
15770 (*ashlsi_se): Rename from *ashldi_se. Define as sign
15771 extension of SImode operation. Use const123_operand predicate.
15772 (*saddsi_1): Remove.
15773 (*saddl_se_1): Ditto.
15774 (*ssubsi_1): Ditto.
15775 (*ssubl_se_1): Ditto.
15776 * config/alpha/predicates.md (const123_operand): New predicate.
15777 * config/alpha/constraints.md (P): Use IN_RANGE.
15779 2018-05-25 Richard Biener <rguenther@suse.de>
15781 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
15783 (ref_maybe_used_by_stmt_p): Likewise.
15784 (stmt_may_clobber_ref_p): Likewise.
15785 (stmt_may_clobber_ref_p_1): Likewise.
15786 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
15788 (ref_maybe_used_by_stmt_p): Likewise.
15789 (stmt_may_clobber_ref_p): Likewise.
15790 (stmt_may_clobber_ref_p_1): Likewise.
15791 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
15792 the alias oracle to disambiguate DRs with stmts DR analysis
15794 (vect_analyze_data_refs): Do not give up on not analyzable
15795 DRs for BB vectorization. Remove code truncating the dataref
15798 2018-05-25 Jakub Jelinek <jakub@redhat.com>
15801 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
15802 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
15803 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
15805 2018-05-25 Richard Biener <rguenther@suse.de>
15807 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
15808 function, combining stmt data ref gathering and fatal analysis
15810 (vect_analyze_data_refs): Remove now redudnant code and simplify.
15811 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
15812 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
15813 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
15814 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
15816 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15818 PR tree-optimization/85720
15819 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
15820 SCC if all partitions are builtins.
15821 (version_loop_by_alias_check): New parameter. Generate cancelable
15822 runtime alias check if all partitions are builtins.
15823 (distribute_loop): Update call to above function.
15825 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15827 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
15828 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
15829 (parm_default_def_partition_arg): Ditto.
15830 (set_parm_default_def_partition): Ditto.
15831 (get_parm_default_def_partitions): Ditto and make it static.
15832 (get_undefined_value_partitions): Ditto and make it static.
15833 (remove_ssa_form): Refactor call to init_var_map here.
15834 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
15835 computation for loop region.
15836 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
15837 (register_default_def): Delete.
15838 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
15839 (parm_default_def_partition_arg): Ditto.
15840 (set_parm_default_def_partition): Ditto.
15841 (get_parm_default_def_partitions): Ditto and make it static.
15842 (get_undefined_value_partitions): Ditto and make it static.
15843 (coalesce_with_default, coalesce_with_default): Update comment.
15844 (create_coalesce_list_for_region): New func factored out from
15845 create_outofssa_var_map.
15846 (populate_coalesce_list_for_outofssa): New func factored out from
15847 create_outofssa_var_map and coalesce_ssa_name.
15848 (create_outofssa_var_map): Delete.
15849 (coalesce_ssa_name): Refactor to support live range computation.
15850 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
15851 (get_parm_default_def_partitions): Delete.
15852 (get_undefined_value_partitions): Ditto.
15853 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
15854 computation for loop region.
15855 (new_tree_live_info, loe_visit_block): Ditto.
15856 (live_worklist, set_var_live_on_entry): Ditto.
15857 (calculate_live_on_exit, verify_live_on_entry): Ditto.
15858 * tree-ssa-live.h (struct _var_map): New fields.
15859 (init_var_map): Change decl.
15860 (region_contains_p): New.
15862 2018-05-25 Bin Cheng <bin.cheng@arm.com>
15864 * tree-ssa-live.h (live_merge_and_clear): Delete.
15866 2018-05-25 Richard Biener <rguenther@suse.de>
15869 * tree-dump.c (dequeue_and_dump): Remove access to removed
15870 operand 2 of a SWITCH_EXPR.
15872 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15874 * doc/sourcebuild.texi (vect_double_cond_arith): Include
15875 multiplication and division.
15876 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
15877 (cond_udiv@var{m}, cond_umod@var{m}): Document.
15878 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
15879 (cond_udiv_optab, cond_umod_optab): New optabs.
15880 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
15881 (IFN_COND_RDIV): New internal functions.
15882 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
15883 TRUNC_MOD_EXPR and RDIV_EXPR.
15884 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
15885 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
15887 (SVE_INT_BINARY): Include mult.
15888 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
15889 (optab, sve_int_op): Handle mult.
15890 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
15892 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
15893 for SVE_INT_BINARY_SD.
15895 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15897 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
15898 (optab, sve_int_op): Handle div and udiv.
15899 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
15900 for SVE_INT_BINARY_SD.
15901 (*<optab><mode>3): New insn for the same.
15903 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15905 * tree-vect-patterns.c: Include predict.h.
15906 (vect_recog_divmod_pattern): Restrict check for division support
15907 to when optimizing for size.
15909 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15911 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
15912 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
15913 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
15914 (gimple_match_op::set_op): Likewise.
15915 (gimple_resimplify4): Declare.
15916 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
15917 (expr::gen_transform): Likewise.
15918 (decision_tree::gen): Generate a simplification routine for 4 operands.
15919 * gimple-match-head.c (gimple_simplify): Add an overload for
15920 4 operands. In the top-level function, handle up to 4 call
15921 arguments and call gimple_resimplify4.
15922 (gimple_resimplify4): New function.
15923 (build_call_internal): Pass a fourth operand.
15924 (maybe_push_to_seq): Likewise.
15925 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
15926 Fold VEC_COND_EXPRs of an operation and a default value into
15927 an IFN_COND_* function if possible.
15928 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
15930 (SVE_COND_FP_BINARY): Include them.
15931 (optab, sve_fp_op): Handle them.
15932 (SVE_INT_BINARY_REV): New code iterator.
15933 (SVE_COND_FP_BINARY_REV): New int iterator.
15934 (commutative): New int attribute.
15935 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
15937 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
15939 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
15940 (*cond_<optab><mode>): New patterns for reversed operands.
15942 2018-05-25 Richard Biener <rguenther@suse.de>
15944 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
15945 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
15946 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
15947 (STMT_VINFO_GROUPED_ACCESS): Adjust.
15948 * tree-vect-data-refs.c (everywhere): Adjust users.
15949 * tree-vect-loop.c (everywhere): Likewise.
15950 * tree-vect-slp.c (everywhere): Likewise.
15951 * tree-vect-stmts.c (everywhere): Likewise.
15952 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
15954 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15956 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
15958 (gcc_cv_as_section_exclude): ... this.
15959 Try Solaris as #exclude syntax.
15960 * configure: Regenerate.
15961 * config.in: Regenerate.
15962 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
15964 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
15965 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
15967 * varasm.c (default_elf_asm_named_section): Don't check if
15968 HAVE_GAS_SECTION_EXCLUDE is defined.
15970 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
15972 * doc/md.texi: Update the documentation of the cond_* optabs
15973 to mention the new final operand. Fix GET_MODE_NUNITS call.
15974 Describe the scalar case too.
15975 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
15976 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
15978 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
15979 (get_conditional_internal_fn): Update comment.
15980 * tree-vect-loop.c (vectorizable_reduction): Pass the original
15981 accumulator value as a final argument to conditional functions.
15982 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
15983 a define_expand and add an "else" operand. Assert for now that
15984 the else operand is equal to operand 2. Use SVE_INT_BINARY and
15985 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
15986 (*cond_<optab><mode>): New patterns.
15987 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
15988 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
15989 (UNSPEC_COND_EOR): Delete.
15990 (optab): Remove associated mappings.
15991 (SVE_INT_BINARY): New code iterator.
15992 (sve_int_op): Remove int attribute and add "minus" to the code
15994 (SVE_COND_INT_OP): Delete.
15995 (SVE_COND_FP_OP): Rename to...
15996 (SVE_COND_FP_BINARY): ...this.
15998 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
16000 * optabs.c (can_reuse_operands_p): New function.
16001 (maybe_legitimize_operands): Try to reuse the results for
16004 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
16006 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
16007 Add {q} suffix to insn mnemonic.
16009 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16011 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
16012 (msp430_warn_func_return): New.
16014 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
16016 * fold-const.c (tree_nonzero_bits): New function.
16017 * fold-const.h (tree_nonzero_bits): Likewise.
16018 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
16019 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
16021 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
16025 * varasm.c (assemble_alias): Check ifunc_resolver only on
16028 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
16031 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
16032 when memory input operand is handled.
16034 2018-05-24 Luis Machado <luis.machado@linaro.org>
16036 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
16038 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
16040 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
16042 * match.pd: Delay FMA folds until after vectorization.
16044 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
16047 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
16048 address check not strict.
16050 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
16052 * gimple-match.h (gimple_match_op): New class.
16053 (mprts_hook): Replace parameters with a gimple_match_op *.
16054 (maybe_build_generic_op): Likewise.
16055 (gimple_simplified_result_is_gimple_val): Replace parameters with
16056 a const gimple_match_op *.
16057 (gimple_simplify): Replace code_helper * and tree * parameters with
16058 a gimple_match_op * parameter.
16059 (gimple_resimplify1): Replace code_helper *, tree and tree *
16060 parameters with a gimple_match_op * parameter.
16061 (gimple_resimplify2): Likewise.
16062 (gimple_resimplify3): Likewise.
16063 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
16064 parameters with a gimple_match_op * parameter.
16065 * gimple-match-head.c (gimple_simplify): Change prototypes of
16066 auto-generated functions to take a gimple_match_op * instead of
16067 separate code_helper * and tree * parameters. Make the same
16068 change in the top-level overload and update calls to the
16069 gimple_resimplify routines. Update calls to the auto-generated
16070 functions and to maybe_push_res_to_seq in the publicly-facing
16071 operation-specific gimple_simplify overloads.
16072 (gimple_match_op::MAX_NUM_OPS): Define.
16073 (gimple_resimplify1): Replace rcode and ops with a single res_op
16074 parameter. Update call to gimple_simplify.
16075 (gimple_resimplify2): Likewise.
16076 (gimple_resimplify3): Likewise.
16077 (mprts_hook): Replace parameters with a gimple_match_op *.
16078 (maybe_build_generic_op): Likewise.
16079 (build_call_internal): Replace type, nargs and ops with
16080 a gimple_match_op *.
16081 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
16082 with a single gimple_match_op *. Update calls to mprts_hook,
16083 build_call_internal and gimple_simplified_result_is_gimple_val.
16084 Factor out code that is common to the tree_code and combined_fn cases.
16085 * genmatch.c (expr::gen_transform): Replace tem_code and
16086 tem_ops with a gimple_match_op called tem_op. Update calls
16087 to the gimple_resimplify functions and maybe_push_res_to_seq.
16088 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
16089 res_ops. Update call to the gimple_resimplify functions.
16090 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
16091 (decision_tree::gen): Make the functions take a gimple_match_op *
16092 called res_op instead of separate res_code and res_ops parameters.
16093 Update call accordingly.
16094 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
16095 and ops with a single res_op parameter. Update calls to
16096 maybe_build_generic_op and maybe_push_res_to_seq.
16097 (fold_stmt_1): Update calls to gimple_simplify and
16098 replace_stmt_with_simplification.
16099 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
16100 and gimple_simplified_result_is_gimple_val.
16101 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
16103 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
16104 with a gimple_match_op *.
16105 (vn_nary_build_or_lookup): Likewise. Update call to
16106 vn_nary_build_or_lookup_1.
16107 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
16108 gimple_match_op *. Update calls to the gimple_resimplify routines
16109 and to gimple_simplified_result_is_gimple_val.
16110 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
16111 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
16112 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
16113 (visit_nary_op): Likewise.
16114 (visit_reference_op_load): Likewise.
16116 2018-05-23 Luis Machado <luis.machado@linaro.org>
16118 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
16119 modifier for printing the step amount.
16121 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16124 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
16127 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
16129 * doc/sourcebuild.texi (Endianness): New subsubsection.
16131 2018-05-23 Luis Machado <luis.machado@linaro.org>
16133 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16134 <prefetch_dynamic_strides>: New const bool field.
16135 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16136 prefetch_dynamic_strides.
16137 (exynosm1_prefetch_tune): Likewise.
16138 (thunderxt88_prefetch_tune): Likewise.
16139 (thunderx_prefetch_tune): Likewise.
16140 (thunderx2t99_prefetch_tune): Likewise.
16141 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
16143 (aarch64_override_options_internal): Update to set
16144 PARAM_PREFETCH_DYNAMIC_STRIDES.
16145 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
16146 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
16147 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
16148 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
16149 prefetch-dynamic-strides setting.
16151 2018-05-23 Luis Machado <luis.machado@linaro.org>
16153 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
16154 <minimum_stride>: New const int field.
16155 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
16156 minimum_stride field defaulting to -1.
16157 (exynosm1_prefetch_tune): Likewise.
16158 (thunderxt88_prefetch_tune): Likewise.
16159 (thunderx_prefetch_tune): Likewise.
16160 (thunderx2t99_prefetch_tune): Likewise.
16161 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
16162 <default_opt_level>: Set to 3.
16163 (aarch64_override_options_internal): Update to set
16164 PARAM_PREFETCH_MINIMUM_STRIDE.
16165 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
16166 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
16167 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
16168 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
16169 stride is constant and is below the minimum stride threshold.
16171 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16173 * config/arm/arm-cpus.in (mode26): Delete.
16174 (armv4): Delete mode26 reference.
16175 * config/arm/arm.c (arm_configure_build_target): Delete use of
16178 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
16180 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
16182 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
16183 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
16185 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
16186 (floatunsdidf2): Ditto.
16188 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
16190 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
16191 (fixuns_trunc<mode>si2_avx512f): Ditto.
16192 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
16193 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
16194 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
16196 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
16198 PR rtl-optimization/79985
16199 * df-scan.c (df_insn_refs_collect): Remove special case for
16200 global registers and asm statements.
16202 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
16204 * extend.texi (Global Register Variables): Rewrite the bullet list.
16205 Note that the register is available for allocation. Note that access
16206 via inline asm must use constraints. Add note about async-signal
16207 handlers. Remove paragraph about automagic register selection.
16209 2018-05-23 Richard Biener <rguenther@suse.de>
16211 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
16212 of fixed offset from memset VN.
16214 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
16216 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
16217 first_interp field.
16218 (alloc_cand_and_find_basis): Initialize first_interp field.
16219 (slsr_process_mul): Modify first_interp field.
16220 (slsr_process_add): Likewise.
16221 (slsr_process_cast): Modify first_interp field for each new
16223 (slsr_process_copy): Likewise.
16224 (dump_candidate): Dump first_interp field.
16225 (replace_mult_candidate): Process all interpretations, not just
16227 (replace_rhs_if_not_dup): Likewise.
16228 (replace_one_candidate): Likewise.
16230 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
16232 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
16234 (aarch64_needs_frame_chain): New function.
16235 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
16237 2018-05-23 Sudakshina Das <sudi.das@arm.com>
16240 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
16241 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
16242 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
16243 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
16244 as true for strict-align.
16245 (aarch64_can_inline_p): Perform checks even when callee has no
16246 attributes to check for strict alignment.
16247 * doc/extend.texi (AArch64 Function Attributes): Document
16249 * doc/invoke.texi: (AArch64 Options): Likewise.
16251 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
16253 PR tree-optimization/85853
16254 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
16255 the handling of the root of the node to...
16256 (vect_slp_analyze_node_operations_1): ...this new function,
16257 and run the whole thing with the child nodes' def types
16258 set according to their SLP node's def type.
16260 2018-05-23 Richard Biener <rguenther@suse.de>
16262 PR middle-end/85874
16263 * tree-data-ref.c (create_runtime_alias_checks): Defer
16264 and ignore overflow warnings.
16266 2018-05-23 Yury Gribov <tetra2005@gmail.com>
16268 PR tree-optimization/85822
16269 * tree-vrp.c (is_masked_range_test): Fix handling of negative
16272 2018-05-23 Richard Biener <rguenther@suse.de>
16274 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
16275 memset constants via native_interpret_expr.
16277 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
16280 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
16282 (cgraph_node::create_alias): Likewise.
16283 (cgraph_node::get_availability): Check ifunc_resolver instead
16284 of looking up ifunc attribute.
16285 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
16286 * varasm.c (do_assemble_alias): Likewise.
16287 (assemble_alias): Likewise.
16288 (default_binds_local_p_3): Likewise.
16289 * cgraph.h (cgraph_node): Add ifunc_resolver.
16290 (cgraph_node::only_called_directly_or_aliased_p): Return false
16291 for IFUNC resolver.
16292 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
16294 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
16295 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
16296 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
16297 instead of looking up ifunc attribute.
16299 2018-05-22 Luis Machado <luis.machado@linaro.org>
16301 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
16303 2018-05-22 Martin Sebor <msebor@redhat.com>
16305 PR middle-end/85359
16306 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
16307 only when expasion succeeds.
16308 (expand_builtin_strcmp): Same.
16309 (expand_builtin_strncmp): Same.
16311 2018-05-22 Martin Sebor <msebor@redhat.com>
16313 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
16315 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
16316 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16318 * config/aarch64/aarch64-ldpstp.md: Replace uses of
16319 aarch64_mem_pair_operand with memory_operand and delete operand swapping
16321 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
16322 Add check for legitimate_address.
16323 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
16324 (aarch64_swap_ldrstr_operands): New.
16325 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
16328 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
16329 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16331 * config/aarch64/aarch64.md: New patterns to generate stp
16333 (store_pair_sw, store_pair_dw): New patterns to generate stp for
16334 single words and double words.
16335 (load_pair_sw, load_pair_dw): Likewise.
16336 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
16338 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
16340 * config/aarch64/aarch64-ldpstp.md: Modify peephole
16341 for different mode ldpstp and add peephole for merged zero stores.
16342 Likewise for loads.
16343 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
16345 (aarch64_gen_store_pair): Rename calls to match new patterns.
16346 (aarch64_gen_load_pair): Rename calls to match new patterns.
16347 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
16348 (load_pair<DREG:mode><DREG2:mode>): ... This.
16349 (store_pair<mode>): Rename to...
16350 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
16351 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
16352 New mode iterators.
16353 (V_INT_EQUIV): Handle SImode.
16354 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
16357 2018-05-22 Martin Sebor <msebor@redhat.com>
16360 * calls.c (maybe_warn_nonstring_arg): Use string length to set
16361 or ajust the presumed bound on an operation to avoid unnecessary
16364 2018-05-22 Martin Sebor <msebor@redhat.com>
16366 PR tree-optimization/85826
16367 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
16368 assuming that a DECL necesarily has a constant size.
16370 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
16372 PR middle-end/85862
16373 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
16375 2018-05-22 Richard Biener <rguenther@suse.de>
16377 PR tree-optimization/85834
16378 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
16379 non-constant and non-zero memset arguments.
16381 2018-05-22 Martin Liska <mliska@suse.cz>
16384 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
16386 2018-05-22 Richard Biener <rguenther@suse.de>
16388 PR tree-optimization/85863
16389 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
16390 comparisons when vectype is specified.
16391 (vectorizable_condition): Do not specify vectype for
16392 vect_is_simple_cond when SLP vectorizing.
16394 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
16397 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
16398 define __ibm128 as long double.
16399 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
16400 as a distinct type when IEEE 128-bit support is enabled.
16401 (init_float128_ieee): Fix up conversions between IFmode and IEEE
16402 128-bit types to use the correct functions.
16403 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
16404 convert between 128-bit floating point types that have different
16405 modes but the same representation, instead of using gen_lowpart to
16407 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
16409 (IFKF_reg): New attributes to give the register constraints for
16411 (extend<mode>tf2_internal): New insns to mark an explicit
16412 conversion between 128-bit floating point types that have a
16413 different mode but share the same representation.
16415 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
16417 PR tree-optimization/85814
16418 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
16419 a null return from get_strinfo when unsharing the next
16420 strinfo in the chain.
16422 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
16425 * varasm.c (weak_finish): Clean up weak_decls.
16427 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16429 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
16430 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
16431 UNSPEC_UADALP values.
16432 * config/aarch64/iterators.md (ABAL): New int iterator.
16435 (sur): Add mappings for the above.
16436 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
16438 (aarch64_<sur>abal<mode>_4): Likewise.
16439 (aarch64_<sur>adalp<mode>_3): Likewise.
16440 (<sur>sadv16qi): New define_expand.
16442 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
16444 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
16445 (*movdf_internal): Ditto.
16446 (*rcpsf2_sse): Ditto.
16447 (*rsqrtsf2_sse): Ditto.
16448 (*sqrt<mode>2_sse): Ditto.
16450 2018-05-21 Tamar Christina <tamar.christina@arm.com>
16452 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
16454 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
16455 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
16456 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
16457 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
16459 * config/aarch64/arm_neon.h: Likewise.
16460 * config/aarch64/iterators.md (VQ_I): New.
16462 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
16464 * config.gcc: Add arc/t-multilib-linux to tmake_file for
16466 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
16469 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16471 * config/nds32/constraints.md (S): New constraint.
16472 * config/nds32/nds32.md (call_internal): Use constraint S.
16473 (call_value_internal): Likewise.
16474 (sibcall_internal): Likewise.
16475 (sibcall_value_internal): Likewise.
16477 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16478 Chung-Ju Wu <jasonwucj@gmail.com>
16480 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
16481 into consideration.
16483 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
16484 Chung-Ju Wu <jasonwucj@gmail.com>
16486 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
16487 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
16488 (nds32_rtx_costs_impl): Simplify.
16489 (nds32_address_cost_impl): Simplify.
16490 (nds32_init_rtx_costs): New function.
16491 (nds32_rtx_costs_speed_prefer): Likewise.
16492 (nds32_rtx_costs_size_prefer): Likewise.
16493 (nds32_address_cost_speed_prefer): Likewise.
16494 (nds32_address_cost_speed_fwprop): Likewise.
16495 (nds32_address_cost_size_prefer): Likewise.
16496 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
16497 * config/nds32/nds32.c (nds32_option_override): Use
16498 nds32_init_rtx_costs function.
16500 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
16502 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
16503 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
16504 (TARGET_PIPELINE_N8): Likewise.
16505 (TARGET_PIPELINE_N10): Likewise.
16506 (TARGET_PIPELINE_N13): Likewise.
16507 (TARGET_PIPELINE_GRAYWOLF): Likewise.
16509 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
16511 * config/nds32/nds32-fpu.md: Update copyright year.
16513 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16515 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
16517 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16519 * config/nds32/nds32.c
16520 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
16521 * config/nds32/nds32.opt (minline-asm-r15): New option.
16523 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16525 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
16527 * config/nds32/nds32.md (abssi2): New pattern.
16529 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
16531 * config/i386/i386.md (rex64namesuffix): New mode attribute.
16532 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
16533 Merge insn pattern from sse_cvtsi2ss<round_name> and
16534 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
16535 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
16536 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
16537 using SWI48 mode iterator.
16538 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
16539 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
16540 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
16541 pattern from sse_cvttss2si<round_saeonly_name>
16542 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
16543 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
16544 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
16545 using SWI48 mode iterator.
16546 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
16547 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
16548 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
16549 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
16550 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
16551 using SWI48 mode iterator.
16552 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
16553 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
16554 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
16555 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
16556 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
16557 SWI48 mode iterator.
16558 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
16559 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
16560 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
16561 pattern from sse_cvttsd2si<round_saeonly_name>
16562 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
16564 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16566 * config/nds32/nds32-md-auxiliary.c
16567 (nds32_valid_smw_lwm_base_p): Refine.
16568 (nds32_output_smw_single_word): Refine.
16569 (nds32_output_smw_double_word): New.
16570 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
16572 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16574 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
16575 (nds32_output_stack_pop): Refine.
16576 (nds32_expand_unaligned_load): Refine.
16577 (nds32_expand_unaligned_store): Refine.
16579 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16580 Chung-Ju Wu <jasonwucj@gmail.com>
16582 * config/nds32/constants.md: Add TP_REGNUM constant.
16583 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
16584 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
16586 * config/nds32/nds32-doubleword.md: Consider flag_pic.
16587 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
16588 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
16589 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
16590 and PIC code generation.
16591 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
16593 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
16595 * config/nds32/nds32.md: Support TLS and PIC.
16596 * config/nds32/nds32.c: Support TLS and PIC.
16597 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
16598 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
16601 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
16603 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
16604 mode with E_ prefix.
16606 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16607 Chung-Ju Wu <jasonwucj@gmail.com>
16609 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
16610 * config/nds32/nds32-md-auxiliary.c
16611 (symbolic_reference_mentioned_p): New.
16612 (nds32_legitimize_ict_address): New.
16613 (nds32_expand_ict_move): New.
16614 (nds32_indirect_call_referenced_p): New.
16615 (nds32_symbol_binds_local_p): Delete.
16616 (nds32_long_call_p): Modify.
16617 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
16618 * config/nds32/nds32-protos.h
16619 (symbolic_reference_mentioned_p): Declare.
16620 (nds32_legitimize_ict_address): Declare.
16621 (nds32_expand_ict_move): Declare.
16622 (nds32_indirect_call_referenced_p): Declare.
16623 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
16624 (nds32_relax_group): Use nds32_ict_const_p as condition.
16625 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
16626 (nds32_asm_file_start): Output ict_model directive in asm code.
16627 (nds32_legitimate_address_p): Consider indirect call.
16628 (nds32_print_operand): Consider indirect call.
16629 (nds32_print_operand_address): Consider indirect call.
16630 (nds32_insert_attributes): Handle "indirect_call" attribute.
16631 (TARGET_LEGITIMATE_ADDRESS_P): Define.
16632 (TARGET_LEGITIMATE_CONSTANT_P): Define.
16633 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
16634 (TARGET_DELEGITIMIZE_ADDRESS): Define.
16635 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
16636 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
16637 (TARGET_ICT_MODEL_SMALL): Define.
16638 (TARGET_ICT_MODEL_LARGE): Define.
16639 * config/nds32/nds32.md (movsi): Consider ict model.
16640 (call, call_value): Consider ict model.
16641 (sibcall, sibcall_value): Consider ict model.
16642 * config/nds32/nds32.opt (mict-model): New option.
16643 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
16646 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
16647 Monk Chiang <sh.chiang04@gmail.com>
16648 Jim Wilson <jimw@sifive.com>
16650 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
16651 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
16652 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
16653 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
16654 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
16655 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
16656 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
16657 compute save_libcall_adjustment properly.
16658 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
16659 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
16660 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
16661 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
16662 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
16663 (ABI_SPEC): Handle mabi=ilp32e.
16664 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
16665 (RVE): Add RVE mask.
16666 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
16667 <-march>: Add rv32e as an example.
16669 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
16672 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
16673 (intra_create_variable_infos): Handle C++ constructors.
16675 2018-05-18 Martin Liska <mliska@suse.cz>
16677 * passes.def: Remove a redundant pass.
16679 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
16682 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
16684 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16686 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
16688 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
16690 (armv2, armv2a, armv3, armv3m): Delete architectures.
16691 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
16692 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
16693 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
16695 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
16696 (*mulsidi3adddi): Likewise.
16697 (mulsidi3): Likewise.
16698 (*mulsidi3_nov6): Likewise.
16699 (umulsidi3): Likewise.
16700 (umulsidi3_nov6): Likewise.
16701 (umaddsidi4): Likewise.
16702 (*umulsidi3adddi): Likewise.
16703 (smulsi3_highpart): Likewise.
16704 (*smulsi3_highpart_nov6): Likewise.
16705 (umulsi3_highpart): Likewise.
16706 (*umulsi3_highpart_nov6): Likewise.
16707 * config/arm/arm.h (arm_arch3m): Delete.
16708 * config/arm/arm.c (arm_arch3m): Delete.
16709 (arm_option_override_internal): Update armv3-related comment.
16710 (arm_configure_build_target): Delete use of isa_bit_mode32.
16711 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
16712 (arm_rtx_costs_internal): Delete check of arm_arch3m.
16713 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
16714 (mulsa3): Likewise.
16715 (mulusa3): Likewise.
16716 * config/arm/arm-protos.h (arm_arch3m): Delete.
16717 * config/arm/arm-tables.opt: Regenerate.
16718 * config/arm/arm-tune.md: Likewise.
16719 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
16720 deleted architectures.
16722 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16724 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
16725 (armv5t, armv5te): New features.
16726 (ARMv5, ARMv5e): Delete fgroups.
16727 (ARMv5t, ARMv5te): Adjust for above changes.
16728 (ARMv6m): Likewise.
16729 (armv5, armv5e): Delete arches.
16730 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
16732 (*call_reg_arm): Likewise.
16733 (*call_value_reg_armv5): Likewise.
16734 (*call_value_reg_arm): Likewise.
16735 (*call_symbol): Likewise.
16736 (*call_value_symbol): Likewise.
16737 (*sibcall_insn): Likewise.
16738 (*sibcall_value_insn): Likewise.
16739 (clzsi2): Likewise.
16740 (prefetch): Likewise.
16741 (define_split and define_peephole2 dependent on arm_arch5):
16743 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
16745 (TARGET_ARM_QBIT): Likewise.
16746 (TARGET_DSP_MULTIPLY): Likewise.
16747 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
16748 (arm_arch5, arm_arch5e): Delete.
16749 (arm_arch5t, arm_arch5te): Declare.
16750 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
16751 (arm_arch5t): Declare.
16752 (arm_option_reconfigure_globals): Update for the above.
16753 (arm_options_perform_arch_sanity_checks): Update comment, replace
16754 use of arm_arch5 with arm_arch5t.
16755 (use_return_insn): Likewise.
16756 (arm_emit_call_insn): Likewise.
16757 (output_return_instruction): Likewise.
16758 (arm_final_prescan_insn): Likewise.
16759 (arm_coproc_builtin_available): Likewise.
16760 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
16761 arm_arch5e with arm_arch5t and arm_arch5te.
16762 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
16763 (arm_arch5t, arm_arch5te): Declare.
16764 * config/arm/arm-tables.opt: Regenerate.
16765 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
16766 * config/arm/t-multilib: Likewise.
16767 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
16768 instead of arm_arch5.
16769 (*call_reg_thumb1): Likewise.
16770 (*call_value_reg_thumb1_v5): Likewise.
16771 (*call_value_reg_thumb1): Likewise.
16772 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
16774 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
16776 2018-05-18 Martin Liska <mliska@suse.cz>
16778 PR gcov-profile/84846
16779 * doc/gcov.texi: Document -t option of gcov tool.
16781 2018-05-18 Martin Liska <mliska@suse.cz>
16783 PR gcov-profile/84846
16784 * gcov.c (print_usage): Add new -t option.
16785 (process_args): Handle the option.
16786 (generate_results): Use stdout as output when requested by
16789 2018-05-18 Martin Liska <mliska@suse.cz>
16791 PR gcov-profile/84846
16792 * coverage.c (coverage_init): Write PWD to .gcno file.
16793 * doc/gcov.texi: Document how working directory is printed.
16794 * gcov-dump.c (dump_gcov_file): Print PWD.
16795 * gcov.c (output_intermediate_file): Likewise.
16796 (read_graph_file): Read PWD string.
16797 (output_lines): Print PWD.
16799 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16801 PR middle-end/85817
16802 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
16803 for retval and return false if all args to phi are zero.
16805 2018-05-18 Richard Biener <rguenther@suse.de>
16807 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
16809 (evrp_dom_walker::before_dom_children): Call it.
16811 2018-05-18 Richard Biener <rguenther@suse.de>
16813 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
16814 results when processing array refs with variable index.
16816 2018-05-18 Toon Moene <toon@moene.org>
16818 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
16819 directly after that of -floop-interchange. Indicate that both
16820 options are enabled by default when specifying -O3.
16822 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16824 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
16825 iterator. Delete separate integer-mode vec_set<mode> expander.
16826 (aarch64_simd_vec_setv2di): Delete.
16827 (vec_setv2di): Delete.
16828 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
16829 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
16830 the "w, r" alternative.
16832 2018-05-18 Martin Liska <mliska@suse.cz>
16834 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
16835 * tree-pass.h (make_pass_lower_switch_O0): New function.
16836 * tree-switch-conversion.c (node_has_low_bound): Remove.
16837 (node_has_high_bound): Likewise.
16838 (node_is_bounded): Likewise.
16839 (class pass_lower_switch): Make it a template type and create
16841 (pass_lower_switch::execute): Add template argument.
16842 (make_pass_lower_switch): New function.
16843 (make_pass_lower_switch_O0): New function.
16844 (do_jump_if_equal): Remove.
16845 (emit_case_nodes): Simplify to just handle all 3 cases and leave
16846 all the hard work to tree optimization passes.
16848 2018-05-18 Martin Liska <mliska@suse.cz>
16850 * dbgcnt.c (limit_low): Renamed from limit.
16851 (limit_high): New variable.
16852 (dbg_cnt_is_enabled): Check for upper limit.
16853 (dbg_cnt): Adjust dumping.
16854 (dbg_cnt_set_limit_by_index): Add new argument for high
16856 (dbg_cnt_set_limit_by_name): Likewise.
16857 (dbg_cnt_process_single_pair): Parse new format.
16858 (dbg_cnt_process_opt): Use strtok.
16859 (dbg_cnt_list_all_counters): Remove 'value' and add
16861 * doc/invoke.texi: Document changes.
16863 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
16865 * doc/sourcebuild.texi (scalar_all_fma): Document.
16866 * tree.def (FMA_EXPR): Delete.
16867 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
16868 * internal-fn.c (ternary_direct): New macro.
16869 (expand_ternary_optab_fn): Likewise.
16870 (direct_ternary_optab_supported_p): Likewise.
16871 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
16872 * builtins.c (fold_builtin_fma): Delete.
16873 (fold_builtin_3): Don't call it.
16874 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
16875 * expr.c (expand_expr_real_2): Likewise.
16876 * fold-const.c (operand_equal_p): Likewise.
16877 (fold_ternary_loc): Likewise.
16878 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
16879 * gimple.c (DEFTREECODE): Likewise.
16880 * gimplify.c (gimplify_expr): Likewise.
16881 * optabs-tree.c (optab_for_tree_code): Likewise.
16882 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
16883 * tree-eh.c (operation_could_trap_p): Likewise.
16884 (stmt_could_throw_1_p): Likewise.
16885 * tree-inline.c (estimate_operator_cost): Likewise.
16886 * tree-pretty-print.c (dump_generic_node): Likewise.
16887 (op_code_prio): Likewise.
16888 * tree-ssa-loop-im.c (stmt_cost): Likewise.
16889 * tree-ssa-operands.c (get_expr_operands): Likewise.
16890 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
16891 * fold-const-call.h (fold_fma): Delete.
16892 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
16893 CFN_FNMA and CFN_FNMS.
16894 (fold_fma): Delete.
16895 * genmatch.c (combined_fn): New enum.
16896 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
16897 (commutative_op): New function.
16898 (commutate): Use it. Handle more than 2 operands.
16899 (dt_operand::gen_gimple_expr): Use commutative_op.
16900 (parser::parse_expr): Allow :c to be used with non-binary
16901 operators if the commutative operand is known.
16902 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
16903 CFN_FMS, CFN_FNMA and CFN_FNMS.
16904 (backprop::process_assign_use): Remove FMA_EXPR handling.
16905 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
16906 (gen_hsa_fma): New function.
16907 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
16908 IFN_FNMA and IFN_FNMS.
16909 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
16910 * gimple-fold.h (follow_all_ssa_edges): Declare.
16911 * gimple-fold.c (follow_all_ssa_edges): New function.
16912 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
16913 gimple_build interface and use follow_all_ssa_edges to fold the result.
16914 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
16915 instead of checking for optabs directly.
16916 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
16917 rather than FMA_EXPRs.
16918 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
16919 call to IFN_FMA instead of an FMA_EXPR.
16921 2018-05-17 Jim Wilson <jimw@sifive.com>
16923 * expr.c (do_tablejump): When converting index to Pmode, if we have a
16924 sign extended promoted subreg, and the range does not have the sign bit
16925 set, then do a sign extend.
16927 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
16928 test, check for sign extended subreg and/or constant operands, and
16929 do a sign extend in that case.
16931 2018-05-17 Steve Ellcey <sellcey@cavium.com>
16933 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
16934 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
16936 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
16937 Change logics_shift_reg to logics_shift_imm.
16938 (thunderx2t99_fp_loadpair_basic): Delete.
16939 (thunderx2t99_fp_storepair_basic): Delete.
16940 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
16941 (thunderx2t99_asimd_polynomial): Delete.
16942 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
16943 and neon_fp_mul_d_scalar_q.
16944 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
16945 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
16946 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
16947 (thunderx2t99_asimd_lut): Add missing tbl types.
16948 (thunderx2t99_asimd_ext): Delete.
16949 (thunderx2t99_asimd_load1_1_mult): Delete.
16950 (thunderx2t99_asimd_load1_2_mult): Delete.
16951 (thunderx2t99_asimd_load1_ldp): New.
16952 (thunderx2t99_asimd_load1): New.
16953 (thunderx2t99_asimd_load2): Add missing *load2* types.
16954 (thunderx2t99_asimd_load3): New.
16955 (thunderx2t99_asimd_load4): New.
16956 (thunderx2t99_asimd_store1_1_mult): Delete.
16957 (thunderx2t99_asimd_store1_2_mult): Delete.
16958 (thunderx2t99_asimd_store2_mult): Delete.
16959 (thunderx2t99_asimd_store2_onelane): Delete.
16960 (thunderx2t99_asimd_store_stp): New.
16961 (thunderx2t99_asimd_store1): New.
16962 (thunderx2t99_asimd_store2): New.
16963 (thunderx2t99_asimd_store3): New.
16964 (thunderx2t99_asimd_store4): New.
16966 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
16968 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
16969 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
16971 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
16972 Segher Boessenkool <segher@kernel.crashing.org>
16975 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
16978 2018-05-17 Richard Biener <rguenther@suse.de>
16980 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
16981 for pruning loop and prune defs feeding only already visited PHIs.
16983 2018-05-17 Richard Biener <rguenther@suse.de>
16985 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
16987 2018-05-17 Bin Cheng <bin.cheng@arm.com>
16988 Richard Biener <rguenther@suse.de>
16990 PR tree-optimization/85793
16991 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
16992 for VMAT_ELEMENTWISE.
16994 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
16996 * internal-fn.h (lookup_internal_fn): Declare
16997 * internal-fn.c (lookup_internal_fn): New function.
16998 * gimple.c (gimple_build_call_from_tree): Handle calls to
16999 internal functions.
17000 * gimple-pretty-print.c (dump_gimple_call): Print "." before
17001 internal function names.
17002 * tree-pretty-print.c (dump_generic_node): Likewise.
17003 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
17005 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
17007 * gimple-fold.h (gimple_build): Make the function forms take
17008 combined_fn rather than built_in_function.
17009 (gimple_simplify): Likewise.
17010 * gimple-match-head.c (gimple_simplify): Likewise.
17011 * gimple-fold.c (gimple_build): Likewise.
17012 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
17013 rather than gimple_build_call_internal.
17014 (get_initial_defs_for_reduction): Likewise.
17015 (vect_create_epilog_for_reduction): Likewise.
17016 (vectorizable_live_operation): Likewise.
17018 2018-05-17 Martin Liska <mliska@suse.cz>
17020 * gimple-ssa-sprintf.c (format_directive): Do not use
17021 space in between 'G_' and '('.
17023 2018-05-17 Jakub Jelinek <jakub@redhat.com>
17026 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
17027 even if the mask is not all ones.
17030 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
17032 (ix86_gimple_fold_builtin): Likewise.
17035 * config/i386/i386.c: Include tree-vector-builder.h.
17036 (ix86_vector_shift_count): New function.
17037 (ix86_fold_builtin): Fold shift builtins by scalar count.
17038 (ix86_gimple_fold_builtin): Likewise.
17040 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
17041 _mm512_setzero): New intrinsics.
17043 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
17044 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17046 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
17047 code generation for cases where splatting a value is not useful.
17048 * simplify-rtx.c (simplify_ternary_operation): Simplify
17049 vec_merge across a vec_duplicate and a paradoxical subreg forming
17050 a vector mode to a vec_concat.
17052 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
17054 * config.gcc: Support "goldmont-plus".
17055 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17057 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17058 PROCESSOR_GOLDMONT_PLUS.
17059 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
17060 (processor_target_table): Add "goldmont-plus".
17061 (PTA_GOLDMONT_PLUS): Define.
17062 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
17063 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
17064 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
17065 (fold_builtin_cpu): Add "goldmont-plus".
17066 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
17067 (ix86_option_override_internal): Add "goldmont-plus".
17068 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
17069 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
17070 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
17071 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
17073 2018-05-17 Richard Biener <rguenther@suse.de>
17075 PR tree-optimization/85757
17076 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
17077 remove defs that only feed that PHI from further processing.
17079 2018-05-16 Jim Wilson <jimw@sifive.com>
17081 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
17083 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
17085 2018-05-16 Mark Wielaard <mark@klomp.org>
17087 * dwarf2out.c (count_index_strings): New function.
17088 (output_indirect_strings): Call count_index_strings and generate
17089 header for dwarf_version >= 5.
17091 2018-05-16 Mark Wielaard <mark@klomp.org>
17093 * dwarf2out.c (dwarf_FORM): New function.
17094 (set_indirect_string): Use dwarf_FORM.
17095 (reset_indirect_string): Likewise.
17096 (size_of_die): Likewise.
17097 (value_format): Likewise.
17098 (output_die): Likewise.
17099 (add_skeleton_AT_string): Likewise.
17100 (output_macinfo_op): Likewise.
17101 (index_string): Likewise.
17102 (output_index_string_offset): Likewise.
17103 (output_index_string): Likewise.
17104 (count_index_strings): Likewise.
17106 2018-05-16 Carl Love <cel@us.ibm.com>
17108 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
17109 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
17111 2018-05-16 Martin Jambor <mjambor@suse.cz>
17113 * ipa-prop.c (ipa_free_all_edge_args): Remove.
17114 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
17116 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
17118 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
17119 (fnma<mode>4): Likewise.
17120 (fms<mode>4): Likewise.
17121 (fnms<mode>4): Likewise.
17122 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
17123 (aarch64_fnma<mode>4): Likewise.
17124 (aarch64_fms<mode>4): Likewise.
17125 (aarch64_fnms<mode>4): Likewise.
17126 (aarch64_fnmadd<mode>4): Likewise.
17128 2018-05-16 Jason Merrill <jason@redhat.com>
17130 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
17132 2018-05-16 Richard Biener <rguenther@suse.de>
17134 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
17135 (dump_stmt_cost): Declare.
17136 (add_stmt_cost): Dump cost we add.
17137 (add_stmt_costs): New function.
17138 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
17139 No longer exported.
17140 (vect_analyze_stmt): Adjust prototype.
17141 (vectorizable_condition): Likewise.
17142 (vectorizable_live_operation): Likewise.
17143 (vectorizable_reduction): Likewise.
17144 (vectorizable_induction): Likewise.
17145 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
17146 cost vector to pass to vectorizable_ and record afterwards.
17147 (vect_model_reduction_cost): Take cost vector argument and adjust.
17148 (vect_model_induction_cost): Likewise.
17149 (vectorizable_reduction): Likewise.
17150 (vectorizable_induction): Likewise.
17151 (vectorizable_live_operation): Likewise.
17152 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
17153 SLP_TREE_NUMBER_OF_VEC_STMTS.
17154 (vect_analyze_slp_cost_1): Remove.
17155 (vect_analyze_slp_cost): Likewise.
17156 (vect_slp_analyze_node_operations): Take visited args and
17157 a target cost vector. Avoid processing already visited stmt sets.
17158 (vect_slp_analyze_operations): Use a local cost vector to gather
17159 costs and register those of non-discarded instances.
17160 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
17161 (vect_schedule_slp_instance): Remove copying of
17162 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
17164 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
17165 adding cost. Record cost entry location.
17166 (vect_prologue_cost_for_slp_op): Function to compute cost of
17167 a constant or invariant generated for SLP vect in the prologue,
17168 split out from vect_analyze_slp_cost_1.
17169 (vect_model_simple_cost): Make static. Adjust for SLP costing.
17170 (vect_model_promotion_demotion_cost): Likewise.
17171 (vect_model_store_cost): Likewise, make static.
17172 (vect_model_load_cost): Likewise.
17173 (vectorizable_bswap): Add cost vector arg and adjust.
17174 (vectorizable_call): Likewise.
17175 (vectorizable_simd_clone_call): Likewise.
17176 (vectorizable_conversion): Likewise.
17177 (vectorizable_assignment): Likewise.
17178 (vectorizable_shift): Likewise.
17179 (vectorizable_operation): Likewise.
17180 (vectorizable_store): Likewise.
17181 (vectorizable_load): Likewise.
17182 (vectorizable_condition): Likewise.
17183 (vectorizable_comparison): Likewise.
17184 (can_vectorize_live_stmts): Likewise.
17185 (vect_analyze_stmt): Likewise.
17186 (vect_transform_stmt): Adjust calls to vectorizable_*.
17187 * tree-vectorizer.c: Include gimple-pretty-print.h.
17188 (dump_stmt_cost): New function.
17190 2018-05-16 Richard Biener <rguenther@suse.de>
17192 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
17193 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
17194 * tree-ssa-dse.c: Include tree-ssa-loop.h.
17195 (check_name): New callback.
17196 (dse_classify_store): Track cycles via a visited bitmap of PHI
17197 defs and simplify handling of in-loop and across loop dead stores
17198 and properly fail for loop-variant refs. Handle byte-tracking with
17199 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
17202 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
17204 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
17205 (vect_get_mask_type_for_stmt): Likewise.
17206 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
17208 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
17209 to determine the statement's vector type and the vector type that
17210 should be used for calculating nunits. Deal with cases in which
17211 the type has to be deferred.
17212 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
17213 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
17214 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
17215 (vect_determine_vf_for_stmt): New functions, split out from...
17216 (vect_determine_vectorization_factor): ...here.
17217 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
17218 (vect_get_mask_type_for_stmt): New functions, split out from
17219 vect_determine_vectorization_factor.
17221 2018-05-16 Richard Biener <rguenther@suse.de>
17223 * tree-cfg.c (verify_gimple_assign_ternary): Properly
17224 verify the [VEC_]COND_EXPR embedded comparison.
17226 2018-05-15 Martin Sebor <msebor@redhat.com>
17228 PR tree-optimization/85753
17229 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
17230 RECORD_TYPE in addition to ARRAY_TYPE.
17232 2018-05-15 Martin Sebor <msebor@redhat.com>
17234 PR middle-end/85643
17235 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
17237 2018-05-15 Richard Biener <rguenther@suse.de>
17239 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
17240 add by_clobber_p one. Change algorithm to collect all defs
17241 representing uses we need to walk and try reducing them to
17242 a single one before failing.
17243 (dse_dom_walker::dse_optimize_stmt): Adjust.
17245 2018-05-13 Mark Wielaard <mark@klomp.org>
17247 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
17248 (size_of_loc_descr): Likewise.
17249 (output_loc_operands): Likewise.
17250 (output_loc_operands_raw): Likewise.
17251 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
17252 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
17253 (hash_loc_operands): Likewise.
17254 (compare_loc_operands): Likewise.
17256 2018-05-14 Mark Wielaard <mark@klomp.org>
17258 * dwarf2out.c (count_index_addrs): New function.
17259 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
17261 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17263 PR tree-optimization/83648
17264 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
17265 return value as malloc candidate.
17267 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17270 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
17271 param as true in call to suggest_attribute.
17273 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
17275 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
17276 -mreadonly-in-sdata.
17278 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17280 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
17282 (aarch64_crypto_aesd_fused): Likewise.
17284 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
17286 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
17287 (movsi_aarch64): Likewise.
17288 (load_pairsi): Likewise.
17289 (load_pairdi): Likewise.
17290 (store_pairsi): Likewise.
17291 (store_pairdi): Likewise.
17292 (load_pairsf): Likewise.
17293 (load_pairdf): Likewise.
17294 (store_pairsf): Likewise.
17295 (store_pairdf): Likewise.
17296 (zero_extend): Likewise.
17297 (trunc): Swap alternatives.
17298 (fcvt_target): Add '?' to prefer w over r.
17300 2018-05-14 Jakub Jelinek <jakub@redhat.com>
17303 * config/i386/i386.md: Disallow non-commutative arithmetics in
17304 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
17305 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
17306 in the peephole2 before it.
17308 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
17310 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
17311 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
17312 (ix86_handle_option): Handle -mcldemote.
17313 * config.gcc: New header.
17314 * config/i386/cldemoteintrin.h: New file.
17315 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
17316 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
17318 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17319 OPTION_MASK_ISA_CLDEMOTE.
17320 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
17321 (ix86_valid_target_attribute_inner_p): Ditto.
17322 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
17323 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
17324 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
17325 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
17326 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
17328 * config/i386/i386.opt: Add -mcldemote.
17329 * config/i386/x86intrin.h: New header.
17330 * doc/invoke.texi: Add -mcldemote.
17332 2018-05-14 Richard Biener <rguenther@suse.de>
17334 * doc/match-and-simplify.texi: Adjust :s documentation.
17336 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
17338 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
17339 intended memcpy size.
17340 (REORDER_45): Likewise.
17342 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
17344 * sort.cc: New file.
17345 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
17346 * vec.c (qsort_chk): Use gcc_qsort.
17347 * Makefile.in (OBJS-libcommon): Add sort.o.
17348 (build/sort.o): New target. Use it...
17349 (BUILD_RTL): ... here, and...
17350 (build/gencfn-macros): ... here, and...
17351 (build/genmatch): ... here.
17353 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17354 Chung-Ju Wu <jasonwucj@gmail.com>
17356 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
17357 * config/nds32/nds32-graywolf.md: New file.
17358 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
17359 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
17361 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
17362 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
17363 * config/nds32/nds32.md (pipeline_model): Add graywolf.
17364 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
17365 * config/nds32/pipelines.md: Include n15 settings.
17367 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17368 Chung-Ju Wu <jasonwucj@gmail.com>
17370 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
17371 * config/nds32/nds32-n13.md: New file.
17372 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
17373 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
17375 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
17376 * config/nds32/nds32.md (pipeline_model): Add n13.
17377 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
17378 * config/nds32/pipelines.md: Include n13 settings.
17380 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
17381 Chung-Ju Wu <jasonwucj@gmail.com>
17383 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
17384 * config/nds32/nds32-n10.md: New file.
17385 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
17386 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
17388 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
17389 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
17390 * config/nds32/nds32.md (pipeline_model): Add n10.
17391 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
17392 * config/nds32/pipelines.md: Include n10 settings.
17394 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
17395 Kito Cheng <kito.cheng@gmail.com>
17396 Chung-Ju Wu <jasonwucj@gmail.com>
17398 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
17399 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
17400 Add enum values for DSP extension instructions.
17401 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
17403 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
17404 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
17405 New code iterators.
17406 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
17407 * config/nds32/nds32-dspext.md: New file for DSP implementation.
17408 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
17409 * config/nds32/nds32-intrinsic.md: Likewise.
17410 * config/nds32/nds32_intrinsic.h: Likewise.
17411 * config/nds32/nds32-md-auxiliary.c: Likewise.
17412 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
17413 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
17414 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
17415 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
17416 * config/nds32/nds32-protos.h: New declarations for DSP extension.
17417 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
17418 TYPE_DMAC in switch statement.
17419 * config/nds32/nds32.c: New checking and implementation for DSP
17420 extension instructions.
17421 * config/nds32/nds32.h: Likewise.
17422 * config/nds32/nds32.md: Likewise.
17423 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
17424 * config/nds32/predicates.md: Implement new predicates for DSP
17427 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
17429 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
17430 Reformat alternatives and attributes so it is easier to identify
17431 which constraints/attributes go with which instruction.
17432 (mov<mode>_hardfloat32, FMOVE64): Likewise.
17433 (mov<mode>_softfloat32, FMOVE64): Likewise.
17434 (mov<mode>_hardfloat64, FMOVE64): Likewise.
17435 (mov<mode>_softfloat64, FMOVE64): Likewise.
17437 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
17439 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17441 (Basic PowerPC Built-in Functions): The new name of the
17442 subsection previously known as "PowerPC Built-in Functions".
17443 (Basic PowerPC Built-in Functions Available on all Configurations):
17445 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
17446 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
17447 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
17448 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
17450 2018-05-11 Martin Jambor <mjambor@suse.cz>
17453 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
17456 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17459 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
17461 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
17463 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
17464 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
17465 (ix86_handle_option): Handle -mwaitpkg.
17466 * config.gcc: New header.
17467 * config/i386/cpuid.h (bit_WAITPKG): New bit.
17468 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
17469 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
17471 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17472 OPTION_MASK_ISA_WAITPKG.
17473 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
17474 (ix86_option_override_internal): Add PTA_WAITPKG.
17475 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
17476 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
17477 IX86_BUILTIN_TPAUSE.
17478 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
17479 __builtin_ia32_umwait and __builtin_ia32_tpause.
17480 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
17481 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
17482 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
17483 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
17484 UNSPECV_TPAUSE): New.
17485 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
17486 * config/i386/i386.opt: Add -mwaitpkg.
17487 * config/i386/waitpkgintrin.h: New file.
17488 * config/i386/x86intrin.h: New header.
17489 * doc/invoke.texi: Add -mwaitpkg.
17491 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
17494 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
17496 (cortex-m0): Use armv6s-m isa.
17497 (cortex-m0plus): Likewise.
17498 (cortex-m1): Likewise.
17499 (cortex-m0.small-multiply): Likewise.
17500 (cortex-m0plus.small-multiply): Likewise.
17501 (cortex-m1.small-multiply): Likewise.
17503 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
17504 Jakub Jelinek <jakub@redhat.com>
17506 PR tree-optimization/85692
17507 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
17508 source permute as well.
17510 2018-05-11 Martin Liska <mliska@suse.cz>
17513 * doc/extend.texi: Document LLVM style format for no_sanitize
17516 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
17518 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
17519 mode_supports_vsx_dform_quad to mode_supports_dq_form.
17520 (mode_supports_vsx_dform_quad): Likewise.
17521 (mode_supports_vmx_dform): Move these functions to be next to the
17522 other mode_supports functions.
17523 (mode_supports_dq_form): Likewise.
17524 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
17525 mode_supports_dq_form.
17526 (reg_offset_addressing_ok_p): Likewise.
17527 (offsettable_ok_by_alignment): Likewise.
17528 (rs6000_legitimate_offset_address_p): Likewise.
17529 (legitimate_lo_sum_address_p): Likewise.
17530 (rs6000_legitimize_address): Likewise.
17531 (rs6000_legitimize_reload_address): Likewise.
17532 (rs6000_secondary_reload_inner): Likewise.
17533 (rs6000_preferred_reload_class): Likewise.
17534 (rs6000_output_move_128bit): Likewise.
17536 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17538 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
17539 Generate SImode target register for null target.
17540 <case IX86_BUILTIN_XGETBV>: Ditto.
17541 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
17542 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
17544 2018-05-10 Carl Love <cel@us.ibm.com>
17546 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
17547 dcbtt and dcbtstt if operands[2] is 0.
17549 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
17552 * config/i386/sse.md (usadv64qi): New expander.
17554 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
17556 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
17557 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
17558 -maltivec=be support.
17559 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
17560 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
17561 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17562 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
17563 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
17564 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
17565 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
17566 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
17567 altivec_vsumsws): Adjust.
17568 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
17569 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
17570 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
17572 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
17573 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
17574 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
17575 (altivec_lve<VI_char>x): Delete expand.
17576 (*altivec_lve<VI_char>x_internal): Rename to...
17577 (altivec_lve<VI_char>x): ... this.
17578 (altivec_lvxl_<mode>): Delete expand.
17579 (*altivec_lvxl_<mode>_internal): Rename to ...
17580 (altivec_lvxl_<mode>): ... this.
17581 (altivec_stvxl_<mode>): Delete expand.
17582 (*altivec_stvxl_<mode>_internal): Rename to ...
17583 (altivec_stvxl_<mode>): ... this.
17584 (altivec_stve<VI_char>x): Delete expand.
17585 (*altivec_stve<VI_char>x_internal): Rename to ...
17586 (altivec_stve<VI_char>x): ... this.
17587 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
17588 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
17589 reduc_plus_scal_<mode>): Adjust.
17590 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
17592 (rs6000_cpu_cpp_builtins): Adjust.
17593 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
17594 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
17595 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
17596 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
17597 -maltivec=be support.
17598 (rs6000_split_vec_extract_var): Adjust.
17599 (rs6000_split_v4si_init): Adjust.
17600 (swap_selector_for_mode): Delete.
17601 (altivec_expand_lvx_be, altivec_expand_stvx_be,
17602 altivec_expand_stvex_be): Delete.
17603 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
17604 -maltivec=be support.
17605 (rs6000_gimple_fold_builtin): Ditto.
17606 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
17608 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
17609 (TARGET_DIRECT_MOVE_64BIT): Adjust.
17610 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
17611 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
17612 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
17613 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
17614 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
17615 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
17616 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
17617 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
17618 anonymous split): Adjust.
17619 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
17620 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
17622 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
17624 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
17625 when --with-gxx-include-dir is also specified.
17626 * configure: Regenerate.
17628 2018-05-09 Jim Wilson <jimw@sifive.com>
17631 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
17632 * config/riscv/t-withmultilib: New.
17633 * config/riscv/withmultilib.h: New.
17634 * doc/install.texi: Document RISC-V --with-multilib-list support.
17636 2018-05-09 Richard Biener <rguenther@suse.de>
17638 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
17640 (vect_bb_vectorization_profitable_p): Adjust. Compute
17641 actual scalar cost using the cost vector and the add_stmt_cost
17644 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17646 PR rtl-optimization/85645
17647 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
17648 in the REG_CFA_REGISTER note for LR, don't leave it empty.
17650 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17652 PR rtl-optimization/85645
17653 * shrink-wrap.c (spread_components): Return a boolean saying if
17654 anything was changed.
17655 (try_shrink_wrapping_separate): Iterate spread_components until
17656 nothing changes anymore.
17658 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17660 PR rtl-optimization/85645
17661 * regrename.c (build_def_use): Also kill the chains that include the
17662 destination of a REG_CFA_REGISTER note.
17664 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
17666 PR rtl-optimization/85645
17667 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
17668 insn that has a REG_CFA_REGISTER note.
17670 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
17672 * cfgexpand.c (expand_clobber): New function.
17673 (expand_gimple_stmt_1): Use it.
17674 * tree-vect-stmts.c (vect_clobber_variable): New function,
17676 (vectorizable_simd_clone_call): ...here.
17677 (vectorizable_store): Emit a clobber either side of an
17678 IFN_STORE_LANES sequence.
17679 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
17681 2018-05-09 Tom de Vries <tom@codesourcery.com>
17684 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
17685 (define_insn "trap_if_false"): Add exit after trap.
17687 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
17689 PR rtl-optimization/85638
17690 * bb-reorder.c: Include common/common-target.h.
17691 (create_forwarder_block): New function extracted from...
17692 (fix_up_crossing_landing_pad): ...here. Rename into...
17693 (dw2_fix_up_crossing_landing_pad): ...this.
17694 (sjlj_fix_up_crossing_landing_pad): New function.
17695 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
17696 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
17697 from both partitions and exit the loop after one iteration.
17699 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17702 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17704 (Basic PowerPC Built-in Functions): The new name of the
17705 subsection previously known as "PowerPC Built-in Functions".
17706 (Basic PowerPC Built-in Functions Available on all Configurations):
17708 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17710 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17712 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17714 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17717 2018-05-08 Jim Wilson <jimw@sifive.com>
17719 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
17720 (LD_EMUL_SUFFIX): New.
17721 (LINK_SPEC): Use it.
17723 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
17725 * doc/extend.texi (PowerPC Built-in Functions): Rename this
17727 (Basic PowerPC Built-in Functions): The new name of the
17728 subsection previously known as "PowerPC Built-in Functions".
17729 (Basic PowerPC Built-in Functions Available on all Configurations):
17731 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
17733 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
17735 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
17737 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
17740 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17743 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
17744 after cmpelim optimization.
17746 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
17748 * config.gcc: Support "goldmont".
17749 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
17750 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
17751 PROCESSOR_GOLDMONT.
17752 * config/i386/i386.c (m_GOLDMONT): Define.
17753 (processor_target_table): Add "goldmont".
17754 (PTA_GOLDMONT): Define.
17755 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
17756 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
17757 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
17758 (fold_builtin_cpu): Add "goldmont".
17759 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
17760 (ix86_option_override_internal): Add "goldmont".
17761 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
17762 (processor_type): Add PROCESSOR_GOLDMONT.
17763 * config/i386/i386.md: Add CPU "glm".
17764 * config/i386/glm.md: New file.
17765 * config/i386/x86-tune.def: Add m_GOLDMONT.
17766 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
17768 2018-05-08 Jakub Jelinek <jakub@redhat.com>
17771 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
17773 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
17774 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
17775 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
17778 * config/i386/i386.c (ix86_fold_builtin): Handle
17779 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
17782 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
17783 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
17785 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
17788 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
17789 (check_arch): Likewise.
17790 (check_fpu): Return the result rather than printing it.
17791 (end arch): Fix operator precedence.
17792 (end cpu): Likewise.
17793 (END): Print the result from check_fpu.
17795 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17796 Alan Hayward <alan.hayward@arm.com>
17797 David Sherwood <david.sherwood@arm.com>
17799 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
17800 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
17801 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
17802 (*fcmuo<mode>_and): New patterns.
17804 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
17806 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
17807 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
17808 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
17809 (cmp_op, sve_imm_con): New code attributes.
17810 (SVE_COND_INT_CMP, imm_con): Delete.
17811 (cmp_op): Remove above unspecs from int attribute.
17812 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
17814 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
17815 comparison-specific unspecs.
17816 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
17817 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
17818 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
17819 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
17820 (*vec_fcm<cmp_op><mode>): Rename to...
17821 (*fcm<cmp_op><mode>): ...this and adjust likewise.
17822 (*vec_fcmuo<mode>): Rename to...
17823 (*fcmuo<mode>): ...this and adjust likewise.
17824 (*pred_fcm<cmp_op><mode>): New pattern.
17825 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
17826 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
17828 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
17830 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
17831 (aarch64_emit_sve_predicated_cond): New function.
17832 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
17833 (aarch64_emit_unspec_cond_or): Replace with...
17834 (aarch64_emit_sve_or_conds): ...this new function. Use
17835 aarch64_emit_sve_ptrue_op for the individual comparisons and
17836 aarch64_emit_binop to OR them together.
17837 (aarch64_emit_inverted_unspec_cond): Replace with...
17838 (aarch64_emit_sve_inverted_cond): ...this new function. Use
17839 aarch64_emit_sve_ptrue_op for the comparison and
17840 aarch64_emit_unop to invert the result.
17841 (aarch64_expand_sve_vec_cmp_float): Update after the above
17842 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
17844 2018-05-07 Nathan Sidwell <nathan@acm.org>
17846 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
17847 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
17848 (Backwards Compatibility): Likewise.
17850 2018-05-07 Luis Machado <luis.machado@linaro.org>
17854 2018-05-07 Luis Machado <luis.machado@linaro.org>
17856 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17857 <prefetch_dynamic_strides>: New const bool field.
17858 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17859 prefetch_dynamic_strides.
17860 (exynosm1_prefetch_tune): Likewise.
17861 (thunderxt88_prefetch_tune): Likewise.
17862 (thunderx_prefetch_tune): Likewise.
17863 (thunderx2t99_prefetch_tune): Likewise.
17864 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
17866 (aarch64_override_options_internal): Update to set
17867 PARAM_PREFETCH_DYNAMIC_STRIDES.
17868 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
17869 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
17870 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
17871 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
17872 prefetch-dynamic-strides setting.
17874 2018-05-07 Luis Machado <luis.machado@linaro.org>
17876 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17877 <minimum_stride>: New const int field.
17878 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17879 minimum_stride field.
17880 (exynosm1_prefetch_tune): Likewise.
17881 (thunderxt88_prefetch_tune): Likewise.
17882 (thunderx_prefetch_tune): Likewise.
17883 (thunderx2t99_prefetch_tune): Likewise.
17884 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
17885 (aarch64_override_options_internal): Update to set
17886 PARAM_PREFETCH_MINIMUM_STRIDE.
17887 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
17888 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
17889 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
17890 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
17891 stride is constant and is below the minimum stride threshold.
17893 2018-05-07 Luis Machado <luis.machado@linaro.org>
17895 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
17898 2018-05-07 Luis Machado <luis.machado@linaro.org>
17900 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17901 <prefetch_dynamic_strides>: New const bool field.
17902 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17903 prefetch_dynamic_strides.
17904 (exynosm1_prefetch_tune): Likewise.
17905 (thunderxt88_prefetch_tune): Likewise.
17906 (thunderx_prefetch_tune): Likewise.
17907 (thunderx2t99_prefetch_tune): Likewise.
17908 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
17910 (aarch64_override_options_internal): Update to set
17911 PARAM_PREFETCH_DYNAMIC_STRIDES.
17912 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
17913 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
17914 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
17915 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
17916 prefetch-dynamic-strides setting.
17918 2018-05-07 Luis Machado <luis.machado@linaro.org>
17920 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17921 <minimum_stride>: New const int field.
17922 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17923 minimum_stride field.
17924 (exynosm1_prefetch_tune): Likewise.
17925 (thunderxt88_prefetch_tune): Likewise.
17926 (thunderx_prefetch_tune): Likewise.
17927 (thunderx2t99_prefetch_tune): Likewise.
17928 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
17929 (aarch64_override_options_internal): Update to set
17930 PARAM_PREFETCH_MINIMUM_STRIDE.
17931 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
17932 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
17933 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
17934 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
17935 stride is constant and is below the minimum stride threshold.
17937 2018-05-06 Jakub Jelinek <jakub@redhat.com>
17940 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
17941 the type is addressable. Don't force op into register if it has
17944 2018-05-05 Roland McGrath <mcgrathr@google.com>
17947 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
17948 any section for which we don't know a specific type it should have,
17949 regardless of name. Previously this was done only for the exact
17950 names ".init_array", ".fini_array", and ".preinit_array".
17951 (default_elf_asm_named_section): Add comment about
17952 relationship with default_section_type_flags and SECTION_NOTYPE.
17953 (get_section): Don't consider it a type conflict if one side has
17954 SECTION_NOTYPE and the other doesn't, as long as neither has the
17955 SECTION_BSS et al used in the default_section_type_flags logic.
17957 2018-05-05 Tom de Vries <tom@codesourcery.com>
17960 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
17961 (workaround_barsyncs): New function.
17962 (nvptx_reorg): Use workaround_barsyncs.
17963 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
17964 (define_expand "nvptx_membar_cta"): New define_expand.
17965 (define_insn "*nvptx_membar_cta"): New insn.
17967 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
17969 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
17970 To improve optimization opportunities.
17971 * builtin-types.def: The new needed builtin types for the above.
17973 2018-05-04 Richard Biener <rguenther@suse.de>
17975 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
17976 * gimple-ssa-store-merging.c
17977 (imm_store_chain_info::output_merged_store): Remove redundant create,
17978 release split_store vector contents on failure.
17979 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
17980 scalar stmt vector on cache hit.
17982 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
17984 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
17986 * config.gcc (powerpc-xilinx-eabi*): Remove.
17987 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
17989 (fusion_addis_mem_combo_load): Ditto.
17990 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
17992 (rs6000_cpu_cpp_builtins): Ditto.
17993 * config/rs6000/rs6000-linux.c
17994 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
17995 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
17996 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
17998 (rs6000_setup_reg_addr_masks): Ditto.
17999 (rs6000_init_hard_regno_mode_ok): Ditto.
18000 (rs6000_option_override_internal): Ditto.
18001 (legitimate_lo_sum_address_p): Ditto.
18002 (rs6000_legitimize_address): Ditto.
18003 (rs6000_legitimize_reload_address): Ditto.
18004 (rs6000_legitimate_address_p): Ditto.
18005 (abi_v4_pass_in_fpr): Ditto.
18006 (setup_incoming_varargs): Ditto.
18007 (rs6000_gimplify_va_arg): Ditto.
18008 (rs6000_split_multireg_move): Ditto.
18009 (rs6000_savres_strategy): Ditto.
18010 (rs6000_emit_prologue_components): Ditto.
18011 (rs6000_emit_epilogue_components): Ditto.
18012 (rs6000_emit_prologue): Ditto.
18013 (rs6000_emit_epilogue): Ditto.
18014 (rs6000_elf_file_end): Ditto.
18015 (rs6000_function_value): Ditto.
18016 (rs6000_libcall_value): Ditto.
18017 * config/rs6000/rs6000.h: Ditto.
18018 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
18019 (TARGET_MINMAX): ... this. New.
18020 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
18021 * config/rs6000/rs6000.md: Remove Xilinx FP support.
18022 (*movsi_internal1_single): Delete.
18023 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
18024 mfpu=, mxilinx-fpu): Delete.
18025 * config/rs6000/singlefp.h: Delete.
18026 * config/rs6000/sysv4.h: Remove Xilinx FP support.
18027 * config/rs6000/t-rs6000: Ditto.
18028 * config/rs6000/t-xilinx: Delete.
18029 * config/rs6000/titan.md: Adjust for fp_type removal.
18030 * config/rs6000/vsx.md: Remove Xilinx FP support.
18031 (VStype_simple): Delete.
18032 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
18033 * config/rs6000/xfpu.h: Delete.
18034 * config/rs6000/xfpu.md: Delete.
18035 * config/rs6000/xilinx.h: Delete.
18036 * config/rs6000/xilinx.opt: Delete.
18037 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
18038 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
18040 2018-05-04 Tom de Vries <tom@codesourcery.com>
18043 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
18046 2018-05-04 Richard Biener <rguenther@suse.de>
18048 PR middle-end/85627
18049 * tree-complex.c (update_complex_assignment): We are always in SSA form.
18050 (expand_complex_div_wide): Likewise.
18051 (expand_complex_operations_1): Likewise.
18052 (expand_complex_libcall): Preserve EH info of the original stmt.
18053 (tree_lower_complex): Handle removed blocks.
18054 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
18055 on complex multiplication and division libcall builtins.
18057 2018-05-04 Richard Biener <rguenther@suse.de>
18059 PR middle-end/85574
18060 * fold-const.c (negate_expr_p): Restrict negation of operand
18061 zero of a division to when we know that can happen without
18063 (fold_negate_expr_1): Likewise.
18065 2018-05-04 Jakub Jelinek <jakub@redhat.com>
18068 * real.h (real_nextafter): Declare.
18069 * real.c (real_nextafter): New function.
18070 * fold-const-call.c (fold_const_nextafter): New function.
18071 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
18072 CASE_CFN_NEXTTOWARD.
18073 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
18074 even when arg1_mode is different from arg0_mode.
18076 2018-05-03 Nathan Sidwell <nathan@acm.org>
18078 * doc/extend.texi (Deprecated Features): Remove
18079 -ffriend-injection.
18080 (Backwards Compatibility): Likewise.
18081 * doc/invoke.texi (C++ Language Options): Likewise.
18082 (C++ Dialect Options): Likewise.
18084 2018-05-03 Jakub Jelinek <jakub@redhat.com>
18087 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
18088 _mm512_mask_mullox_epi64): New intrinsics.
18090 2018-05-03 Tom de Vries <tom@codesourcery.com>
18093 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
18094 dump files): Add offload-tree.
18096 2018-05-03 Richard Biener <rguenther@suse.de>
18098 PR tree-optimization/85615
18099 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
18100 to loops not nested in BBs loop father to avoid creating multi-entry
18103 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18105 PR tree-optimization/70291
18106 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
18107 arguments. Change return type to tree. Emit libcall as a new
18108 statement rather than replacing existing one when inplace_p is true.
18109 (expand_complex_multiplication_components): New function.
18110 (expand_complex_multiplication): Expand floating-point complex
18111 multiplication using the above.
18112 (expand_complex_division): Rename inner_type parameter to type.
18113 Update expand_complex_libcall call-site.
18114 (expand_complex_operations_1): Update expand_complex_multiplication
18115 and expand_complex_division call-sites.
18117 2018-05-02 Jakub Jelinek <jakub@redhat.com>
18120 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
18121 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
18122 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
18123 the highest significant bit of the shift count mask is clear. In
18124 check whether and[sq]i3 is needed verify that all significant bits
18125 of the shift count other than the highest are set.
18127 2018-05-02 Tom de Vries <tom@codesourcery.com>
18130 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
18131 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
18132 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
18133 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
18134 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
18135 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
18136 __builtin_goacc_parlevel_size.
18138 2018-05-02 Richard Biener <rguenther@suse.de>
18140 PR tree-optimization/85597
18141 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
18142 do not use split vect_get_vec_defs call but call vect_get_slp_defs
18145 2018-05-02 Tom de Vries <tom@codesourcery.com>
18148 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
18149 dump files): Add ltrans-tree.
18151 2018-05-02 Tom de Vries <tom@codesourcery.com>
18154 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
18155 dump files): Add wpa-ipa.
18157 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
18159 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
18160 powerpc*-*-linux*paired* target.
18161 * config/rs6000/750cl.h: Delete.
18162 * config/rs6000/paired.h: Delete.
18163 * config/rs6000/paired.md: Delete.
18164 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
18166 * config/rs6000/rs6000-builtin.def: Remove paired float support.
18167 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
18168 comment. Remove paired float support.
18169 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
18170 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
18172 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
18173 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
18175 * config/rs6000/rs6000.c: Remove paired float support.
18176 (paired_expand_vector_init, paired_expand_vector_move,
18177 paired_emit_vector_compare, paired_emit_vector_cond_expr,
18178 (paired_expand_lv_builtin, paired_expand_stv_builtin,
18179 paired_expand_builtin, paired_expand_predicate_builtin,
18180 paired_init_builtins): Delete.
18181 * config/rs6000/rs6000.h: Remove paired float support.
18182 * config/rs6000/rs6000.md: Remove paired float support.
18183 (move_from_CR_ov_bit): Delete.
18184 * config/rs6000/rs6000.opt (mpaired): Delete.
18185 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
18186 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
18188 2018-05-02 Richard Biener <rguenther@suse.de>
18190 PR middle-end/85567
18191 * gimplify.c (gimplify_save_expr): When in SSA form allow
18192 SAVE_EXPRs to compute to SSA vars.
18194 2018-05-02 Jakub Jelinek <jakub@redhat.com>
18197 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
18198 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
18199 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
18200 clobber operands[2], instead use a new pseudo. Formatting fixes.
18202 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
18204 PR tree-optimization/85586
18205 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
18206 exit early for statements in the same group if the accesses are
18209 2018-05-02 Tom de Vries <tom@codesourcery.com>
18212 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
18215 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
18217 PR tree-optimization/85143
18218 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
18220 2018-05-01 Tom de Vries <tom@codesourcery.com>
18223 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
18224 not found" error message.
18226 2018-05-01 Tom de Vries <tom@codesourcery.com>
18229 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
18230 * vec.c (test_ordered_remove_if): New function.
18231 (vec_c_tests): Call test_ordered_remove_if.
18232 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
18233 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
18234 * tree-vect-patterns.c (vect_pattern_recog_1): Use
18235 VEC_ORDERED_REMOVE_IF.
18237 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18239 PR tree-optimization/82665
18240 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
18241 pointer subtraction where arguments come from a memchr call.
18243 2018-05-01 Jakub Jelinek <jakub@redhat.com>
18245 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
18246 --push-state --as-needed and --pop-state instead of --as-needed and
18247 --no-as-needed if ld supports it.
18248 * configure: Regenerated.
18251 * doc/install.texi2html: Replace _002d with - and _002a with * in
18252 generated html files using sed.
18254 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18257 * gcc-rich-location.c (blank_line_before_p): New function.
18258 (use_new_line): New function.
18259 (gcc_rich_location::add_fixit_insert_formatted): New function.
18260 * gcc-rich-location.h
18261 (gcc_rich_location::add_fixit_insert_formatted): New function.
18263 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18265 * selftest.c (assert_streq): Rename "expected" and "actual" to
18266 "val1" and "val2". Extend NULL-handling to cover both inputs
18267 symmetrically, while still requiring both to be non-NULL for a pass.
18268 * selftest.h (assert_streq): Rename "expected" and "actual" to
18270 (ASSERT_EQ): Likewise.
18271 (ASSERT_EQ_AT): Likewise.
18272 (ASSERT_KNOWN_EQ): Likewise.
18273 (ASSERT_KNOWN_EQ_AT): Likewise.
18274 (ASSERT_NE): Likewise.
18275 (ASSERT_MAYBE_NE): Likewise.
18276 (ASSERT_MAYBE_NE_AT): Likewise.
18277 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
18278 the assertion to pass.
18279 (ASSERT_STREQ_AT): Likewise.
18281 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
18283 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
18284 interaction with -pie.
18286 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18288 * selftest.h: Fix alphabetization of per-source-file selftest
18291 2018-04-30 Jason Merrill <jason@redhat.com>
18293 PR c++/61982 - dead stores to destroyed objects.
18294 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
18297 2018-04-30 Jason Merrill <jason@redhat.com>
18299 * tree.c (build_clobber): New.
18300 * tree.h: Declare it.
18301 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
18303 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18305 * diagnostic-show-locus.c (layout::layout): Update for
18306 location_get_source_line returning a char_span.
18307 (struct char_span): Move to input.h.
18308 (struct correction): Update for fields in char_span becoming
18310 (struct source_line): Update for location_get_source_line
18311 returning a char_span.
18312 (layout::print_line): Likewise.
18313 * edit-context.c (edited_file::print_content): Likewise.
18314 (edited_file::print_diff_hunk): Likewise.
18315 (edited_file::print_run_of_changed_lines): Likewise.
18316 (edited_file::get_num_lines): Likewise.
18317 (edited_line::edited_line): Likewise.
18318 * final.c (asm_show_source): Likewise.
18319 * input.c (location_get_source_line): Convert return type
18320 from const char * to char_span, losing the final "line_len"
18322 (dump_location_info): Update for the above.
18323 (get_substring_ranges_for_loc): Likewise. Use a char_span
18324 when handling the literal within the line.
18325 (test_reading_source_line): Update for location_get_source_line
18326 returning a char_span.
18327 * input.h (class char_span): Move here from
18328 diagnostic-show-locus.c, converting from a struct to a class.
18329 Make data members private.
18330 (char_span::operator bool): New.
18331 (char_span::length): New.
18332 (char_span::get_buffer): New.
18333 (char_span::operator[]): New.
18334 (char_span::subspan): Make const.
18335 (char_span::xstrdup): New.
18336 (location_get_source_line): Convert return type from const char *
18337 to char_span, losing the final "line_size" param.
18339 2018-04-30 Jan Hubicka <jh@suse.cz>
18341 * lto-wrapper.c (ltrans_priorities): New static var.
18342 (cmp_priority): New.
18343 (run_gcc): Read priorities and if doing parallel build order
18344 the Makefile by them.
18346 2018-04-30 David Malcolm <dmalcolm@redhat.com>
18348 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
18350 2018-04-30 Richard Biener <rguenther@suse.de>
18352 * tree-cfg.c (verify_address): Remove base argument, add
18353 flag whether to check TREE_ADDRESSABLE and do that.
18354 (verify_expr): Remove.
18355 (verify_types_in_gimple_reference): Add pieces from verify_expr.
18356 (verify_gimple_assign_single): Likewise.
18357 (verify_gimple_switch): Likewise.
18358 (verify_expr_location_1): Dereference tp once. Add (disabled)
18359 piece from verify_expr.
18360 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
18362 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18364 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
18366 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18368 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
18369 (small_data_pattern): Likewise.
18370 (arc_rewrite_small_data): Likewise.
18371 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
18372 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
18373 (get_symbol_alignment): New function.
18374 (legitimate_small_data_address_p): Likewise.
18375 (legitimate_scaled_address): Update, call
18376 legitimate_small_data_address_p.
18377 (output_sdata): New static variable.
18378 (arc_print_operand): Update how we handle small data operands.
18379 (arc_print_operand_address): Likewise.
18380 (arc_legitimate_address_p): Update, use
18381 legitimate_small_data_address_p.
18382 (arc_rewrite_small_data_p): Remove.
18383 (arc_rewrite_small_data_1): Likewise.
18384 (arc_rewrite_small_data): Likewise.
18385 (small_data_pattern): Likewise.
18386 (compact_sda_memory_operand): Update to use
18387 legitimate_small_data_address_p and get_symbol_alignment.
18388 (prepare_move_operands): Don't rewite sdata pattern.
18389 (prepare_extend_operands): Remove.
18390 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
18392 (zero_extendqisi2): Likewise.
18393 (zero_extendhisi2): Likewise.
18394 (extendqihi2): Likewise.
18395 (extendqisi2): Likewise.
18396 (extendhisi2): Likewise.
18397 (addsi3): Likewise.
18398 (subsi3): Likewise.
18399 (andsi3): Likewise.
18400 * config/arc/constraints.md (Usd): Change it to memory constraint.
18402 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
18404 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
18405 as source of std instructions.
18406 * config/arc/arc.md (movsi_insn): Update pattern predicate to
18407 allow 6-bit constants as source for store instructions.
18408 (movdi_insn): Update instruction pattern to allow 6-bit constants
18409 as source for store instructions.
18411 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
18413 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
18415 2018-04-30 Nathan Sidwell <nathan@acm.org>
18416 Sandra Loosemore <sandra@codesourcery.com>
18418 * dumpfile.c (dump_open): Allow '-' for stdout.
18419 * doc/invoke.texi (Developer Options): Document dump filename
18420 determination early. Document stdin/stdout selection.
18422 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
18424 Microblaze Target: PIC data text relative
18426 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
18427 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
18429 * config/microblaze/microblaze.h (microblaze_constant_address_p):
18430 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
18431 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
18432 New addressing mode for data-text relative position indepenedent code.
18433 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
18434 'ADDRESS_SYMBOLIC_TXT_REL'.
18435 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
18436 (microblaze_legitimate_pic_operand): Exclude function calls from
18437 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
18438 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
18440 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18441 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
18442 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
18443 for 'address + offset'.
18444 (microblaze_expand_prologue): Add new function prologue call for
18446 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
18447 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
18448 table in case of TARGET_PIC_DATA_TEXT_REL.
18449 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
18450 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
18451 Add new macros 'UNSPEC_TEXT',
18452 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
18453 + exclude function calls from 'UNSPEC_PLT' in case of data text
18455 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18456 new target hook for generating address diff vector tables in case of
18458 * doc/tm.texi : Regenerate.
18459 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
18460 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
18461 of addr diff vector generation.
18462 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
18463 target hook definition.
18464 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
18465 Add default function for generate_pic_addr_diff_vec -> flag_pic.
18466 * doc/invoke.texi (Add new pic option): Add new microblaze pic
18467 option for data text relative.
18469 2018-04-30 Richard Biener <rguenther@suse.de>
18471 * tree-chrec.h (evolution_function_is_constant_p): Remove
18473 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
18475 2018-04-30 Richard Biener <rguenther@suse.de>
18478 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
18480 2018-04-30 Richard Biener <rguenther@suse.de>
18482 PR tree-optimization/28364
18483 PR tree-optimization/85275
18484 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
18485 copying first exit test.
18487 2018-04-28 Mark Wielaard <mark@klomp.org>
18489 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
18490 dwarf_version >= 5.
18491 (dwarf_AT): Handle DW_AT_addr_base.
18492 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
18494 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
18497 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
18498 (*ashl<dwi>3_doubleword_mask_1): Ditto.
18499 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
18500 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
18502 2018-04-28 Richard Biener <rguenther@suse.de>
18504 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
18505 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
18506 to reflect use. Only add interesting stmts.
18508 2018-04-27 Martin Jambor <mjambor@suse.cz>
18511 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
18512 the jump function allows for passing through aggregate values.
18514 2018-04-27 David Malcolm <dmalcolm@redhat.com>
18516 * input.h (in_system_header_at): Convert from macro to inline
18518 (from_macro_expansion_at): Likewise.
18519 (from_macro_definition_at): Likewise.
18521 2018-04-27 Jeff Law <law@redhat.com>
18523 * config.gcc: Mark tile* targets as deprecated/obsolete.
18525 2018-04-27 Richard Biener <rguenther@suse.de>
18527 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
18530 2018-04-27 Richard Biener <rguenther@suse.de>
18532 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
18534 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
18536 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
18537 with Yd constraint. Set "preferred_for_speed" attribute from
18538 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
18539 with Yd constraint.
18540 (*movdi_internal): Ditto.
18541 (movti_interunit splitters): Remove
18542 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
18543 (movdi_interunit splitters): Ditto.
18544 * config/i386/constraints.md (Ye): Remove.
18545 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
18547 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18550 * config/aarch64/constraints.md (Usg): Limit to 31.
18551 (Usj): Limit to 63.
18553 2018-04-27 Jakub Jelinek <jakub@redhat.com>
18555 PR tree-optimization/85529
18556 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
18557 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
18558 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
18559 zero extension or masking of the MSB bit.
18560 (optimize_range_tests): Add FIRST_BB argument, pass it through
18561 to optimize_range_tests_var_bound.
18562 (maybe_optimize_range_tests, reassociate_bb): Adjust
18563 optimize_range_tests callers.
18565 2018-04-26 Richard Biener <rguenther@suse.de>
18566 Jakub Jelinek <jakub@redhat.com>
18568 * cgraph.h (symbol_table): Just declare debug method here.
18569 * symtab.c (symbol_table::debug): Define.
18571 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
18573 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
18575 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
18577 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
18578 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
18579 (*movdi_internal): Substitute Yi and Yj constraint with x
18580 and Ym and Yn constraint with y constraint. Update "isa"
18581 attribute and set "preferred_for_speed" attribute from
18582 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
18583 (*movsi_internal): Ditto.
18584 (*movdf_internal): Ditto.
18585 (*movsf_internal): Ditto.
18586 (*zero_extendsidi2): Ditto.
18587 * config/i386/sse.md (vec_set<mode>_0): Ditto.
18588 (sse2_loadld): Ditto.
18589 (*vec_extract<ssevecmodelower>_0): Ditto.
18590 (*vec_extractv4si_0_zext_sse4): Ditto.
18591 (vec_concatv2di): Ditto.
18592 (*vec_dup<mode>): Ditto.
18593 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
18594 * config/i386/constraints.md (Yi): Remove.
18599 2018-04-26 Nathan Sidwell <nathan@acm.org>
18601 * dumpfile.c (dump_open): New.
18602 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
18603 (dump_finish): Detect stdio/stderr by value not name.
18605 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
18607 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
18609 2018-04-26 Tom de Vries <tom@codesourcery.com>
18612 * config/nvptx/nvptx.c (verify_neutering_jumps)
18613 (verify_neutering_labels): New function
18614 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
18616 2018-04-26 Tom de Vries <tom@codesourcery.com>
18619 * config/nvptx/nvptx.c (needs_neutering_p): New function.
18620 (nvptx_single): Use needs_neutering_p to skip over insns that do not
18623 2018-04-26 Richard Biener <rguenther@suse.de>
18624 Tom de Vries <tom@codesourcery.com>
18627 * lto-streamer-out.c (output_function): Fixup loops if required to match
18628 discovery done in the reader.
18630 2018-04-26 Richard Biener <rguenther@suse.de>
18632 PR tree-optimization/85116
18633 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
18634 have a loop exit from the single latch predecessor. Remove
18635 case of header with just condition.
18636 (ch_base::copy_headers): Exclude infinite loops from any
18638 (pass_ch::execute): Record exits.
18640 2018-04-26 Richard Biener <rguenther@suse.de>
18642 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
18643 prologue cost vector and pass it to vect_get_load_cost.
18644 (vect_get_peeling_costs_all_drs): Likewise.
18645 (vect_peeling_hash_get_lowest_cost): Likewise.
18646 (vect_enhance_data_refs_alignment): Likewise.
18648 2018-04-26 Richard Biener <rguenther@suse.de>
18650 PR middle-end/85450
18651 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
18652 checking of integer<->pointer conversions.
18653 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
18654 sign-/zero-extending pointer types.
18655 (expand_omp_for_static_chunk): Likewise.
18657 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
18658 Jean Lee <xiaoyur347@gmail.com>
18660 * config/mips/mips.c (mips_asan_shadow_offset): New function.
18661 (TARGET_ASAN_SHADOW_OFFSET): Define.
18662 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
18663 true for -fsanitize=address.
18665 2018-04-25 Mark Wielaard <mark@klomp.org>
18667 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
18670 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18672 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
18673 than "alu", remove explicit "memory" and "imm_disp" attributes.
18674 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
18676 PR middle-end/85414
18677 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
18678 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
18679 gen_lowpart_no_emit.
18681 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
18684 * config/i386/i386.c (ix86_expand_builtin): Change memory
18685 operand to XI, extend p0 to Pmode.
18686 * config/i386/i386.md: Change unspec volatile and operand
18687 1 mode to XI, change operand 0 mode to P.
18689 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18691 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
18692 GET_MODE_MASK before any checking.
18693 (nds32_can_use_bset_p): Likewise.
18694 (nds32_can_use_btgl_p): Likewise.
18696 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18698 * config/nds32/nds32-doubleword.md: New define_split pattern for
18699 illegal register number.
18701 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18703 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
18705 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
18707 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
18709 2018-04-25 Richard Biener <rguenther@suse.de>
18711 * lto-streamer.h (LTO_major_version): Bump to 8.
18713 2018-04-25 Jakub Jelinek <jakub@redhat.com>
18715 * BASE-VER: Set to 9.0.0.
18717 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
18719 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
18720 in __abskf2 and __powikf2.
18722 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18725 * config/aarch64/constraints.md (Usg, Usj): New constraints.
18726 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
18727 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
18728 Use the above on operand 2. Reindent.
18729 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
18731 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18734 * common/config/i386/i386-common.c (ix86_handle_option): Don't
18736 * config/i386/i386.opt (mcet): Removed.
18737 * doc/install.texi: Remove -mcet documentation.
18738 * doc/invoke.texi: Likewise.
18740 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
18743 * doc/install.texi: Remove -mcet from bootstrap-cet.
18745 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18748 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
18749 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
18753 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
18754 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
18755 containing a CONST_VECTOR.
18757 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
18759 * doc/install.texi: Update newlib dependency for nvptx.
18761 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18764 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
18765 instead of INTVAL when shifting x left.
18767 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
18769 PR tree-optimization/85478
18770 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
18771 vect_grouped_store_supported for single element vectors.
18773 2018-04-24 Richard Biener <rguenther@suse.de>
18776 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
18777 load cost increase to the case of non-constant step.
18779 2018-04-24 Jakub Jelinek <jakub@redhat.com>
18782 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
18783 destination if any_malformed_asm.
18785 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
18787 PR middle-end/85496
18788 * expr.c (store_field): In the bitfield case, if the value comes from
18789 a function call and is returned in registers by means of a PARALLEL,
18790 do not change the mode of the temporary unless BLKmode and VOIDmode.
18792 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
18794 PR rtl-optimization/85423
18795 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
18796 dependencies to debug insns when the previous insn is non-debug.
18798 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18800 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
18801 enums into a single definition.
18802 (fls): Fix predicates and printing.
18805 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
18807 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
18808 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
18809 and short u6 immediate.
18810 (check_if_valid_sleep_operand): Remove.
18811 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
18813 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18815 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
18816 flag_always_save_lp condition.
18817 * config/nds32/nds32.opt (malways-save-lp): New option.
18819 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18821 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
18822 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
18823 * config/nds32/nds32.h
18824 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
18825 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
18827 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18829 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
18830 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
18832 2018-04-22 Shiva Chen <shiva0217@gmail.com>
18833 Chung-Ju Wu <jasonwucj@gmail.com>
18835 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
18837 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
18838 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
18840 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18842 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
18844 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18846 * config/nds32/nds32-protos.h (nds32_data_alignment,
18847 nds32_local_alignment): Declare.
18848 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
18849 nds32_local_alignment): New functions.
18850 (TARGET_CONSTANT_ALIGNMENT): Define.
18851 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
18853 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18855 * config/nds32/nds32.c
18856 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
18857 (TARGET_MODES_TIEABLE_P): Likewise.
18859 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
18861 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
18862 level Ofast and Og.
18864 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
18865 Chung-Ju Wu <jasonwucj@gmail.com>
18867 * config/nds32/constants.md (unspec_volatile_element): Add enum values
18868 for unaligned access.
18869 * config/nds32/nds32-intrinsic.c: Implementation of expanding
18871 * config/nds32/nds32-intrinsic.md: Likewise.
18872 * config/nds32/nds32_intrinsic.h: Likewise.
18873 * config/nds32/nds32.h (nds32_builtins): Likewise.
18874 * config/nds32/nds32.opt (munaligned-access): New option.
18875 * config/nds32/nds32.c (nds32_asm_file_start): Display
18876 flag_unaligned_access status.
18878 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
18880 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
18881 -mno-relax is present.
18882 * config/riscv/linux.h (LINK_SPEC): Ditto.
18884 2018-04-20 Martin Sebor <msebor@redhat.com>
18887 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
18889 (gimple_fold_builtin_stxcpy_chk): Same.
18890 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
18892 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
18895 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
18896 __powikf2 when long double is IEEE 128-bit.
18898 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
18900 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
18901 step to make sure stack always aligned.
18903 2018-04-20 Carl Love <cel@us.ibm.com>
18906 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
18907 size check for arg0.
18909 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
18910 Tom de Vries <tom@codesourcery.com>
18913 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
18914 Emit insns for calls too.
18915 (nvptx_find_par): Always look for worker-level predecessor insn.
18916 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
18918 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
18919 (nvptx_process_pars): Propagate frames for calls.
18921 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
18924 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
18926 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
18927 (ix86_handle_option): Don't handle OPT_mibt.
18928 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
18930 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
18932 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
18933 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
18934 (ix86_target_macros): Define __CET__ with flag_cf_protection
18935 for -fcf-protection.
18936 * config/i386/i386.c (isa2_opts): Remove -mibt.
18937 * config/i386/i386.h (TARGET_IBT): Removed.
18938 (TARGET_IBT_P): Likewise.
18939 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
18940 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
18941 * config/i386/i386.opt (mcet): Update help message.
18942 (mshstk): Likewise.
18944 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
18945 -mcet as an alias for -mshstk.
18947 2018-04-20 Richard Biener <rguenther@suse.de>
18949 PR middle-end/85475
18950 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
18951 complexity by forcing a single use of the multiply operand.
18953 2018-04-20 Martin Jambor <mjambor@suse.cz>
18956 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
18957 recursion dependency to only apply to non-clones.
18959 2018-04-20 Martin Jambor <mjambor@suse.cz>
18962 * ipa-cp.c (create_specialized_node): Check that clones of
18963 self-recursive edges exist during IPA-CP.
18965 2018-04-19 Toon Moene <toon@moene.org>
18967 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
18970 2018-04-19 Jakub Jelinek <jakub@redhat.com>
18972 PR tree-optimization/85467
18973 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
18974 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
18975 VECTOR_CST element to type.
18977 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18980 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
18981 * config/i386/i386.md (builtin_setjmp_setup): Removed.
18982 (builtin_longjmp): Likewise.
18983 (save_stack_nonlocal): New pattern.
18984 (restore_stack_nonlocal): Likewise.
18986 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18989 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
18990 Replace ASM_OUTPUT_LABEL with fprintf.
18992 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
18995 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
18996 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
18997 * config/i386/i386-c.c (ix86_target_macros_internal): Also
18998 define __IBT__ and __SHSTK__ for -fcf-protection.
18999 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
19001 (ix86_trampoline_init): Likewise.
19002 (x86_output_mi_thunk): Likewise.
19003 (ix86_notrack_prefixed_insn_p): Likewise.
19004 (ix86_option_override_internal): Don't disallow -fcf-protection.
19005 * config/i386/i386.md (rdssp<mode>): Also enable for
19007 (incssp<mode>): Likewise.
19008 (nop_endbr): Likewise.
19009 * config/i386/i386.opt (mcet): Change help message to built-in
19012 (mshstk): Likewise.
19013 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
19014 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
19015 enable CET built-in functions.
19017 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
19019 * common/config/i386/i386-common.c
19020 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
19021 OPTION_MASK_ISA_MOVDIRI_UNSET,
19022 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
19023 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
19024 * config.gcc (movdirintrin.h): New header.
19025 * config/i386/cpuid.h (bit_MOVDIRI,
19026 bit_MOVDIR64B): New bits.
19027 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
19029 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
19030 (VOID, PVOID, PCVOID)): New function types.
19031 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
19032 __builtin_ia32_directstoreu_u64,
19033 __builtin_ia32_movdir64b): New builtins.
19034 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
19035 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
19037 (ix86_valid_target_attribute_inner_p): Ditto.
19038 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
19039 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
19040 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
19041 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
19042 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
19043 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
19044 (movdiri<mode>, movdir64b_<mode>): New.
19045 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
19046 * config/i386/immintrin.h: Include movdirintrin.h.
19047 * config/i386/movdirintrin.h: New file.
19048 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
19050 2018-04-19 Richard Biener <rguenther@suse.de>
19052 PR middle-end/85455
19053 * cfg.c (clear_bb_flags): When loop state says we have
19054 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
19056 2018-04-19 Richard Biener <rguenther@suse.de>
19058 PR tree-optimization/84737
19059 * tree-vect-data-refs.c (vect_copy_ref_info): New function
19060 copying restrict info.
19061 (vect_setup_realignment): Use it.
19062 * tree-vectorizer.h (vect_copy_ref_info): Declare.
19063 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
19064 the first DR to all generated stores.
19065 (vectorizable_load): Likewise for loads.
19067 2018-04-19 Jakub Jelinek <jakub@redhat.com>
19069 PR tree-optimization/85446
19070 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
19071 the integral and pointer types to have the same precision.
19073 * doc/install.texi: Document --disable-cet being the default and
19076 2018-04-18 Martin Liska <mliska@suse.cz>
19078 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
19081 2018-04-18 Martin Liska <mliska@suse.cz>
19084 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
19087 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
19088 arguments if they are comparable.
19090 2018-04-18 Martin Liska <mliska@suse.cz>
19093 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
19096 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
19099 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
19102 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
19103 ENDBR after calling __morestack.
19105 2018-04-18 David Malcolm <dmalcolm@redhat.com>
19108 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
19109 by using gcc_base_ver to generate a gcc_driver_version, and use
19110 it when generating GCC_DRIVER_NAME.
19111 * configure: Regenerate.
19113 2018-04-18 Jakub Jelinek <jakub@redhat.com>
19116 * config.gcc: Obsolete powerpc*-*-*spe*.
19118 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19121 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
19122 (stabstr_D): Change type of unum from unsigned int to
19123 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
19126 2018-04-17 Jim Wilson <jimw@sifive.com>
19129 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
19130 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
19131 Set arg_pointer_offset after using pretend_args_size.
19133 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19135 PR rtl-optimization/85431
19136 * dse.c (record_store): Ignore zero width stores.
19139 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
19140 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
19141 __builtin_stack_restore rather than after it.
19142 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
19143 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
19144 argument instead of virtual_dynamic_stack_rtx.
19146 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
19148 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
19150 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
19151 Add note to error message to explain internal mapping of overloaded
19152 built-in function name to non-overloaded built-in function name.
19153 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
19156 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
19159 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
19160 where the inputs overlap with the output.
19162 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19164 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
19165 (=v, v) alternative and explicit "memory" attribute.
19166 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
19167 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
19169 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
19170 "sselog1" type instead of "sselog".
19171 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
19172 "sselog". Remove explicit "memory" attribute.
19173 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
19174 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
19176 (vec_extract_hi_v32hi): Merge all alternatives into one, use
19177 "sselog1" type instead of "sselog". Remove explicit "memory"
19179 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
19180 use "sselog1" type instead of "sselog". Remove explicit "memory"
19182 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
19183 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
19185 (vec_extract_hi_v64qi): Merge all alternatives into one, use
19186 "sselog1" type instead of "sselog". Remove explicit "memory"
19188 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
19189 use "sselog1" type instead of "sselog". Remove explicit "memory"
19193 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
19195 PR middle-end/85414
19196 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
19199 2018-04-17 Martin Jambor <mjambor@suse.cz>
19202 * ipa-cp.c (create_specialized_node): Call
19203 expand_all_artificial_thunks if necessary.
19205 2018-04-17 Martin Liska <mliska@suse.cz>
19208 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
19209 in message, remote space in between '_G' and '('.
19211 2018-04-17 Jakub Jelinek <jakub@redhat.com>
19214 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
19215 avx512f_vmcmp<mode>3<round_saeonly_name>,
19216 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
19217 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
19218 avx512f_rndscale<mode><round_saeonly_name>,
19219 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
19220 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
19221 Use %<iptr>2 instead of %2 for -masm=intel.
19222 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
19223 avx512f_vcvttss2usi<round_saeonly_name>,
19224 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
19226 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
19227 avx512f_vcvttsd2usi<round_saeonly_name>,
19228 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
19229 Use %q1 instead of %1 for -masm=intel.
19230 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
19231 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
19232 of %3 for -masm=intel.
19233 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
19235 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
19237 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
19239 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
19240 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
19242 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
19244 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
19245 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
19246 %g1 and one with %0 and %1.
19247 (avx512er_vmrcp28<mode><round_saeonly_name>,
19248 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
19249 %1 for -masm=intel.
19250 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
19251 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
19252 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
19253 of %0 and %{%4%} for -masm=intel.
19254 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
19255 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
19256 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
19257 order of %0 and %{%5%}%{z%} for -masm=intel.
19259 2018-04-17 Jan Hubicka <jh@suse.cz>
19262 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
19264 2018-04-17 Martin Liska <mliska@suse.cz>
19267 * multiple_target.c (create_dispatcher_calls): Set apostrophes
19268 for target_clone error message. Make default implementation
19269 clone to be a local declaration.
19270 (separate_attrs): Add new argument and check for an empty
19272 (expand_target_clones): Handle it.
19273 (ipa_target_clone): Make redirection just for target_clones
19276 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
19277 Tom de Vries <tom@codesourcery.com>
19279 PR middle-end/84955
19280 * omp-expand.c (expand_oacc_for): Add dummy false branch for
19281 tiled basic blocks without omp continue statements.
19283 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
19286 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
19287 vec_extract expression as having side effects to make sure it gets
19290 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
19293 * config/i386/i386.c (get_builtin_code_for_version): Check
19296 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
19299 * config.gcc: Support "skylake".
19300 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19302 * config/i386/i386.c (m_SKYLAKE): Define.
19303 (processor_target_table): Add "skylake".
19304 (ix86_option_override_internal): Add "skylake".
19305 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
19306 PROCESSOR_CANNONLAKE.
19307 (get_builtin_code_for_version): Fix priority for
19308 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
19309 PROCESSOR_SKYLAKE-AVX512.
19310 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
19311 (processor_type): Add PROCESSOR_SKYLAKE.
19313 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
19314 Jason Merrill <jason@redhat.com>
19317 * convert.c (convert_to_integer_1): Use direct recursion for
19318 enumeral types and types with a precision less than the number
19319 of bits in their mode.
19321 2018-04-16 Julia Koval <julia.koval@intel.com>
19324 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
19325 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
19327 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
19330 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
19331 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
19332 and -mno-direct-move.
19334 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
19337 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
19338 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
19339 Ensure negative shifts result in {0}.
19341 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
19343 PR rtl-optimization/79916
19344 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
19345 regs (if any) to define how to gnerate SD moves when LRA is in
19348 2018-04-13 Jakub Jelinek <jakub@redhat.com>
19350 PR rtl-optimization/85393
19351 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
19352 * except.c (expand_dw2_landing_pad_for_region): Make static.
19353 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
19354 a label and unconditional jump to old_bb, rather than
19355 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
19358 PR rtl-optimization/85376
19359 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
19360 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
19361 instead of a specific value.
19363 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
19364 Bin Cheng <bin.cheng@arm.com>
19366 PR tree-optimization/82965
19367 PR tree-optimization/83991
19368 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
19369 by_profile_only parameter.
19370 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
19371 information if the loop was predicted to iterate too many times.
19372 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
19374 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
19377 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
19380 2018-04-13 Martin Liska <mliska@suse.cz>
19381 Jakub Jelinek <jakub@redhat.com>
19383 PR middle-end/81657
19384 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
19385 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
19386 * builtins.c (expand_builtin_memory_copy_args): Use
19387 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
19388 handle dest_addr == pc_rtx.
19390 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
19393 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
19394 asked to not generate direct moves.
19395 (fix_trunc<mode>si2_stfiwx): Similar.
19396 (fix_trunc<mode>si2_internal): Similar.
19398 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19401 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
19402 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
19403 lookup if dest in some wider mode is known to be const0_rtx and
19404 if so, record permanent equivalence for it to be ZERO_EXTEND of
19405 the narrower mode destination.
19407 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19409 * lto-streamer-out.c (output_function): Revert 259346.
19410 * omp-expand.c (expand_oacc_for): Likewise.
19412 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
19414 PR rtl-optimization/85354
19415 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
19416 * sel-sched.c (sel_global_init): ... here.
19418 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
19421 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
19422 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
19423 mode for PE-COFF targets.
19424 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
19425 (i386_pe_asm_lto_end): Likewise.
19426 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
19427 (TARGET_ASM_LTO_END): Likewise.
19428 * config/i386/winnt.c (saved_debug_info_level): New static variable.
19429 (i386_pe_asm_lto_start): New function.
19430 (i386_pe_asm_lto_end): Likewise.
19432 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
19433 Richard Biener <rguenther@suse.de>
19435 PR middle-end/84955
19436 * lto-streamer-out.c (output_function): Fix CFG loop state before
19438 * omp-expand.c (expand_oacc_for): Handle calls to internal
19439 functions like regular functions.
19441 2018-04-12 Richard Biener <rguenther@suse.de>
19444 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
19445 for the early LTO debug to properly generate references to it
19446 during DIE emission. Do not re-use that for the skeleton for
19448 (dwarf2out_early_finish): Likewise.
19450 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19453 * config/i386/sse.md
19454 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
19455 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
19456 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
19457 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
19458 and output is a reg, avoid creating invalid lowpart subreg, but
19459 instead split into a 512-bit move. Don't split if not AVX512VL,
19460 input is xmm16+ reg and output is a mem.
19461 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
19462 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
19463 xmm16+ reg and output is a mem.
19465 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19467 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
19468 also for flag_dwarf2_cfi_asm.
19470 2018-04-12 Jakub Jelinek <jakub@redhat.com>
19472 PR rtl-optimization/85342
19473 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
19474 a bool scalar var inside of the loop instead. Don't try to update
19475 recog_data.operand after failed apply_change_group.
19477 2018-04-12 Tom de Vries <tom@codesourcery.com>
19480 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
19481 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
19482 array with flexible array member as array without given dimension.
19483 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
19484 argument for undefined param to true.
19486 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
19489 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19490 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
19491 from PowerPC section.
19492 * config/rs6000/sysv4.opt (mcall-): Improve help text.
19493 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
19494 help text that is too long.
19495 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
19496 help text that is too long.
19497 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
19498 help text that is too long.
19500 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
19502 * config/alpha/alpha.md (stack_probe_internal): Rename
19503 from "probe_stack". Update all callers.
19505 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19507 PR rtl-optimization/84566
19508 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
19509 sched_macro_fuse_insns.
19511 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19514 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
19515 (compute_block_dependences): ... from here.
19517 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19519 PR tree-optimization/85331
19520 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
19521 from int to HOST_WIDE_INT.
19523 2018-04-11 Martin Jambor <mjambor@suse.cz>
19526 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
19527 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
19528 not the same as the source val.
19529 (cgraph_edge_brings_value_p): New parameter.
19530 (gather_edges_for_value): Pass destination value to
19531 cgraph_edge_brings_value_p.
19532 (perhaps_add_new_callers): Likewise.
19533 (get_info_about_necessary_edges): Likewise and exclude values brought
19534 only by self-recursive edges.
19535 (create_specialized_node): Redirect only clones of self-calling edges.
19536 (+self_recursive_pass_through_p): New function.
19537 (find_more_scalar_values_for_callers_subset): Use it.
19538 (find_aggregate_values_for_callers_subset): Likewise.
19539 (known_aggs_to_agg_replacement_list): Removed.
19540 (decide_whether_version_node): Re-calculate known constants for all
19541 remaining context clones.
19543 2018-04-11 Richard Biener <rguenther@suse.de>
19546 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
19547 from early DWARF output.
19548 (dwarf2out_early_finish): Output line info unconditionally into
19549 early DWARF and add reference to it.
19551 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19554 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
19555 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
19556 other than V2DFmode using iptr mode attribute.
19557 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
19559 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
19561 PR rtl-optimization/84659
19562 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
19564 2018-04-11 Jakub Jelinek <jakub@redhat.com>
19567 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
19569 (output_loc_list): Pass address of a dummy size variable even in the
19570 locview handling loop.
19571 (index_location_lists): Add comment on why skip_loc_list_entry can't
19574 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
19577 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
19580 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
19583 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
19584 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
19585 and -mstring-compare-inline-limit.
19587 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19590 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
19591 for stack clash protection in a register whenever we need it to be in
19594 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19596 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
19597 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
19599 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
19602 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
19604 (mlong-double-): Ditto.
19605 * config/rs6000/sysv4.opt (msdata=): Ditto.
19606 (mtls-size=): Ditto.
19608 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19610 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
19611 erroneous entries for
19612 "vector int vec_ldl (int, long int *)", and
19613 "vector unsigned int vec_ldl (int, unsigned long int *)".
19614 Add comments and entries for
19615 "vector bool char vec_ldl (int, bool char *)",
19616 "vector bool short vec_ldl (int, bool short *)",
19617 "vector bool int vec_ldl (int, bool int *)",
19618 "vector bool long long vec_ldl (int, bool long long *)",
19619 "vector pixel vec_ldl (int, pixel *)",
19620 "vector long long vec_ldl (int, long long *)",
19621 "vector unsigned long long vec_ldl (int, unsigned long long *)".
19622 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
19623 type tree bool_long_long_type_node and correct definition of
19624 bool_V2DI_type_node to make reference to this new type tree.
19625 (rs6000_mangle_type): Replace erroneous reference to
19626 bool_long_type_node with bool_long_long_type_node.
19627 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
19628 comments to emphasize sign distinctions for char and int types and
19629 replace RS6000_BTI_bool_long constant with
19630 RS6000_BTI_bool_long_long constant. Also add comment to restrict
19631 use of RS6000_BTI_pixel.
19632 (bool_long_type_node): Remove this macro definition.
19633 (bool_long_long_type_node): New macro definition
19635 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19637 PR rtl-optimization/85300
19638 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
19639 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
19640 simplify_unary_operation fails.
19642 2018-04-10 Martin Liska <mliska@suse.cz>
19644 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
19645 cgraph_edge and ipa_ref.
19647 2018-04-10 Jakub Jelinek <jakub@redhat.com>
19651 * config/i386/sse.md
19652 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
19653 computation of the VEC_MERGE selector from mask.
19654 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
19655 Fix decoding of the VEC_MERGE selector into mask.
19657 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19659 PR tree-optimization/85286
19660 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
19662 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
19664 * final.c (final_1): Set insn_last_address as well as
19665 insn_current_address.
19667 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19670 * explow.c (emit_stack_probe): Call validize_mem on memory location
19671 before passing it to gen_probe_stack. Create address operand and
19672 legitimize it for the probe_stack_address case.
19674 2018-04-09 Jan Hubicka <jh@suse.cz>
19677 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
19678 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
19679 * tree.c (free_lang_data_in_type): Fix handling of binfos;
19681 (free_lang_data): Rebuild type inheritance graph.
19683 2018-04-09 Martin Sebor <msebor@redhat.com>
19685 * invoke.texi (-finline-small-functions): Mention other optimization
19687 (-findirect-inlining, -fpartial-inlining): Same.
19688 (-finline-functions-called-once): Same.
19689 (-freorder-blocks-and-partition): Same.
19691 2018-04-09 Jan Hubicka <jh@suse.cz>
19694 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
19695 jumps; choose last target that matches the criteria (i.e.
19696 no partition changes for non-crossing jumps).
19697 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
19698 support for redirecting crossing jumps to non-crossing.
19700 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
19702 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
19703 also for naked functions.
19705 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
19707 * config/arc/arc.md (add_shift): New pattern.
19708 (add_shift2): Likewise.
19709 (sub_shift): Likewise.
19710 (sub_shift_cmp0_noout): Likewise.
19711 (compare_si_ashiftsi): Likewise.
19712 (xbfu_cmp0_noout): New combine pattern.
19713 (xbfu_cmp0"): Likewise.
19714 (movsi_set_cc_insn): Place the predicable variant first.
19715 (commutative_binary_cmp0_noout): Remove clobber.
19716 (commutative_binary_cmp0): New pattern.
19717 (noncommutative_binary_cmp0): Likewise.
19718 (noncommutative_binary_cmp0_noout): Likewise.
19719 (noncommutative_binary_comparison_result_used): Removed.
19720 (rsub_cmp0): New pattern.
19721 (rsub_cmp0_noout): Likewise.
19722 (extzvsi): Changed, keep only meaningful variants.
19723 (SQH, SEZ): New iterators.
19724 (SQH_postfix): New mode attribute.
19725 (SEZ_prefix): New code attribute.
19726 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
19727 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
19728 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
19729 of numerical value.
19730 (noncommutative_operator): Check the availability of barrel
19733 2018-04-09 Richard Biener <rguenther@suse.de>
19735 PR tree-optimization/85284
19736 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
19737 Only use the niter constraining form of simple_iv when the exit
19738 is always executed.
19740 2018-04-09 Tom de Vries <tom@codesourcery.com>
19743 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
19744 (define_expand "*memory_barrier"): New define_expand.
19745 (define_insn "memory_barrier"): New insn.
19747 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19749 PR rtl-optimization/80463
19750 PR rtl-optimization/83972
19751 PR rtl-optimization/83480
19753 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
19754 correct producer for the insn.
19755 (tidy_control_flow): Fixup seqnos in case of debug insns.
19757 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19759 PR rtl-optimization/83913
19761 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
19762 different sched-times when merging exprs.
19764 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19766 PR rtl-optimization/83962
19768 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
19769 tidy_fallthru_edge and tidy_control_flow.
19771 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
19773 PR rtl-optimization/83530
19775 * sel-sched.c (force_next_insn): New global variable.
19776 (remove_insn_for_debug): When force_next_insn is true, also leave only
19777 next insn in the ready list.
19778 (sel_sched_region): When the region wasn't scheduled, make another pass
19779 over it with force_next_insn set to 1.
19781 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19783 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
19785 * config/nds32/constants.md (unspec_volatile_element): Add enum values
19786 for interrupt control.
19787 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19788 functions for interrupt control.
19789 * config/nds32/nds32-intrinsic.md: Likewise.
19790 * config/nds32/nds32_intrinsic.h: Likewise.
19791 * config/nds32/nds32.h (nds32_builtins): Likewise.
19793 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19795 * config/nds32/nds32.c (nds32_init_machine_status,
19796 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
19797 strict_aligned_p field.
19798 (nds32_expand_to_rtl_hook): New function.
19799 (TARGET_EXPAND_TO_RTL_HOOK): Define.
19800 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
19802 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19803 Chung-Ju Wu <jasonwucj@gmail.com>
19805 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
19806 * config/nds32/nds32-n7.md: New file.
19807 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
19808 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
19810 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
19811 * config/nds32/nds32.md (pipeline_model): Add n7.
19812 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
19813 * config/nds32/pipelines.md: Include n7 settings.
19815 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19816 Chung-Ju Wu <jasonwucj@gmail.com>
19818 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
19819 * config/nds32/nds32-e8.md: New file.
19820 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
19821 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
19823 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
19824 * config/nds32/nds32.md (pipeline_model): Add e8.
19825 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
19826 * config/nds32/pipelines.md: Include e8 settings.
19828 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19829 Chung-Ju Wu <jasonwucj@gmail.com>
19831 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
19832 * config/nds32/nds32-n8.md: New file.
19833 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
19834 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
19836 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
19837 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
19838 * config/nds32/nds32.md (pipeline_model): Add n8.
19839 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
19840 * config/nds32/pipelines.md: Include n8 settings.
19842 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
19843 Chung-Ju Wu <jasonwucj@gmail.com>
19845 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
19846 * config/nds32/nds32-n9-2r1w.md: New file.
19847 * config/nds32/nds32-n9-3r2w.md: New file.
19848 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
19849 nds32_register_ports): New or modify for cpu n9.
19850 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
19852 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
19853 * config/nds32/nds32-utils.c: New file.
19854 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
19855 TARGET_MUL_SLOW): Define.
19856 * config/nds32/nds32.md (pipeline_model): New attribute.
19857 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
19858 New options that support cpu n9.
19859 * config/nds32/pipelines.md: Include n9 settings.
19860 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
19862 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
19864 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
19865 information if necessary.
19866 (output_cond_branch_compare_zero): Likewise.
19867 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
19868 (nds32_target_alignment): Refine for alignment.
19869 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
19870 (FUNCTION_BOUNDARY): Modify.
19871 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
19873 * config/nds32/nds32.opt (malways-align, malign-functions): New.
19875 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
19877 * config/nds32/constants.md (unspec_volatile_element): Add values for
19878 TLB operation and data prefetch.
19879 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
19880 functions for TLB operation and data prefetch.
19881 * config/nds32/nds32-intrinsic.md: Likewise.
19882 * config/nds32/nds32_intrinsic.h: Likewise.
19883 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
19884 (nds32_print_operand): Likewise.
19885 * config/nds32/nds32.h (nds32_builtins): Likewise.
19887 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
19888 Andrew Pinski <pinsika@gcc.gnu.org>
19890 PR middle-end/82976
19891 * match.pd: Use constant_boolean_node of correct type instead of
19892 boolean_true_node or boolean_false_node for simplifying
19893 pointer comparisons to zero.
19895 2018-04-07 Jakub Jelinek <jakub@redhat.com>
19897 PR tree-optimization/80021
19898 * tree.c (verify_type_variant): Make error call in verify_variant_match
19899 translatable and remove final full stop.
19901 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19903 * config/nds32/constants.md (unspec_volatile_element): Add
19904 UNSPEC_VOLATILE_EH_RETURN.
19905 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
19906 nds32_output_stack_pop): Support dwarf exception handling process.
19907 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
19908 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
19909 exception handling process.
19910 (nds32_compute_stack_frame): Likewise.
19911 (nds32_return_addr_rtx): Likewise.
19912 (nds32_initial_elimination_offset): Likewise.
19913 (nds32_expand_prologue): Likewise.
19914 (nds32_expand_epilogue): Likewise.
19915 (nds32_dynamic_chain_address): New function.
19916 * config/nds32/nds32.h (machine_function): Add fields for dwarf
19917 exception handling.
19918 (DYNAMIC_CHAIN_ADDRESS): Define.
19919 (EH_RETURN_DATA_REGNO): Define.
19920 (EH_RETURN_STACKADJ_RTX): Define.
19921 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
19922 patterns for dwarf exception handling.
19924 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19926 * config/nds32/nds32.h: Clean up obsolete macros.
19928 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19930 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19931 Add enum values for particular instructions.
19932 * config/nds32/nds32-intrinsic.c: Implementation of expanding
19933 particular intrinsic functions.
19934 * config/nds32/nds32-intrinsic.md: Likewise.
19935 * config/nds32/nds32_intrinsic.h: Likewise.
19936 * config/nds32/nds32.h (nds32_builtins): Likewise.
19937 * config/nds32/nds32.md (type): Add pbsad and pbsada.
19938 (btst, ave): New patterns for particular instructions.
19940 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19942 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
19943 Add enum values for atomic load/store and memory sync.
19944 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
19946 * config/nds32/nds32-intrinsic.md: Likewise.
19947 * config/nds32/nds32_intrinsic.h: Likewise.
19948 * config/nds32/nds32.h (nds32_builtins): Likewise.
19950 2018-04-07 Jakub Jelinek <jakub@redhat.com>
19952 PR tree-optimization/85257
19953 * fold-const.c (native_encode_vector): If not all elts could fit
19954 and off is -1, return 0 rather than offset.
19955 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
19956 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
19957 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
19958 adjust buffer in native_interpret_expr call.
19960 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19962 * config/nds32/constants.md (unspec_volatile_element): Add cache
19963 control enum values.
19964 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
19965 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
19966 * config/nds32/nds32.c (nds32_cctl_names): New.
19967 (nds32_print_operand): Handle cache control register names.
19968 * config/nds32/nds32.h (nds32_builtins): New enum values.
19969 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
19971 * config/nds32/nds32.md (type): Add mmu.
19972 * config/nds32/pipelines.md (simple_insn): Add mmu.
19974 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
19976 * config/nds32/nds32.md (type): Remove call.
19977 * config/nds32/pipelines.md (simple_insn): Likewise.
19979 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
19981 * config/nds32/constants.md (unspec_volatile_element): Add
19982 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
19983 UNSPEC_VOLATILE_FMFCFG.
19984 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
19985 description for fmfcfg and fmfcsr.
19986 (bdesc_1arg): Add fmtcsr.
19987 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
19988 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
19989 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
19990 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
19991 unspec_fmfcfg): New patterns.
19992 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
19993 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
19994 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
19995 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
19996 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
19997 __nds32__fmfcfg): Define.
19999 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
20001 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
20002 intrinsic register names.
20003 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
20004 intrinsic register enum values and macros.
20006 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
20008 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
20009 for load/store addressing form.
20010 (nds32_print_operand_address): Likewise.
20012 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
20015 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
20016 based on LABEL_REF. Remove useless assertion.
20017 (pic_address_needs_scratch): Fix formatting.
20018 (sparc_legitimize_pic_address): Minor tweaks.
20019 (sparc_delegitimize_address): Adjust assertion accordingly.
20020 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
20021 into symbolic_operand.
20022 (movsi_high_pic_label_ref): Likewise.
20023 (movsi_lo_sum_pic_label_ref): Likewise.
20024 (movdi_pic_label_ref): Likewise.
20025 (movdi_high_pic_label_ref): Likewise.
20026 (movdi_lo_sum_pic_label_ref): Likewise.
20028 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
20030 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
20031 custom LIB_SPEC setup.
20033 2018-04-06 Ruslan Bukin <br@bsdpad.com>
20034 Kito Cheng <kito.cheng@gmail.com>
20036 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
20037 * config/riscv/freebsd.h: New.
20039 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
20041 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
20042 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
20045 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
20046 Kito Cheng <kito.cheng@gmail.com>
20048 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
20049 nds32_output_call, nds32_symbol_binds_local_p): New functions.
20050 * config/nds32/nds32-protos.h (nds32_output_call,
20051 nds32_output_return): Declare.
20052 * config/nds32/nds32.md: Refine all the call and return patterns.
20054 2018-04-06 Jakub Jelinek <jakub@redhat.com>
20057 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
20058 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
20060 PR rtl-optimization/84872
20061 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
20062 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
20063 EDGE_CROSSING edge.
20065 2018-04-06 Tamar Christina <tamar.christina@arm.com>
20067 * expr.c (copy_blkmode_to_reg): Revert 254862.
20068 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
20070 2018-04-06 Richard Biener <rguenther@suse.de>
20072 PR middle-end/85244
20073 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
20074 after seeing a component reference with an adjacent field. Treat
20075 refs to arrays at struct end of external decls similar to
20076 refs to unconstrained commons.
20078 2018-04-06 Jakub Jelinek <jakub@redhat.com>
20081 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
20082 look through SAVE_EXPRs with non-side-effects argument. Adjust
20084 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
20087 2018-04-06 Richard Biener <rguenther@suse.de>
20089 PR middle-end/85180
20090 * alias.c (find_base_term): New wrapper around find_base_term
20091 unwinding CSELIB_VAL_PTR changes.
20092 (find_base_term): Do not restore CSELIB_VAL_PTR during the
20095 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20097 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
20099 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
20100 constant definitions.
20101 ("nop"): lr 0,0 -> nopr r0
20102 ("nop_lr0", "nop_lr1"): New insn definitions.
20104 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
20106 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
20107 NDS32_V3PUSH_AVAILABLE_P macro.
20109 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
20110 Chung-Ju Wu <jasonwucj@gmail.com>
20112 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
20113 (nds32*-*-*): Add float and fpu_config into supported_defaults.
20114 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
20115 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
20116 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
20117 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
20118 * config/nds32/constraints.md: New constraints and checking for hard
20119 float configuration.
20120 * config/nds32/iterators.md: New mode iterator and attribute for hard
20121 float configuration.
20122 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
20124 * config/nds32/nds32-fpu.md: New file.
20125 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
20126 deal with hard float code generation.
20127 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
20129 (abi_type, float_reg_number): New enum type.
20130 * config/nds32/nds32-predicates.c: New predicates for hard float.
20131 * config/nds32/nds32-protos.h: Declare functions for hard float.
20132 * config/nds32/nds32.c: Implementation for hard float configuration.
20133 * config/nds32/nds32.h: Definitions for hard float configuration.
20134 * config/nds32/nds32.md: Include hard float machine description and
20135 modify patterns for hard float configuration.
20136 * config/nds32/nds32.opt: New options for hard float configuration.
20137 * config/nds32/predicates.md: New predicates for hard float
20140 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20142 * common/config/nds32/nds32-common.c
20143 (nds32_option_optimization_table): Enable -mreleax-hint by default.
20145 2018-04-05 Jakub Jelinek <jakub@redhat.com>
20147 PR middle-end/85195
20148 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
20149 CONSTRUCTOR_ELT (ctor, ...)->value.
20151 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
20154 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
20156 2018-04-05 Tom de Vries <tom@codesourcery.com>
20159 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
20162 2018-04-05 Shiva Chen <shiva0217@gmail.com>
20163 Kito Cheng <kito.cheng@gmail.com>
20165 * config/nds32/constraints.md (U33): Fine-tune checking condition.
20166 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
20167 * config/nds32/nds32.h (nds32_16bit_address_type): Add
20168 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
20170 2018-04-05 Shiva Chen <shiva0217@gmail.com>
20171 Kito Cheng <kito.cheng@gmail.com>
20173 * config/nds32/constraints.md (Ufe): New memory constraint.
20174 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
20175 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
20176 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
20178 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
20179 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
20181 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20183 * config/nds32/nds32.md: Use optimize_size in the condition for
20184 alu-shift instructions.
20186 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20188 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
20190 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20192 * config/nds32/nds32.md (negsi2): Refine pattern.
20194 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
20195 Chung-Ju Wu <jasonwucj@gmail.com>
20197 * config/nds32/iterators.md (shift_rotate): New code iterator.
20198 (shift): New code attribute.
20199 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
20200 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
20201 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
20202 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
20203 bit-wise operations.
20204 (andsi3, *andsi3): Ditto.
20205 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
20206 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
20207 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
20208 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
20209 nds32_ior_operand, nds32_xor_operand): New predicates.
20211 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20213 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
20214 (addsi3, subsi3): ... this.
20216 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20218 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
20220 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20222 * config/nds32/nds32.md: Adjust indention.
20224 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
20226 * config/nds32/nds32.md (feature): New attribute.
20228 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
20230 * config/nds32/nds32.md (subtype): New attribute.
20232 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20235 * config/arm/arm-builtins.c (arm_expand_builtin): Change
20236 expansion to perform a bitwise AND of the argument followed by a
20237 boolean negation of the result.
20239 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
20241 PR rtl-optimization/84878
20242 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
20243 the basic block. Assert the use reference is not artificial and that
20244 it has an associated insn.
20246 2018-04-04 Michael Matz <matz@suse.de>
20248 * builtins.c (compute_objsize): Pass correct operand
20249 to array_at_struct_end_p.
20251 2018-04-04 Richard Biener <rguenther@suse.de>
20254 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
20255 from contexts for DINFO_LEVEL_TERSE and below.
20257 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
20259 * config/nds32/nds32-doubleword.md (move_<mode>): Require
20260 resiter_operand condition.
20261 * config/nds32/nds32.md (*move<mode>): Ditto.
20263 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
20264 Monk Chiang <sh.chiang04@gmail.com>
20266 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
20268 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20270 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
20272 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20273 Kito Cheng <kito.cheng@gmail.com>
20275 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
20276 nds32_cond_code_str, output_cond_branch,
20277 output_cond_branch_compare_zero, nds32_expand_cbranch,
20278 nds32_expand_cstore, nds32_expand_movcc,
20279 nds32_output_cbranchsi4_equality_zero,
20280 nds32_output_cbranchsi4_equality_reg,
20281 nds32_output_cbranchsi4_equality_reg_or_const_int,
20282 nds32_output_cbranchsi4_greater_less_zero: New functions.
20283 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
20284 nds32_expand_cstore, nds32_expand_movcc,
20285 nds32_output_cbranchsi4_equality_zero,
20286 nds32_output_cbranchsi4_equality_reg,
20287 nds32_output_cbranchsi4_equality_reg_or_const_int,
20288 nds32_output_cbranchsi4_greater_less_zero): Declare.
20289 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
20290 nds32_rimm11s_operand): New predicates.
20291 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
20292 * config/nds32/nds32.md: Rewrite all the branch and conditional move
20295 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
20297 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
20298 * config/nds32/nds32.md: Ditto.
20299 * config/nds32/pipelines.md: Ditto.
20301 2018-04-04 Richard Biener <rguenther@suse.de>
20303 PR tree-optimization/85168
20304 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
20305 propagating abnormals.
20307 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20309 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
20311 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
20312 Kito Cheng <kito.cheng@gmail.com>
20314 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
20315 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
20316 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
20317 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
20318 * config/nds32/nds32.md (sibcall_internal): New.
20319 (sibcall_register): Remove.
20320 (sibcall_immediate): Remove.
20321 (sibcall_value_internal): New.
20322 (sibcall_value_register): Remove.
20323 (sibcall_value_immediate): Remove.
20324 * config/nds32/predicates.md (nds32_general_register_operand): New.
20325 (nds32_call_address_operand): New.
20327 2018-04-03 Jakub Jelinek <jakub@redhat.com>
20329 PR rtl-optimization/85167
20330 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
20331 bb_defs if *split_p, instead preinitialize it to NULL.
20333 PR tree-optimization/85156
20334 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
20335 evaluating the argument multiple times.
20337 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
20339 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
20341 (_mm_cvtpd_ps): Likewise.
20342 (_mm_cvttpd_epi32): Likewise.
20343 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
20344 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
20345 vector, pixel, and bool following altivec.h include.
20347 2018-04-03 Martin Sebor <msebor@redhat.com>
20349 * doc/extend.texi (Common Function Attributes): Clarify.
20350 (const attribute): Likewise.
20351 (pure attribute): Likewise.
20353 2018-04-03 Jakub Jelinek <jakub@redhat.com>
20356 * config/i386/i386.c (ix86_expand_vector_set): Use
20357 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
20359 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
20361 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
20362 instructions when changing rounding bits to preserve precision bits
20363 in the x87 control word.
20365 2018-04-03 Martin Liska <mliska@suse.cz>
20367 PR tree-optimization/82491
20368 * rtl.h (strip_offset_and_add): Replace += suboffset with
20369 poly_uint64 () + suboffset.
20371 2018-03-29 Martin Liska <mliska@suse.cz>
20372 Martin Jambor <mjambor@suse.cz>
20375 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
20376 param_type is not an integral or pointer type.
20378 2018-04-03 Richard Biener <rguenther@suse.de>
20380 * sese.h (recompute_all_dominators): Remove.
20382 2018-04-02 Martin Sebor <msebor@redhat.com>
20384 * doc/invoke.texi (-Wrestrict): Fix typos.
20386 2018-04-02 Jim Wilson <jimw@sifive.com>
20388 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
20389 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
20390 (<optab>di3, <optab>si3_extend): Likewise.
20391 (<optab>si3_mask, <optab>si3_mask_1): New.
20392 (<optab>di3_mask, <optab>di3_mask_1): New.
20393 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
20394 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
20395 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
20397 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
20399 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
20402 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
20404 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
20405 (nds32_canonicalize_comparison): New function.
20407 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20408 Kito Cheng <kito.cheng@gmail.com>
20409 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20411 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
20412 * config/nds32/constants.md (unspec_volatile_element): Add
20413 UNSPEC_VOLATILE_RELAX_GROUP.
20414 * config/nds32/nds32-relax-opt.c: New file.
20415 * config/nds32/nds32-predicates.c
20416 (nds32_symbol_load_store_p): New function.
20417 * config/nds32/nds32-protos.h
20418 (nds32_symbol_load_store_p): Declare function.
20419 (make_pass_nds32_relax_opt): Declare new rtl pass function.
20420 * config/nds32/nds32.c
20421 (nds32_register_pass): New function to register pass.
20422 (nds32_register_passes): New function to register passes.
20423 * config/nds32/nds32.md (relax_group): New pattern.
20424 * config/nds32/nds32.opt (mrelax-hint): New option.
20425 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
20427 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
20429 * config/nds32/t-nds32: Modify files dependency.
20431 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20433 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
20434 (PROFILE_HOOK): Define its implementation.
20436 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
20438 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
20439 type and 32-bit size.
20441 2018-04-01 Jakub Jelinek <jakub@redhat.com>
20443 PR middle-end/85090
20444 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
20445 (V_128_256): New mode iterator.
20446 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
20447 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
20448 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
20450 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
20451 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
20453 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
20456 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
20457 NaN inputs correctly.
20459 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
20462 * config/rs6000/vsx.md (??r): New mode attribute.
20463 (*vsx_mov<mode>_64bit): Use it.
20464 (*vsx_mov<mode>_32bit): Likewise.
20466 2018-03-30 Martin Sebor <msebor@redhat.com>
20468 PR tree-optimization/84818
20469 * builtins.c (check_access): Use warning_n.
20471 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20474 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
20476 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
20479 2018-03-30 Julia Koval <julia.koval@intel.com>
20482 * x86-tune.def (movx, partial_reg_dependency): Enable for
20485 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
20487 PR inline-asm/84985
20488 * lra-constraints.c (process_alt_operands): Move setting
20489 this_alternative_matches below.
20491 2018-03-29 Martin Liska <mliska@suse.cz>
20494 * doc/invoke.texi: Document how LTO works with debug info.
20495 Describe auto-load support of binutils. Mention 'x86-64'
20496 as valid option value of -march option.
20498 2018-03-29 Jakub Jelinek <jakub@redhat.com>
20500 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
20503 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
20504 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
20505 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
20508 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20511 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
20512 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
20513 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
20514 for RS6000_BTM_POWERPC64.
20515 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
20516 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
20517 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
20520 (DIVDEU): Likewise.
20522 2018-03-28 Carl Love <cel@us.ibm.com>
20525 2017-09-27 Carl Love <cel@us.ibm.com>
20527 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
20528 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
20529 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
20532 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20534 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
20535 instead of __vector bool.
20536 (_mm_max_pu8): Likewise.
20537 (_mm_min_pi16): Likewise.
20539 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
20542 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
20543 (DIVWEUO): Likewise.
20544 (DIVDEO): Likewise.
20545 (DIVDEUO): Likewise.
20546 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
20547 DIVWEUO and DIVDEUO.
20548 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
20549 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
20550 (div_extend): Likewise.
20551 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
20553 (__builtin_divweuo): Likewise.
20554 (__builtin_divdeo): Likewise.
20555 (__builtin_divdeuo): Likewise.
20557 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20560 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
20561 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
20563 PR tree-optimization/82004
20564 * gimple-match-head.c (optimize_pow_to_exp): New function.
20565 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
20566 Don't fold to exp if optimize_pow_to_exp is false.
20568 2018-03-28 Martin Liska <mliska@suse.cz>
20571 * calls.c (initialize_argument_information): Fix trailing space.
20572 * common.opt: Fix typo and provide better explanation for
20573 -fsanitize-coverage option.
20574 * config/i386/i386.opt: Fix typo.
20576 2018-03-28 Jakub Jelinek <jakub@redhat.com>
20577 Martin Liska <mliska@suse.cz>
20580 * gimplify.c (asan_poison_variable): Don't do the check for
20581 gimplify_omp_ctxp here.
20582 (gimplify_decl_expr): Do it here.
20583 (gimplify_target_expr): Likewise.
20585 2018-03-28 Martin Liska <mliska@suse.cz>
20588 * config/i386/i386.c (ix86_function_arg_advance): Do not call
20589 chkp_type_bounds_count if MPX is not enabled.
20591 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
20593 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
20595 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
20598 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
20599 function to create the function decl for complex long double
20600 multiply and divide for -mabi=ieeelongdouble.
20601 (init_float128_ieee): Call it.
20603 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
20606 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
20607 -fcf-protection=branch -mibt.
20608 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
20610 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20613 * config/arm/arm.c (arm_valid_symbolic_address): Handle
20614 arm_word_relocations.
20616 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
20619 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
20620 extern array declarations.
20622 2018-03-27 Richard Biener <rguenther@suse.de>
20624 PR middle-end/84067
20625 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
20626 explicit single_use checks.
20628 2018-03-27 Richard Biener <rguenther@suse.de>
20630 PR tree-optimization/85082
20631 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
20634 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20636 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
20637 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
20638 Turn on fasynchronous-unwind-tables and funwind-tables.
20640 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
20643 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
20644 (*bmi_blsr_<mode>_ccz): Ditto.
20646 2018-03-26 Tom de Vries <tom@codesourcery.com>
20648 PR tree-optimization/85063
20649 * omp-general.c (offloading_function_p): New function. Factor out
20651 * omp-offload.c (pass_omp_target_link::gate): ... here.
20652 * omp-general.h (offloading_function_p): Declare.
20653 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
20654 with attribute omp declare target for offloading functions.
20656 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
20658 PR tree-optimization/84005
20659 * tree-data-ref.h (get_base_for_alignment): Declare.
20660 * tree-data-ref.c (get_base_for_alignment_1): New function.
20661 (get_base_for_alignment): Likewise.
20662 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
20663 get_base_for_alignment to find a suitable base object, instead
20664 of always using drb->base_address.
20666 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20668 PR inline-asm/85022
20669 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
20670 known size by default.
20672 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
20674 PR inline-asm/85030
20675 * lra-constraints.c (process_alt_operands): Don't match BLKmode
20676 and non BLKmode operands.
20678 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20681 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
20682 Clean up attributes.
20684 2018-03-23 Richard Biener <rguenther@suse.de>
20687 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
20688 we are going to emit early debug for LTO.
20690 2018-03-23 Jakub Jelinek <jakub@redhat.com>
20692 PR inline-asm/85034
20693 * function.c (match_asm_constraints_1): Don't optimize if input
20694 doesn't satisfy general_operand predicate for output's mode.
20696 PR inline-asm/85022
20697 * alias.c (write_dependence_p): Don't require for x_canonicalized
20698 non-VOIDmode if x has VOIDmode.
20701 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
20702 just don't try to optimize it rather than assert it never happens.
20704 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20706 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
20707 macro expansions for definition of ST_INTERNAL_<mode> and
20708 LD_INTERNAL_<mode> builtins.
20709 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
20711 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
20713 (altivec_expand_st_builtin): Likewise.
20714 (altivec_expand_builtin): Remove calls to deleted functions.
20715 (rs6000_address_for_altivec): Delete this function.
20716 * config/rs6000/vector.md: Remove expands for
20717 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
20719 2018-03-22 Sudakshina Das <sudi.das@arm.com>
20722 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
20723 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
20724 re-computing once computed.
20725 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
20726 (arm_init_machine_status): Initialize
20727 machine->static_chain_stack_bytes.
20729 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
20732 * doc/extend.texi: Add four new prototypes for vec_ld.
20733 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
20734 definitions for more logical presentation.
20735 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
20736 entries for V1TI variants of __builtin_altivec_ld builtin.
20737 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
20738 handling of V1TI variant of LVX icode pattern.
20739 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
20740 (rs6000_gimple_fold_builtin): Likewise.
20741 (altivec_init_builtins): Add code to define
20742 __builtin_altivec_lvx_v1ti function.
20744 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20746 PR inline-asm/84941
20747 * function.c (match_asm_constraints_1): Don't do the optimization
20748 if input isn't a REG, SUBREG, MEM or constant.
20750 2018-03-22 Tom de Vries <tom@codesourcery.com>
20752 PR tree-optimization/84956
20753 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
20754 bb_has_abnormal_pred.
20756 2018-03-22 Jakub Jelinek <jakub@redhat.com>
20759 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
20760 DECL_INITIAL (decl) to decl at the end.
20761 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
20762 adjust the comment.
20764 2018-03-21 Joseph Myers <joseph@codesourcery.com>
20766 * doc/extend.texi (__builtin_tgmath): Document when complex
20767 integer types are treated as _Complex _Float64.
20769 2018-03-21 Tom de Vries <tom@codesourcery.com>
20771 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
20773 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20775 PR tree-optimization/84960
20776 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
20777 if it is ENTRY block, move them into single succ of ENTRY in that case.
20779 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
20781 PR tree-optimization/84811
20782 * poly-int.h (poly_span_traits): Remove the T3 parameter and
20783 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
20784 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
20785 (known_subrange_p): Update accordingly. Cast each value involved
20786 in the size comparison, rather than casting the result of the
20789 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20791 PR tree-optimization/84982
20792 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
20793 by flipping the least significant bit rather than all bits from
20794 bitpos to bitpos + bitsize - 1.
20796 2018-03-21 Nathan Sidwell <nathan@acm.org>
20798 * doc/extend.texi (Deprecated Features): Remove mention of
20799 long-deleted deprecations.
20801 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20804 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
20805 * configure: Regenerate.
20807 2018-03-21 Tom de Vries <tom@codesourcery.com>
20809 PR tree-optimization/83126
20810 * tree-parloops.c (num_phis): New function.
20811 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
20813 2018-03-21 Nathan Sidwell <nathan@acm.org>
20815 * doc/extend.texi (Deprecated Features): Update deprecated flags,
20816 mention anon-struct/union members and trailing attributes.
20818 2018-03-21 Bin Cheng <bin.cheng@arm.com>
20820 PR tree-optimization/84969
20821 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
20822 builtin memset partitions if they set different rhs values.
20824 2018-03-21 Jakub Jelinek <jakub@redhat.com>
20826 PR rtl-optimization/84989
20827 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
20828 VEC_DUPLICATE with scalar result mode.
20830 2018-03-21 Martin Liska <mliska@suse.cz>
20833 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
20834 not intended return statement.
20836 2018-03-21 Martin Liska <mliska@suse.cz>
20839 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
20840 (chkp_find_bound_slots_1): Limit number of iterations.
20842 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
20845 * Minor grammar fixes for x86 options.
20847 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20850 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
20851 holding REG_CFA_RESTORE notes, instead turn them into a USE.
20853 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
20856 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
20857 (altivec_lvx_<mode>_1op): Likewise.
20858 (altivec_stvx_<mode>_2op): Likewise.
20859 (altivec_stvx_<mode>_1op): Likewise.
20860 (altivec_lvx_<VM2:mode>): New define_expand.
20861 (altivec_stvx_<VM2:mode>): Likewise.
20862 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
20863 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20864 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
20865 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
20866 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
20867 (rs6000_gen_lvx): Likewise.
20868 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
20869 (altivec_expand_stv_builtin): Likewise.
20870 (altivec_expand_builtin): Likewise.
20871 * config/rs6000/vector.md: Likewise.
20873 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20876 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
20879 2018-03-20 Richard Biener <rguenther@suse.de>
20882 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
20883 sign-conversions as zero, fall back to standard scalar_stmt
20886 2018-03-20 Martin Liska <mliska@suse.cz>
20889 * predict.c (rebuild_frequencies): Handle case when we have
20890 PROFILE_ABSENT, but flag_guess_branch_prob is false.
20892 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20895 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
20896 flag_section_anchors.
20897 * varasm.c (use_blocks_for_decl_p): Remove hack for
20898 dw2_force_const_mem.
20901 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
20903 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
20904 be created, use lowpart_subreg of operands[0] rather than operands[0]
20906 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
20907 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
20908 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
20909 and n constraint instead of aarch64_shift_imm_di and Usd.
20910 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
20911 (*aarch64_<optab>_reg_minus<mode>3): ... this.
20913 2018-03-20 Sudakshina Das <sudi.das@arm.com>
20916 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
20917 to favor GPR over NEON registers.
20918 (<shift>di3_neon): Likewise.
20920 2018-03-20 Tom de Vries <tom@codesourcery.com>
20923 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
20924 (nvptx_process_pars): Emit bar.sync asap and alap.
20926 2018-03-20 Tom de Vries <tom@codesourcery.com>
20929 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
20930 seen_label if seen_label is already set.
20932 2018-03-20 Jakub Jelinek <jakub@redhat.com>
20935 * config/i386/i386.c (fold_builtin_cpu): For features above 31
20936 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
20937 Use 1U instead of 1. Formatting fixes.
20940 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
20941 instead of TREE_TYPE (s1) for the return value.
20943 2018-03-19 Jakub Jelinek <jakub@redhat.com>
20945 PR tree-optimization/84946
20946 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
20947 bitsize + bitsize in poly_uint64 rather than poly_int64.
20950 * dwarf2asm.c: Include fold-const.c.
20951 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
20952 of decl rather than decl itself.
20954 PR rtl-optimization/84643
20955 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
20957 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
20960 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
20961 calling assemble_variable.
20963 2018-03-19 Sudakshina Das <sudi.das@arm.com>
20966 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
20967 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
20969 2018-03-19 Jim Wilson <jimw@sifive.com>
20972 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
20973 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
20974 (riscv_first_stack_step): Likewise.
20975 (riscv_option_override): Use STACK_BOUNDARY instead of
20976 MIN_STACK_BOUNDARY.
20977 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
20978 MIN_STACK_BOUNDARY.
20979 (BIGGEST_ALIGNMENT): Set to 128.
20980 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
20981 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
20984 2018-03-19 Richard Biener <rguenther@suse.de>
20986 PR tree-optimization/84933
20987 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
20988 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
20990 2018-03-19 Richard Biener <rguenther@suse.de>
20992 PR tree-optimization/84859
20993 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
20994 (cond_if_else_store_replacement): Perform sinking operation on
20995 single-store BBs regardless of MAX_STORES_TO_SINK setting.
20996 Generalize what a BB with a single eligible store is.
20998 2018-03-19 Richard Biener <rguenther@suse.de>
21000 PR tree-optimization/84929
21001 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
21002 chrec_is_positive against non-chrec arg.
21004 2018-03-19 Tamar Christina <tamar.christina@arm.com>
21007 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
21009 2018-03-18 Martin Liska <mliska@suse.cz>
21011 PR rtl-optimization/84635
21012 * regrename.c (build_def_use): Use matches_mode only when
21015 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
21017 PR tree-optimization/84913
21018 * tree-vect-loop.c (vectorizable_reduction): Don't try to
21019 vectorize chains of COND_EXPRs.
21021 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
21023 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
21025 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
21027 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
21029 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
21031 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
21033 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
21034 Kito Cheng <kito.cheng@gmail.com>
21036 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
21037 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
21038 (nds32_adjust_reg_alloc_order): New function.
21039 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
21041 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
21043 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
21044 nds32_print_operand, nds32_print_operand_address): Use
21045 HOST_WIDE_INT_PRINT_DEC instead.
21047 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
21049 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
21051 2018-03-17 Jakub Jelinek <jakub@redhat.com>
21054 * config/i386/i386.c (initial_ix86_tune_features,
21055 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
21056 unsigned long long.
21057 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
21058 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
21059 rather than 1u << ix86_tune. Formatting fix.
21060 (ix86_option_override_internal): Change ix86_arch_mask from
21061 unsigned int to unsigned HOST_WIDE_INT, initialize to
21062 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
21063 (ix86_function_specific_restore): Likewise.
21065 2018-03-16 Jakub Jelinek <jakub@redhat.com>
21068 * postreload.c (reload_combine_recognize_pattern): Perform
21069 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
21070 truncate_int_for_mode the result for the destination's mode.
21073 * hsa-gen.c (mem_type_for_type): Fix comment typo.
21074 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
21076 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
21079 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
21082 * lra-assigns.c (lra_split_hard_reg_for): Don't use
21083 regno_allocno_class_array and sorted_pseudos.
21084 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
21085 insns where regno is used.
21087 2018-03-16 Martin Liska <mliska@suse.cz>
21090 * multiple_target.c (create_dispatcher_calls): Redirect
21091 reference in the symbol table.
21093 2018-03-16 Martin Liska <mliska@suse.cz>
21096 * multiple_target.c (create_dispatcher_calls): Redirect also
21099 2018-03-16 Jakub Jelinek <jakub@redhat.com>
21103 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
21104 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
21105 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
21107 2018-03-16 Julia Koval <julia.koval@intel.com>
21109 * doc/invoke.texi (Skylake Server): Add CLWB.
21110 Cannonlake): Remove CLWB.
21112 2018-03-16 Jakub Jelinek <jakub@redhat.com>
21114 PR tree-optimization/84841
21115 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
21117 (FLOAT_ONE_CONST_TYPE): Define.
21118 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
21119 (sort_by_operand_rank): Put entries with higher constant_type last
21120 rather than first to match comments.
21122 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
21124 * config/nios2/nios2.md (movsi_internal): Fix thinko in
21127 2018-03-15 Jakub Jelinek <jakub@redhat.com>
21130 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
21131 check and use address of target always.
21133 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
21136 * config/i386/i386.c (indirect_thunk_needed): Update comments.
21137 (indirect_thunk_bnd_needed): Likewise.
21138 (indirect_thunks_used): Likewise.
21139 (indirect_thunks_bnd_used): Likewise.
21140 (indirect_return_needed): New.
21141 (indirect_return_bnd_needed): Likewise.
21142 (output_indirect_thunk_function): Add a bool argument for
21144 (output_indirect_thunk_function): Don't generate alias for
21145 function return thunk.
21146 (ix86_code_end): Call output_indirect_thunk_function to generate
21147 function return thunks.
21148 (ix86_output_function_return): Set indirect_return_bnd_needed
21149 and indirect_return_needed instead of indirect_thunk_bnd_needed
21150 and indirect_thunk_needed.
21152 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
21154 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
21155 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
21156 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
21158 2018-03-15 David Malcolm <dmalcolm@redhat.com>
21159 Paul Hua <paul.hua.gm@gmail.com>
21162 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
21164 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
21166 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
21167 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
21168 resp. SFmode cases.
21170 2018-03-15 Tamar Christina <tamar.christina@arm.com>
21173 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
21174 instead of GET_MODE_SIZE when comparing Units.
21176 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
21179 * varasm.c (hash_section): Return an unchangeble hash value
21180 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
21181 Return !aarch64_can_use_per_function_literal_pools_p ().
21183 2018-03-15 Jakub Jelinek <jakub@redhat.com>
21186 * optabs.c (emit_conditional_move): Pass address of cmode's copy
21187 rather than address of cmode as last argument to prepare_cmp_insn.
21189 2018-03-15 Julia Koval <julia.koval@intel.com>
21191 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
21192 F_AVX512VNNI, F_AVX512BITALG): New.
21194 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
21197 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
21198 insn for floating-point loads and stores.
21200 2018-03-14 Carl Love <cel@us.ibm.com>
21202 * config/rs6000/rs6000-c.c: Add macro definitions for
21203 ALTIVEC_BUILTIN_VEC_PERMXOR.
21204 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
21205 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
21206 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
21207 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
21209 * config/doc/extend.texi: Add prototypes for vec_permxor.
21211 2018-03-14 David Malcolm <dmalcolm@redhat.com>
21214 * diagnostic-show-locus.c (class layout_point): Convert m_line
21215 from int to linenum_type.
21216 (line_span::comparator): Use linenum "compare" function when
21217 comparing line numbers.
21218 (test_line_span): New function.
21219 (layout_range::contains_point): Convert param "row" from int to
21221 (layout_range::intersects_line_p): Likewise.
21222 (layout::will_show_line_p): Likewise.
21223 (layout::print_source_line): Likewise.
21224 (layout::should_print_annotation_line_p): Likewise.
21225 (layout::print_annotation_line): Likewise.
21226 (layout::print_leading_fixits): Likewise.
21227 (layout::annotation_line_showed_range_p): Likewise.
21228 (struct line_corrections): Likewise for field m_row.
21229 (line_corrections::line_corrections): Likewise for param "row".
21230 (layout::print_trailing_fixits): Likewise.
21231 (layout::get_state_at_point): Likewise.
21232 (layout::get_x_bound_for_row): Likewise.
21233 (layout::print_line): Likewise.
21234 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
21235 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
21236 * input.c (selftest::test_linenum_comparisons): New function.
21237 (selftest::input_c_tests): Call it.
21238 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
21239 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
21240 * selftest.h (ASSERT_GT): New macro.
21241 (ASSERT_GT_AT): New macro.
21242 (ASSERT_LT): New macro.
21243 (ASSERT_LT_AT): New macro.
21245 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
21247 PR rtl-optimization/84780
21248 * combine.c (distribute_links): Don't make a link based on pc_rtx.
21250 2018-03-14 Martin Liska <mliska@suse.cz>
21252 * tree.c (record_node_allocation_statistics): Use
21253 get_stats_node_kind.
21254 (get_stats_node_kind): New function extracted from
21255 record_node_allocation_statistics.
21256 (free_node): Use get_stats_node_kind.
21258 2018-03-14 Richard Biener <rguenther@suse.de>
21260 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
21261 that the value-set of ANTIC_IN doesn't grow.
21264 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
21266 (BB_VISITED_WITH_VISITED_SUCCS): New define.
21267 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
21269 2018-03-14 Julia Koval <julia.koval@intel.com>
21271 * config.gcc (icelake-client, icelake-server): New.
21273 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
21274 (initial_ix86_arch_features): Ditto.
21275 (PTA_SKYLAKE): Add SGX.
21276 (PTA_ICELAKE): Remove.
21277 (PTA_ICELAKE_CLIENT): New.
21278 (PTA_ICELAKE_SERVER): New.
21279 (ix86_option_override_internal): Split up icelake on icelake client and
21281 (get_builtin_code_for_version): Ditto.
21282 (fold_builtin_cpu): Ditto.
21283 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
21284 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
21285 * config/i386/i386.h (processor_type): Ditto.
21286 * doc/invoke.texi: Ditto.
21288 2018-03-14 Jakub Jelinek <jakub@redhat.com>
21291 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
21292 INTEGER_CST offset, add it together with bitpos / 8 and
21293 sign extend based on POINTER_SIZE.
21297 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21300 * config/i386/constraints.md (Yc): New register constraint.
21301 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21302 Use Yc constraint for alternative 2 of operand 0. Remove
21303 preferred_for_speed attribute.
21305 2018-03-14 Richard Biener <rguenther@suse.de>
21307 PR tree-optimization/84830
21308 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
21309 with the old one to avoid oscillations.
21311 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
21314 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
21316 (assign_by_spills): Return a flag of reload assignment failure.
21317 Do not process the reload assignment failures. Do not spill other
21318 reload pseudos if they has the same reg class. Update n if
21320 (lra_assign): Add a return arg. Set up from the result of
21321 assign_by_spills call.
21322 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21323 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21324 usage_insns if it is not NULL.
21325 (spill_hard_reg_in_range): New function.
21326 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21327 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21328 function prototypes.
21329 (lra_assign): Change prototype.
21330 * lra.c (lra): Add code to deal with fails by splitting hard reg
21333 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
21335 * config/riscv/riscv.opt (mrelax): New option.
21336 * config/riscv/riscv.c (riscv_file_start): Emit ".option
21337 "norelax" when riscv_mrelax is disabled.
21338 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
21340 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21343 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
21344 reassociation for int modes.
21346 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21348 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
21349 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
21351 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
21352 * config/aarch64/aarch64-sve.md
21353 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
21354 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
21355 (*extend<mode><Vwide>2): Rename to...
21356 (aarch64_sve_extend<mode><Vwide>2): ...this.
21357 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
21358 renaming the old pattern to...
21359 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
21361 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
21362 define_expand, renaming the old pattern to...
21363 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
21364 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
21365 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
21366 account when deciding which SVE instruction the optab should use.
21367 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
21369 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21371 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
21372 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
21373 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
21374 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
21375 (tlsdesc_small_<mode>): Turn a define_expand and use
21376 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
21377 (tlsdesc_small_advsimd_<mode>): ...this.
21378 (tlsdesc_small_sve_<mode>): New pattern.
21380 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
21382 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
21383 (UNSPEC_UMUL_HIGHPART): New constants.
21384 (MUL_HIGHPART): New int iteraor.
21385 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
21386 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
21388 (*<su>mul<mode>3_highpart): New define_insn.
21390 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
21393 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
21396 2018-03-13 Martin Liska <mliska@suse.cz>
21399 * (sem_item_optimizer::sem_item_optimizer): Initialize new
21401 (sem_item_optimizer::~sem_item_optimizer): Release it.
21402 (sem_item_optimizer::merge_classes): Register variable aliases.
21403 (sem_item_optimizer::fixup_pt_set): New function.
21404 (sem_item_optimizer::fixup_points_to_sets): Likewise.
21405 * ipa-icf.h: Declare new variables and functions.
21407 2018-03-13 Jakub Jelinek <jakub@redhat.com>
21409 PR middle-end/84834
21410 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
21411 integer_pow2p@2 and test integer_pow2p in condition.
21412 (A < 0 ? C : 0): Similarly for @1.
21414 PR middle-end/84831
21415 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
21416 characters starting at p contain '\0' character, don't look beyond
21420 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
21421 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
21424 * reg-stack.c (change_stack): Change update_end var from int to
21425 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
21426 also call set_block_for_insn on the newly added insns and rescan.
21429 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
21430 on the last operand.
21433 * tree.c (stabilize_reference_1): Return save_expr (e) for
21434 STATEMENT_LIST even if it doesn't have side-effects.
21436 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
21438 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
21440 2018-03-12 Renlin Li <renlin.li@arm.com>
21442 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
21443 aarch64_output_scalar_simd_mov_immediate.
21445 2018-03-12 Martin Sebor <msebor@redhat.com>
21447 PR tree-optimization/83456
21448 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
21449 for perfectly overlapping calls to memcpy.
21450 (gimple_fold_builtin_memory_chk): Same.
21451 (gimple_fold_builtin_strcpy): Handle no-warning.
21452 (gimple_fold_builtin_stxcpy_chk): Same.
21453 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
21455 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21457 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
21458 parameter. Use it for SFmode.
21459 (rs6000_function_arg_advance_1): Adjust.
21460 (rs6000_function_arg): Adjust.
21461 (rs6000_gimplify_va_arg): Pass false for that new parameter.
21463 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
21465 PR rtl-optimization/84169
21466 PR rtl-optimization/84780
21467 * combine.c (can_combine_p): Check for a 2-insn combination whether
21468 the destination register is used between the two insns, too.
21470 2018-03-12 Richard Biener <rguenther@suse.de>
21472 PR tree-optimization/84803
21473 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
21474 for refs DR analysis didn't process.
21476 2018-03-12 Richard Biener <rguenther@suse.de>
21478 PR tree-optimization/84777
21479 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
21480 force-vectorize loops ignore whether we are optimizing for size.
21482 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
21484 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
21485 (TARGET_MD_ASM_ADJUST): Define.
21487 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
21488 Kito Cheng <kito.cheng@gmail.com>
21489 Chung-Ju Wu <jasonwucj@gmail.com>
21491 * config/nds32/nds32.c (nds32_compute_stack_frame,
21492 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
21493 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
21494 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
21495 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
21496 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
21497 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
21498 * config/nds32/nds32.md (prologue, epilogue): Use macro
21499 NDS32_V3PUSH_AVAILABLE_P to do checking.
21501 2018-03-11 Jakub Jelinek <jakub@redhat.com>
21504 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
21505 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
21506 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
21507 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
21508 addition of most attributes on !orig_type_die or the attribute not
21509 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
21511 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21512 Chung-Ju Wu <jasonwucj@gmail.com>
21514 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
21515 __NDS32_VH__ macro.
21516 * config/nds32/nds32.opt (mvh): New option.
21518 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21519 Chung-Ju Wu <jasonwucj@gmail.com>
21521 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
21523 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
21524 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
21527 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
21528 Chung-Ju Wu <jasonwucj@gmail.com>
21530 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
21532 * config/nds32/nds32-multiple.md (strlensi): New pattern.
21533 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
21535 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
21536 Kito Cheng <kito.cheng@gmail.com>
21537 Chung-Ju Wu <jasonwucj@gmail.com>
21539 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
21540 UNSPEC_FFMISM and UNSPEC_FLMISM.
21541 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
21542 for ffb, ffmism and flmism.
21543 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
21544 (unspec_ffmism): Ditto.
21545 (unspec_flmism): Ditto.
21546 (nds32_expand_builtin_impl): Check if string extension is available.
21547 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
21548 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
21550 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
21553 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21556 * lra-assigns.c (assign_by_spills): Return a flag of reload
21557 assignment failure. Do not process the reload assignment
21558 failures. Do not spill other reload pseudos if they has the same
21560 (lra_assign): Add a return arg. Set up from the result of
21561 assign_by_spills call.
21562 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21563 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21564 usage_insns if it is not NULL.
21565 (spill_hard_reg_in_range): New function.
21566 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21567 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21568 function prototypes.
21569 (lra_assign): Change prototype.
21570 * lra.c (lra): Add code to deal with fails by splitting hard reg
21573 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
21576 * config/i386/i386.opt: Replace Enforcment with Enforcement.
21578 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
21581 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
21582 (dw_val_node): Add val_symbolic_view.
21583 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
21584 (symview_upper_bound): New.
21585 (new_line_info_table): Initialize symviews_since_reset.
21586 (dwarf2out_source_line): Count symviews_since_reset and set
21587 symview_upper_bound.
21588 (dw_val_equal_p): Handle symview.
21589 (add_AT_symview): New.
21590 (print_dw_val): Handle symview.
21591 (attr_checksum, attr_checksum_ordered): Likewise.
21592 (same_dw_val_p, size_of_die): Likewise.
21593 (value_format, output_die): Likewise.
21594 (add_high_low_attributes): Use add_AT_symview for entry_view.
21595 (dwarf2out_finish): Reset symview_upper_bound, clear
21598 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
21601 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
21602 Add strict argument and use it.
21603 (rs6000_split_multireg_move): Update for new strict argument.
21604 (mem_operand_gpr): Disallow all non-offsettable addresses.
21605 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
21607 2018-03-09 Jakub Jelinek <jakub@redhat.com>
21610 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
21611 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
21612 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
21615 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
21616 decl, use remap_type if we want to use the type.
21618 2018-03-09 Martin Sebor <msebor@redhat.com>
21620 PR tree-optimization/84526
21621 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
21623 (builtin_access::generic_overlap): Be prepared to handle non-array
21626 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
21628 PR rtl-optimization/84682
21629 * lra-constraints.c (process_address_1): Check is_address flag
21630 for address constraints.
21631 (process_alt_operands): Likewise.
21632 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
21633 preprocess_constraints.
21634 * recog.h (preprocess_constraints): Add oploc parameter.
21636 * recog.c (preprocess_constraints): Test address_operand for
21637 CT_ADDRESS constraints.
21639 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
21642 * lra-assigns.c (assign_by_spills): Return a flag of reload
21643 assignment failure. Do not process the reload assignment
21644 failures. Do not spill other reload pseudos if they has the same
21646 (lra_assign): Add a return arg. Set up from the result of
21647 assign_by_spills call.
21648 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
21649 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
21650 usage_insns if it is not NULL.
21651 (spill_hard_reg_in_range): New function.
21652 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
21653 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
21654 function prototypes.
21655 (lra_assign): Change prototype.
21656 * lra.c (lra): Add code to deal with fails by splitting hard reg
21659 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21662 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
21663 Accept complain bool parameter. Only emit errors if it is true.
21664 (arm_parse_cpu_option_name): Likewise.
21665 (arm_target_thumb_only): Adjust callers of the above.
21666 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
21667 prototype to take a default true bool parameter.
21668 (arm_parse_arch_option_name): Likewise.
21670 2018-03-09 David Malcolm <dmalcolm@redhat.com>
21671 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
21675 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
21676 * configure: Regenerate.
21677 * configure.ac ("linker --version-script option"): New.
21678 ("linker soname option"): New.
21680 2018-03-09 Richard Biener <rguenther@suse.de>
21682 PR tree-optimization/84775
21683 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
21684 immediate uses of predicate stmts and mark them modified.
21687 PR tree-optimization/84178
21688 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21690 (version_loop_for_if_conversion): Delay update_ssa call.
21691 (tree_if_conversion): Delay update_ssa until after predicate
21694 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
21697 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
21698 when the function accesses prior frames.
21700 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21703 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
21704 gen_llsym, otherwise call maybe_gen_llsym.
21706 PR inline-asm/84742
21707 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
21708 has ',' character inside of it.
21710 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21713 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
21714 as clobbering CC_REGNUM.
21716 2018-03-08 Richard Biener <rguenther@suse.de>
21718 PR middle-end/84552
21719 * tree-scalar-evolution.c: Include tree-into-ssa.h.
21720 (follow_copies_to_constant): Do not follow SSA names registered
21723 2018-03-08 Richard Biener <rguenther@suse.de>
21725 PR tree-optimization/84178
21726 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
21728 (version_loop_for_if_conversion): Delay update_ssa call.
21729 (tree_if_conversion): Delay update_ssa until after predicate
21732 2018-03-08 David Malcolm <dmalcolm@redhat.com>
21734 PR tree-optimization/84178
21735 * tree-if-conv.c (release_bb_predicate): Remove the
21736 the assertion that the stmts have NULL use_ops.
21737 Discard the statements, asserting that they haven't
21738 yet been added to a BB.
21740 2018-03-08 Richard Biener <rguenther@suse.de>
21742 PR tree-optimization/84746
21743 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
21744 (phi_translate): Pass in destination ANTIC_OUT set.
21745 (phi_translate_1): Likewise. For a simplified result lookup
21746 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
21747 (phi_translate_set): Adjust.
21748 (do_pre_regular_insertion): Likewise.
21749 (do_pre_partial_partial_insertion): Likewise.
21751 2018-03-08 Martin Liska <mliska@suse.cz>
21753 PR gcov-profile/84735
21754 * doc/gcov.texi: Document usage of profile files.
21755 * gcov-io.h: Document changes in the format.
21757 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
21761 * dwarf2out.c (struct dw_line_info_table): Update comments for
21763 (FORCE_RESET_NEXT_VIEW): New.
21764 (FORCE_RESETTING_VIEW_P): New.
21765 (RESETTING_VIEW_P): Check for -1 too.
21766 (ZERO_VIEW_P): Likewise.
21767 (new_line_info_table): Force-reset next view.
21768 (dwarf2out_begin_function): Likewise.
21769 (dwarf2out_source_line): Simplify zero_view_p initialization.
21770 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
21771 view directly. Omit view when omitting .loc at line 0.
21773 2018-03-08 Jakub Jelinek <jakub@redhat.com>
21775 PR tree-optimization/84740
21776 * tree-switch-conversion.c (process_switch): Call build_constructors
21777 only if info.phi_count is non-zero.
21779 PR tree-optimization/84739
21780 * tree-tailcall.c (find_tail_calls): Check call arguments against
21781 DECL_ARGUMENTS (current_function_decl) rather than
21782 DECL_ARGUMENTS (func) when checking for tail recursion.
21784 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21786 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
21787 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
21788 Volker Reichelt's entry and add entries for people that perform
21789 GCC fuzzy testing and report numerous bugs.
21791 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
21794 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
21795 readonly data in sdata, if that is disabled.
21796 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
21797 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
21798 -mreadonly-in-sdata option.
21800 2018-03-07 Martin Sebor <msebor@redhat.com>
21802 PR tree-optimization/84468
21803 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
21804 basic block when looking for nul assignment.
21806 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
21809 * except.h (output_function_exception_table): Adjust prototype.
21810 * except.c (output_function_exception_table): Remove FNNAME parameter
21811 and add SECTION parameter. Ouput one part of the table at a time.
21812 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
21813 the first part of the exception table and emit unwind directives.
21814 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
21815 (i386_pe_seh_cold_init): Likewise.
21816 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
21817 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
21818 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
21819 (ix86_output_call_insn): Emit a nop in one more case for SEH.
21820 * config/i386/winnt.c: Include except.h.
21821 (struct seh_frame_state): Add reg_offset, after_prologue and
21822 in_cold_section fields.
21823 (i386_pe_seh_end_prologue): Set seh->after_prologue.
21824 (i386_pe_seh_cold_init): New function.
21825 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
21826 to seh->in_cold_section.
21827 (seh_emit_push): Record the offset of the push.
21828 (seh_emit_save): Record the offet of the save.
21829 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
21830 Test seh->after_prologue to disregard the epilogue.
21831 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
21832 (i386_pe_end_cold_function): New function.
21834 2018-03-07 Jakub Jelinek <jakub@redhat.com>
21837 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
21838 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
21841 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
21843 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
21844 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
21846 PR middle-end/84723
21847 * multiple_target.c: Include tree-inline.h and intl.h.
21848 (expand_target_clones): Diagnose and fail if node->definition and
21849 !tree_versionable_function_p (node->decl).
21851 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
21853 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
21855 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
21856 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
21857 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
21859 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21862 * combine.c (try_combine): Use reg_or_subregno instead of handling
21863 just paradoxical SUBREGs and REGs.
21865 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
21867 * config/arc/arc.c (arc_finalize_pic): Remove function.
21868 (arc_must_save_register): We use single base PIC register, remove
21869 checks to save/restore the PIC register.
21870 (arc_expand_prologue): Likewise.
21871 * config/arc/arc-protos.h (arc_set_default_type_attributes):
21873 (arc_verify_short): Likewise.
21874 (arc_attr_type): Likewise.
21875 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
21876 (walk_stores): Likewise.
21877 (arc_address_cost): Make it static.
21878 (arc_verify_short): Likewise.
21879 (branch_dest): Likewise.
21880 (arc_attr_type): Likewise.
21881 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
21882 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
21883 (arc_final_prescan_insn): Remove inserting the nops due to
21884 hardware hazards. It is done in reorg step.
21885 (insn_length_variant_t): Remove.
21886 (insn_length_parameters_t): Likewise.
21887 (arc_insn_length_parameters): Likewise.
21888 (arc_get_insn_variants): Likewise.
21889 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
21891 2018-03-06 Jakub Jelinek <jakub@redhat.com>
21893 PR inline-asm/84683
21894 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
21897 PR tree-optimization/84687
21898 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
21900 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
21902 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21904 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
21905 Rename to ppc_speculation_barrier.
21906 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
21907 __builtin_ppc_speculation_barrier.
21909 2018-03-05 Jakub Jelinek <jakub@redhat.com>
21912 * combine.c (combine_simplify_rtx): Don't try to simplify if
21913 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
21916 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
21918 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
21919 to 32 bytes when compiling for POWER9.
21921 2018-03-05 Jakub Jelinek <jakub@redhat.com>
21924 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
21925 regparm >= 3 with no arg reg available also for calls with
21926 flag_force_indirect_call. Pass decl to ix86_function_regparm.
21929 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
21931 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
21933 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
21936 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
21938 2018-03-05 Richard Biener <rguenther@suse.de>
21940 PR tree-optimization/84486
21941 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
21942 When inserting a __builtin_assume_aligned call set the LHS
21943 SSA name alignment info accordingly.
21945 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
21947 PR tree-optimization/84114
21948 * config/aarch64/aarch64.c (aarch64_reassociation_width)
21949 Avoid reassociation of FLOAT_MODE addition.
21951 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
21953 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
21954 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
21955 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
21956 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
21957 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
21958 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
21959 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
21961 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
21962 __builtin_ia32_wbinvd): New builtins.
21963 (SPECIAL_ARGS2): New.
21964 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
21965 (SPECIAL_ARGS2): New.
21966 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
21967 (ix86_valid_target_attribute_inner_p): Ditto.
21968 (ix86_init_mmx_sse_builtins): Add special_args2.
21969 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
21970 TARGET_WBNOINVD_P): New.
21971 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
21972 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
21973 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
21974 * config/i386/immintrin.h (_wbinvd): New intrinsic.
21975 * config/i386/pconfigintrin.h: New file.
21976 * config/i386/wbnoinvdintrin.h: Ditto.
21977 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
21979 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
21981 2018-03-05 Richard Biener <rguenther@suse.de>
21983 PR tree-optimization/84670
21984 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
21986 (BB_VISITED_WITH_VISITED_SUCCS): New define.
21987 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
21988 (compute_antic_aux): Only assert the number of values in ANTIC_IN
21989 doesn't grow if all successors (recursively) were visited at least
21992 2018-03-05 Richard Biener <rguenther@suse.de>
21994 PR tree-optimization/84650
21995 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
21996 if executed in the loop pipeline.
21998 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
22000 * doc/configfiles.texi (Configuration Files): Move info about
22001 conditionalizing $target-protos.h to...
22002 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
22003 differs from $target-protos.h.
22005 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
22006 Chung-Ju Wu <jasonwucj@gmail.com>
22008 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
22009 * config/nds32/nds32-multiple.md (setmemsi): Define.
22010 * config/nds32/nds32-memory-manipulation.c
22011 (nds32_gen_dup_4_byte_to_word_value): New.
22012 (emit_setmem_word_loop): New.
22013 (emit_setmem_byte_loop): New.
22014 (nds32_expand_setmem_loop): New.
22015 (nds32_expand_setmem_loop_v3m): New.
22016 (nds32_expand_setmem_unroll): New.
22017 (nds32_expand_setmem): New.
22019 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
22020 Chung-Ju Wu <jasonwucj@gmail.com>
22022 * config/nds32/nds32-memory-manipulation.c
22023 (nds32_emit_load_store): New.
22024 (nds32_emit_post_inc_load_store): New.
22025 (nds32_emit_mem_move): New.
22026 (nds32_emit_mem_move_block): New.
22027 (nds32_expand_movmemsi_loop_unknown_size): New.
22028 (nds32_expand_movmemsi_loop_known_size): New.
22029 (nds32_expand_movmemsi_loop): New.
22030 (nds32_expand_movmemsi_unroll): New.
22031 (nds32_expand_movmemqi): Rename ...
22032 (nds32_expand_movmemsi): ... to this.
22033 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
22034 (movmemsi): ... to this.
22035 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
22036 (nds32_expand_movmemsi): ... to this.
22038 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
22039 Monk Chiang <sh.chiang04@gmail.com>
22040 Chung-Ju Wu <jasonwucj@gmail.com>
22042 * config/nds32/nds32-protos.h
22043 (nds32_expand_load_multiple): New arguments.
22044 (nds32_expand_store_multiple): Ditto.
22045 (nds32_valid_multiple_load_store): Rename ...
22046 (nds32_valid_multiple_load_store_p): ... to this.
22047 * config/nds32/nds32-memory-manipulation.c
22048 (nds32_expand_load_multiple): Refine implementation.
22049 (nds32_expand_store_multiple): Ditto.
22050 * config/nds32/nds32-multiple.md
22051 (load_multiple): Update nds32_expand_load_multiple interface.
22052 (store_multiple): Update nds32_expand_store_multiple interface.
22053 * config/nds32/nds32-predicates.c
22054 (nds32_valid_multiple_load_store): Rename ...
22055 (nds32_valid_multiple_load_store_p): ... to this and refine
22057 * config/nds32/predicates.md
22058 (nds32_load_multiple_and_update_address_operation): New predicate.
22059 (nds32_store_multiple_and_update_address_operation): New predicate.
22061 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
22062 Chung-Ju Wu <jasonwucj@gmail.com>
22064 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
22065 (combo): New attribute.
22066 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
22068 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
22070 * config/nds32/nds32.opt: Change -mcmodel= default value.
22072 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
22073 Monk Chiang <sh.chiang04@gmail.com>
22074 Chung-Ju Wu <jasonwucj@gmail.com>
22076 * config/nds32/constants.md (unspec_element): New enum.
22077 * config/nds32/constraints.md (Umw): New constraint.
22078 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
22079 * config/nds32/nds32-intrinsic.md: Likewise.
22080 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
22081 (nds32_valid_smw_lwm_base_p): New.
22082 (nds32_output_smw_single_word): New.
22083 (nds32_output_lmw_single_word): New.
22084 (nds32_expand_unaligned_load): New.
22085 (nds32_expand_unaligned_store): New.
22086 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
22087 (nds32_output_smw_single_word): Declare.
22088 (nds32_output_lmw_single_word): Declare.
22089 (nds32_expand_unaligned_load): Declare.
22090 (nds32_expand_unaligned_store): Declare.
22091 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
22092 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
22093 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
22094 NDS32_BUILTIN_UASTORE_DW.
22095 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
22098 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
22099 Kito Cheng <kito.cheng@gmail.com>
22100 Chung-Ju Wu <jasonwucj@gmail.com>
22102 * config/nds32/nds32-intrinsic.c
22103 (nds32_expand_builtin_null_ftype_reg): Delete.
22104 (nds32_expand_builtin_reg_ftype_imm): Ditto.
22105 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
22106 (nds32_read_argument): New.
22107 (nds32_legitimize_target): Ditto.
22108 (nds32_legitimize_argument): Ditto.
22109 (nds32_check_constant_argument): Ditto.
22110 (nds32_expand_unop_builtin): Ditto.
22111 (nds32_expand_unopimm_builtin): Ditto.
22112 (nds32_expand_binop_builtin): Ditto.
22113 (nds32_builtin_decl_impl): Ditto.
22114 (builtin_description): Ditto.
22115 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
22116 (nds32_init_builtins_impl): Ditto.
22117 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
22118 (nds32_builtin_decl): New.
22119 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
22120 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
22122 2018-03-02 Jeff Law <law@redhat.com>
22124 * reorg.c (stop_search_p): Handle DEBUG_INSN.
22125 (redundant_insn, fill_simple_delay_slots): Likewise.
22126 (fill_slots_from_thread): Likewise.
22127 * resource.c (mark_referenced_resources): Likewise.
22128 (mark_set_resources, find_dead_or_set_registers): Likewise.
22130 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22132 * substring-locations.h (format_warning_va): Formatting fix for
22133 ATTRIBUTE_GCC_DIAG.
22134 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
22136 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
22137 * substring-locations.c: Include intl.h.
22138 (format_warning_va): Turned into small wrapper around
22139 format_warning_n_va, renamed to ...
22140 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
22141 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
22143 (format_warning_at_substring_n): New function.
22144 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
22145 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
22146 format_warning_at_substring with just a shorter name instead of
22147 const function pointer.
22148 (fmtwarn_n): New function.
22149 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
22150 appropriate, get rid of all the fmtstr temporaries, move conditionals
22151 with G_() wrapped string literals directly into fmtwarn arguments,
22152 cast dir.len to (int), formatting fixes.
22154 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
22156 * doc/invoke.texi: Remove "Cilk Plus" references.
22158 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22159 Richard Biener <rguenther@suse.de>
22162 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
22163 for error or warning attributes if CALL_FROM_THUNK_P is set.
22166 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22169 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
22170 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
22173 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
22174 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
22176 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
22177 instead of -1U in last predictors element's probability member.
22179 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
22182 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
22183 arguments if they are comparable.
22185 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
22187 PR tree-optimization/84634
22188 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
22189 masks and masked_loop_p with a single loop_masks, making sure it's
22190 null for bb vectorization.
22192 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
22194 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
22195 (vect_analyze_data_ref_access): Use loop->safe_len rather than
22196 loop->force_vectorize to check whether there is no alias.
22198 2018-03-02 Jakub Jelinek <jakub@redhat.com>
22201 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
22203 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
22205 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
22206 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
22207 instead of a loop around prev_real_insn.
22208 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
22211 PR inline-asm/84625
22212 * config/i386/i386.c (ix86_print_operand): Use conditional
22213 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
22216 2018-03-02 Richard Biener <rguenther@suse.de>
22218 PR tree-optimization/84427
22219 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
22220 (bitmap_set_subtract_values): Rewrite to handle multiple
22223 (prune_clobbered_mems): Likewise.
22224 (phi_translate): Take edge instead of pred/phiblock.
22225 (phi_translate_1): Likewise.
22226 (phi_translate_set): Likewise. Insert all translated
22227 exprs for a value into the set, keeping possibly multiple
22228 expressions per value.
22229 (compute_antic_aux): Adjust for phi_translate changes.
22230 When intersecting union the expressions and prune those
22231 not in the final value set, keeping possibly multiple
22232 expressions per value. Do not use value-insertion
22233 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
22234 all expressions. Add verification that the value-sets
22235 only shrink during iteration.
22236 (compute_partial_antic_aux): Adjust for the phi_translate changes.
22237 (do_pre_regular_insertion): Likewise.
22238 (do_pre_partial_partial_insertion): Likewise.
22240 2018-03-02 Richard Biener <rguenther@suse.de>
22243 * config/darwin.c (saved_debug_info_level): New static global.
22244 (darwin_asm_lto_start): Disable debug info generation for LTO out.
22245 (darwin_asm_lto_end): Restore debug info generation settings.
22247 2018-03-01 Martin Liska <mliska@suse.cz>
22250 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
22251 volatile arguments.
22253 2018-03-01 Richard Biener <rguenther@suse.de>
22256 * dwarf2out.c (gen_variable_die): Properly handle late VLA
22257 type annotation with LTO when debug was disabled at compile-time.
22259 2018-03-01 Matthew Fortune <mfortune@gmail.com>
22261 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
22263 (mips_final_postscan_insn): Likewise.
22265 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
22267 PR rtl-optimization/84528
22268 * alias.c (init_alias_target): Add commentary.
22269 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
22270 a unique base value if the frame pointer is not eliminated
22271 to the stack pointer.
22273 2018-03-01 Tom de Vries <tom@codesourcery.com>
22275 PR rtl-optimization/83327
22276 * lra-int.h (hard_regs_spilled_into): Declare.
22277 * lra.c (hard_regs_spilled_into): Define.
22278 (init_reg_info): Init hard_regs_spilled_into.
22279 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
22280 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
22281 (process_bb_lives): Handle hard_regs_spilled_into.
22282 (lra_create_live_ranges_1): Before doing liveness propagation, clear
22283 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
22285 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
22287 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
22288 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
22289 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
22290 * config/rs6000/aix72.h: New file.
22292 2018-02-28 Jakub Jelinek <jakub@redhat.com>
22294 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
22295 instead of warning_at with conditional singular and plural messages
22299 * stor-layout.c (update_alignment_for_field): For
22300 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
22301 && !DECL_PACKED (field), do the alignment update, just use
22302 only desired_align instead of MAX (type_align, desired_align)
22304 (place_field): Don't do known_align < desired_align handling
22305 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
22306 is non-NULL, instead do it after rli->prev_field handling and
22307 only if not within a bitfield word. For DECL_PACKED (field)
22308 use type_align of BITS_PER_UNIT.
22310 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
22312 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
22313 superfluous parentheses and trailing spaces.
22315 2018-02-28 Richard Biener <rguenther@suse.de>
22317 PR tree-optimization/84584
22318 * graphite-scop-detection.c (scop_detection::add_scop): Discard
22319 SCoPs with fake exit edge.
22321 2018-02-28 Martin Liska <mliska@suse.cz>
22324 * timevar.c (timer::print): Fix format to properly print 100%
22327 2018-02-28 Richard Biener <rguenther@suse.de>
22329 PR middle-end/84607
22330 * genmatch.c (capture_info::walk_match): Do not mark
22331 captured expressions without operands as expr_p given
22332 they act more like predicates and should be subject to
22333 "lost tail" side-effect preserving.
22335 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
22337 PR rtl-optimization/81611
22338 * auto-inc-dec.c (attempt_change): Move dead note from
22339 mem_insn if it's the next use of regno
22340 (find_address): Take address use of reg holding
22341 non-incremented value. Add parm to limit search to the named
22343 (merge_in_block): Attempt to use a mem insn that is the next
22344 use of the original regno.
22346 2018-02-27 Martin Sebor <msebor@redhat.com>
22349 * doc/invoke.texi (-Wmissing-attributes): New option.
22350 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
22352 2018-02-27 Martin Sebor <msebor@redhat.com>
22354 PR translation/84207
22355 * diagnostic-core.h (warning_n, error_n, inform_n): Change
22356 n argument to unsigned HOST_WIDE_INT.
22357 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
22358 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
22359 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
22360 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
22362 2018-02-27 Richard Biener <rguenther@suse.de>
22364 PR tree-optimization/84512
22365 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
22366 Do not use the estimate returned from record_stmt_cost for
22367 the scalar iteration cost but sum properly using add_stmt_cost.
22369 2018-02-27 Richard Biener <rguenther@suse.de>
22371 PR tree-optimization/84466
22372 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
22373 Adjust last change to less strictly validate use operands.
22375 2018-02-27 Martin Liska <mliska@suse.cz>
22377 PR gcov-profile/84548
22378 * gcov.c (process_file): Allow partial overlap and consider it
22379 also as group functions.
22380 (output_lines): Properly calculate range of lines for a group.
22382 2018-02-27 Martin Liska <mliska@suse.cz>
22384 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
22385 'ggc' suffixes. Change first column width.
22386 (timer::print): Fix formatting of the column.
22388 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
22390 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
22391 preserve inline entry blocks for the sake of debug inline
22392 entry point markers alone.
22393 (remove_unused_locals): Suggest in comments a better place to
22394 force the preservation of inline entry blocks that are
22395 otherwise unused, but do not preserve them.
22397 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22399 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
22401 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22404 * config/i386/constraints.md (Bs): Replace
22405 ix86_indirect_branch_register with
22406 TARGET_INDIRECT_BRANCH_REGISTER.
22408 * config/i386/i386.md (indirect_jump): Likewise.
22409 (tablejump): Likewise.
22410 (*sibcall_memory): Likewise.
22411 (*sibcall_value_memory): Likewise.
22412 Peepholes of indirect call and jump via memory: Likewise.
22413 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
22414 (*sibcall_value_GOT_32): Likewise.
22415 * config/i386/predicates.md (indirect_branch_operand): Likewise.
22416 (GOT_memory_operand): Likewise.
22417 (call_insn_operand): Likewise.
22418 (sibcall_insn_operand): Likewise.
22419 (GOT32_symbol_operand): Likewise.
22420 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
22422 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22424 PR rtl-optimization/83496
22425 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
22426 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
22427 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
22428 redundant insn, if any.
22429 (relax_delay_slots): Likewise.
22430 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
22432 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
22434 PR tree-optimization/83965
22435 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
22436 that grouped statements are part of a reduction chain. Return
22437 true if the statement is not marked as a reduction itself but
22438 is part of a group.
22439 (vect_recog_dot_prod_pattern): Don't check whether the statement
22440 is part of a group here.
22441 (vect_recog_sad_pattern): Likewise.
22442 (vect_recog_widen_sum_pattern): Likewise.
22444 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
22447 * final.c (rest_of_clean_state): Also look for calls inside sequences.
22449 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
22452 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
22454 (ix86_output_indirect_function_return): New prototype.
22455 (ix86_split_simple_return_pop_internal): Likewise.
22456 * config/i386/i386.c (indirect_return_via_cx): New.
22457 (indirect_return_via_cx_bnd): Likewise.
22458 (indirect_thunk_name): Handle return va CX_REG.
22459 (output_indirect_thunk_function): Create alias for
22460 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
22461 (ix86_output_indirect_jmp): Remove the bool argument.
22462 (ix86_output_indirect_function_return): New function.
22463 (ix86_split_simple_return_pop_internal): Likewise.
22464 * config/i386/i386.md (*indirect_jump): Don't pass false
22465 to ix86_output_indirect_jmp.
22466 (*tablejump_1): Likewise.
22467 (simple_return_pop_internal): Change it to define_insn_and_split.
22468 Call ix86_split_simple_return_pop_internal to split it for
22469 -mfunction-return=.
22470 (simple_return_indirect_internal): Call
22471 ix86_output_indirect_function_return instead of
22472 ix86_output_indirect_jmp.
22474 2018-02-26 Jakub Jelinek <jakub@redhat.com>
22477 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
22478 memset and value initialization afterwards.
22480 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
22482 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
22484 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22487 * common/config/aarch64/aarch64-common.c
22488 (aarch_option_optimization_table[]): Switch
22489 off fomit-frame-pointer
22491 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22492 Chung-Ju Wu <jasonwucj@gmail.com>
22494 * config/nds32/nds32-multiple.md (load_multiple): Disallow
22496 (store_multiple): Ditto.
22498 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
22500 * config.gcc: Add --with-cpu support for nds32 target.
22501 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
22502 * config/nds32/nds32.opt: Add -mcpu= option.
22504 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
22506 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
22507 isel=yes): Warn for these deprecated options.
22509 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
22511 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
22512 ISA_2_5_MASKS_EMBEDDED.
22514 2018-02-23 Jakub Jelinek <jakub@redhat.com>
22516 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
22517 p->max as pointers rather than using iterative_hash_expr.
22519 2018-02-23 Carl Love <cel@us.ibm.com>
22521 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
22522 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
22524 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
22525 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
22526 P8V_BUILTIN_VEC_VUNSIGNED2.
22528 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
22531 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
22532 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
22534 * lra-constraints.c (curr_insn_transform): Set up
22535 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
22537 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
22538 * lra-eliminations.c (spill_pseudos): Ditto.
22539 (process_insn_for_elimination): Ditto.
22540 * lra-lives.c (reg_early_clobber_p): Use the new macros.
22541 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
22542 LRA_NON_CLOBBERED_ALT.
22544 2018-02-22 Martin Sebor <msebor@redhat.com>
22546 PR tree-optimization/84480
22547 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
22548 to maybe_diag_stxncpy_trunc. Call it.
22549 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
22550 from gimple_fold_builtin_strcpy. Print inlining stack.
22551 (handle_builtin_stxncpy): Print inlining stack.
22552 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
22554 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
22557 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
22558 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
22559 and -fcheck-pointer-bounds are used together.
22560 (indirect_thunk_prefix): New enum.
22561 (indirect_thunk_need_prefix): New function.
22562 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
22563 "_nt" instead of "_bnd" for NOTRACK prefix.
22564 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
22565 (output_indirect_thunk_function): Likewise.
22567 (ix86_code_end): Update output_indirect_thunk_function calls.
22568 (ix86_output_indirect_branch_via_reg): Replace
22569 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
22570 (ix86_output_indirect_branch_via_push): Likewise.
22571 (ix86_output_function_return): Likewise.
22572 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
22573 incompatible with -fcf-protection=branch and
22574 -fcheck-pointer-bounds.
22576 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22579 * config/aarch64/aarch64.c (aarch64_print_address_internal):
22580 Change gcc_assert call to output_operand_lossage.
22582 2018-02-22 Steve Ellcey <sellcey@cavium.com>
22584 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
22586 2018-02-22 DJ Delorie <dj@redhat.com>
22587 Sebastian Perta <sebastian.perta@renesas.com>
22588 Oleg Endo <olegendo@gcc.gnu.org>
22590 * config/rx/rx.c (rx_rtx_costs): New function.
22591 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
22593 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
22595 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
22597 2018-02-22 Martin Liska <mliska@suse.cz>
22600 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
22601 Add "native" as a possible value.
22603 2018-02-22 Martin Liska <mliska@suse.cz>
22606 * config/i386/i386.c (ix86_option_override_internal):
22607 Add "native" as a possible value for -march and -mtune.
22609 2018-02-22 Jakub Jelinek <jakub@redhat.com>
22612 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
22613 to all type variants.
22615 PR tree-optimization/84503
22616 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
22617 width as info->bitpos + info->bitsize - start.
22618 (merged_store_group::merge_overlapping): Simplify width computation.
22619 (check_no_overlap): New function.
22620 (imm_store_chain_info::try_coalesce_bswap): Compute expected
22621 start + width and last_order of the group, fail if check_no_overlap
22623 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
22624 to group if check_no_overlap fails.
22626 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22628 * config/rs6000/altivec.md: Delete contraint arguments to
22629 define_expand, define_split, and define_peephole2, and in
22630 define_insn_and_split if always unused.
22631 * config/rs6000/darwin.md: Ditto.
22632 * config/rs6000/dfp.md: Ditto.
22633 * config/rs6000/rs6000.md: Ditto.
22634 * config/rs6000/sync.md: Ditto.
22635 * config/rs6000/vector.md: Ditto.
22636 * config/rs6000/vsx.md: Ditto.
22638 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
22640 * config/rs6000/altivec.md: Write output control strings as braced
22641 blocks instead of double-quoted strings.
22642 * config/rs6000/darwin.md: Ditto.
22643 * config/rs6000/rs6000.md: Ditto.
22644 * config/rs6000/vector.md: Ditto.
22645 * config/rs6000/vsx.md: Ditto.
22647 2018-02-21 Jason Merrill <jason@redhat.com>
22649 PR c++/84314 - ICE with templates and fastcall attribute.
22650 * attribs.c (build_type_attribute_qual_variant): Remove assert.
22652 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22654 * ipa-cp.c (determine_versionability): Fix comment typos.
22656 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
22659 * ipa-cp.c (determine_versionability): Do not version functions caling
22662 2018-02-21 Martin Liska <mliska@suse.cz>
22665 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
22666 Add "native" as a possible value.
22667 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
22668 the macro when native cpu detection is available.
22670 2018-02-21 Martin Liska <mliska@suse.cz>
22673 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
22674 Add "native" as a possible value.
22675 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
22676 when native cpu detection is available.
22678 2018-02-21 Jakub Jelinek <jakub@redhat.com>
22679 Martin Sebor <msebor@redhat.com>
22681 PR tree-optimization/84478
22682 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
22684 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
22685 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
22686 support which is conservatively correct, for 2 only stay conservative
22687 for maxlen. Formatting and comment capitalization fixes. Add STRICT
22688 argument to the 2 argument get_range_strlen, adjust 6 arg
22689 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
22691 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
22692 (gimple_fold_builtin_strlen): Pass true as last argument to
22695 2018-02-20 Martin Sebor <msebor@redhat.com>
22697 PR middle-end/84095
22698 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
22699 (builtin_memref::set_base_and_offset): Same. Handle inner references.
22700 (builtin_memref::builtin_memref): Factor out parts into
22701 set_base_and_offset and call it.
22703 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22705 PR middle-end/84406
22706 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
22707 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
22708 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
22709 search at the associated MODE_INT.
22711 2018-02-20 Jeff Law <law@redhat.com>
22713 PR middle-end/82123
22714 PR tree-optimization/81592
22715 PR middle-end/79257
22716 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
22717 for range data rather than using global data.
22718 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
22719 range data rather than using global data.
22720 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
22721 pass it to children as needed.
22722 (struct directive::fmtresult): Similarly.
22723 (struct directive::set_width): Similarly.
22724 (struct directive::set_precision): Similarly.
22725 (format_integer, format_directive, parse_directive): Similarly.
22726 (format_none): Accept unnamed vr_values parameter.
22727 (format_percent, format_floating, format_character): Similarly.
22728 (format_string, format_plain): Similarly.
22729 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
22730 the EVRP range analyzer for range data rather than using global data.
22731 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
22732 gimple-ssa-evrp-analyze.h
22733 (class sprintf_dom_walker): Add after_dom_children member function.
22734 Add evrp_range_analyzer member.
22735 (sprintf_dom_walker::before_dom_children): Call into the EVRP
22736 range analyzer as needed.
22737 (sprintf_dom_walker::after_dom_children): New member function.
22738 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
22740 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
22741 (evrp_range_analyzer::pop_to_marker): Likewise.
22743 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
22745 PR tree-optimization/84419
22746 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
22747 with the required type if its current type is compatible but
22750 2018-02-20 Jakub Jelinek <jakub@redhat.com>
22752 PR middle-end/82004
22753 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
22754 after vectorization.
22756 2018-02-20 Martin Liska <mliska@suse.cz>
22759 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
22760 possible values if we don't have a hint.
22762 2018-02-20 Martin Liska <mliska@suse.cz>
22766 * final.c (shorten_branches): Build align_tab array with one
22768 * opts.c (finish_options): Add alignment option limit check.
22769 (MAX_CODE_ALIGN): Likewise.
22770 (MAX_CODE_ALIGN_VALUE): Likewise.
22771 * doc/invoke.texi: Document maximum allowed option value for
22772 all -falign-* options.
22774 2018-02-19 Jakub Jelinek <jakub@redhat.com>
22777 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
22778 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
22779 * var-tracking.c (emit_note_insn_var_location): Remove all references
22780 to NOTE_INSN_CALL_ARG_LOCATION.
22781 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
22782 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
22783 Use copy_rtx_if_shared.
22784 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
22785 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
22786 (dwarf2out_var_location): Remove handling of
22787 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
22789 * final.c (final_scan_insn): Remove all references to
22790 NOTE_INSN_CALL_ARG_LOCATION.
22791 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
22792 before dumping final insns.
22793 * except.c (emit_note_eh_region_end): Remove all references to
22794 NOTE_INSN_CALL_ARG_LOCATION.
22795 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
22796 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
22797 * config/arc/arc.c (hwloop_optimize): Likewise.
22798 * config/arm/arm.c (create_fix_barrier): Likewise.
22799 * config/s390/s390.c (s390_chunkify_start): Likewise.
22800 * config/sh/sh.c (find_barrier): Likewise.
22801 * config/i386/i386.c (rest_of_insert_endbranch,
22802 ix86_seh_fixup_eh_fallthru): Likewise.
22803 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
22804 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
22805 * config/frv/frv.c (frv_function_prologue): Likewise.
22806 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
22808 (note_outside_basic_block_p): Remove all references to
22809 NOTE_INSN_CALL_ARG_LOCATION.
22810 * gengtype.c (adjust_field_rtx_def): Likewise.
22811 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
22813 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
22814 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
22817 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
22820 PR tree-optimization/84452
22821 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
22822 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
22825 2018-02-19 Martin Liska <mliska@suse.cz>
22828 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
22830 2018-02-19 Martin Liska <mliska@suse.cz>
22831 Richard Sandiford <richard.sandiford@linaro.org>
22833 PR tree-optimization/82491
22834 * gimple-fold.c (get_base_constructor): Make earlier bail out
22837 2018-02-19 Carl Love <cel@us.ibm.com>
22839 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
22840 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
22842 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
22843 P8V_BUILTIN_VEC_NEG.
22845 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
22847 * config/rl78/rl78.md (movdf): New define expand.
22849 2018-02-19 Martin Liska <mliska@suse.cz>
22852 * doc/invoke.texi: Fix typo.
22853 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
22855 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
22857 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
22858 handle rs6000_single_float and rs6000_double_float specially for
22861 2018-02-16 Jeff Law <law@redhat.com>
22863 * config/rx/rx.c (add_pop_cfi_notes): New function.;
22864 (pop_regs): Use it.
22866 2018-02-16 Jakub Jelinek <jakub@redhat.com>
22869 * ipa-inline.c (inline_small_functions): Fix a typo.
22871 2018-02-16 Nathan Sidwell <nathan@acm.org>
22873 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
22875 2018-02-16 Carl Love <cel@us.ibm.com>
22877 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
22878 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
22879 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
22880 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
22881 expansion to P8V_BUILTIN_VEC_FLOAT2.
22883 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
22885 PR rtl-optimization/70023
22886 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
22887 src_regno into account.
22889 2018-02-16 Carl Love <cel@us.ibm.com>
22891 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
22892 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
22893 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
22894 * config/rs6000/rs6000.c: Remove case statements for
22895 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
22896 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
22897 and P9V_BUILTIN_VEC_VINSERT4B.
22898 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
22899 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
22900 * config/rs6000/vsx.md:
22901 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
22904 2018-02-16 Carl Love <cel@us.ibm.com>
22906 * config/rs6000/altivec.h: Add builtin names vec_extract4b
22908 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
22910 * config/rs6000/rs6000-c.c: Add the definitions for
22911 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
22912 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
22913 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
22914 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
22915 definition for insert4b and define insn *insert3b_internal.
22916 * doc/extend.texi: Add documentation for vec_extract4b.
22918 2018-02-16 Nathan Sidwell <nathan@acm.org>
22920 * doc/extend.texi (Backwards Compatibility): Mention friend
22921 injection. Note for-scope is deprecated.
22922 * doc/invoke.texi (-ffriend-injection): Deprecate.
22924 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
22926 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
22927 that moved to I2, also allow destinations that are a paradoxical
22928 subreg (instead of a normal reg).
22930 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
22933 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
22936 2018-02-16 Richard Biener <rguenther@suse.de>
22938 PR tree-optimization/84037
22939 PR tree-optimization/84016
22941 * config/i386/i386.c (ix86_builtin_vectorization_cost):
22942 Adjust vec_construct for the fact we need additional higher latency
22943 128bit inserts for AVX256 and AVX512 vector builds.
22944 (ix86_add_stmt_cost): Scale vector construction cost for
22947 2018-02-16 Richard Biener <rguenther@suse.de>
22949 PR tree-optimization/84417
22950 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
22951 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
22952 (non_rewritable_lvalue_p): Likewise, use poly-ints.
22954 2018-02-16 Martin Liska <mliska@suse.cz>
22957 * internal-fn.def (ASAN_CHECK): Set proper flags.
22958 (ASAN_MARK): Likewise.
22960 2018-02-16 Julia Koval <julia.koval@intel.com>
22962 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
22963 from PTA_CANNONLAKE.
22965 2018-02-16 Jakub Jelinek <jakub@redhat.com>
22968 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
22969 Use ++iter rather than iter++ for std::list iterators.
22970 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
22971 defer deleting them until all nodes in the forest are processed. Do
22972 free even leaf nodes. Change to_process into auto_vec.
22975 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
22976 * vec.h (vec_default_construct): Use memset instead of placement new
22977 if BROKEN_VALUE_INITIALIZATION is defined.
22978 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
22979 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
22982 PR rtl-optimization/83723
22983 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
22984 * lra.c (lra_substitute_pseudo): Likewise. If true, use
22985 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
22987 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
22989 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
22991 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
22993 PR rtl-optimization/81443
22994 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
22995 from inner REGs to paradoxical SUBREGs.
22997 2018-02-16 Richard Biener <rguenther@suse.de>
22999 PR tree-optimization/84399
23000 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
23001 For operands we can analyze at their definition make sure we can
23002 analyze them at each use as well.
23004 2018-02-16 Richard Biener <rguenther@suse.de>
23006 PR tree-optimization/84190
23007 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
23008 volatile accesses if the decl isn't volatile.
23010 2018-02-15 Jason Merrill <jason@redhat.com>
23012 PR c++/84314 - ICE with templates and fastcall attribute.
23013 * attribs.c (build_type_attribute_qual_variant): Don't clobber
23014 TYPE_CANONICAL on an existing type.
23016 2018-02-15 Jakub Jelinek <jakub@redhat.com>
23018 PR tree-optimization/84383
23019 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
23020 dstoff nor call operand_equal_p if dstbase is NULL.
23022 PR tree-optimization/84334
23023 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
23024 also a CONSTANT_CLASS_P, punt.
23026 2018-02-14 Jim Wilson <jimw@sifive.com>
23028 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
23029 first SMALL_OPERAND check. New local min_second_step. Move assert
23030 to where locals are set. Add TARGET_RVC support.
23031 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
23033 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
23035 * doc/invoke.texi: Correct -Wformat-overflow code sample.
23037 2018-02-14 Martin Sebor <msebor@redhat.com>
23039 PR tree-optimization/83698
23040 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
23041 arrays constrain the offset range to their bounds.
23042 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
23043 (builtin_access::overlap): Avoid setting the size of overlap if it's
23045 (maybe_diag_overlap): Also consider arrays when deciding what values
23046 of offsets to include in diagnostics.
23048 2018-02-14 Martin Sebor <msebor@redhat.com>
23051 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
23052 that correspond to the kind of a declaration.
23054 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
23057 * config/pa/pa.md: Load address of PIC label using the linkage table
23058 if the label is nonlocal.
23060 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
23062 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
23063 warning message if user requests -maltivec=be.
23064 * doc/invoke.texi: Document deprecation of -maltivec=be.
23066 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
23069 * config/rs6000/rs6000-c.c: Update definitions for
23070 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
23071 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
23073 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
23076 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
23077 add _get_ssp intrinsics. Remove argument from
23078 __builtin_ia32_rdssp[d|q].
23079 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
23080 * config/i386/i386-builtin.def: Remove argument from
23081 __builtin_ia32_rdssp[d|q].
23082 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
23083 ix86_expand_special_args_builtin for _rdssp[d|q].
23084 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
23085 Clear register before usage.
23086 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
23087 Add documentation for new _get_ssp and _inc_ssp intrinsics.
23089 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
23091 PR tree-optimization/84357
23092 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
23093 operand 1 of an ARRAY_REF too.
23095 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
23098 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
23099 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
23101 (set_of_reg): New struct.
23102 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
23103 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
23104 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
23106 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
23107 Split into bitclr, bitset, bitinvert patterns if appropriate.
23108 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
23109 use rx_fuse_in_memory_bitop.
23110 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
23111 to named insn, correct maximum insn length.
23113 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
23116 * machmode.def: Define a complex mode for PARTIAL_INT.
23117 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
23119 * doc/rtl.texi: Document CSPImode.
23120 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
23122 (msp430_hard_regno_nregs_with_padding): Likewise.
23124 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
23127 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
23129 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
23131 PR rtl-optimization/84169
23132 * combine.c (try_combine): New variable split_i2i3. Set it to true if
23133 we generated a parallel as new i3 and we split that to new i2 and i3
23134 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
23135 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
23136 those to i2, not i1. Partially rewrite this scan code.
23138 2018-02-13 Jakub Jelinek <jakub@redhat.com>
23141 * stor-layout.c (place_field): For variable length fields, adjust
23142 offset_align afterwards not just based on the field's alignment,
23143 but also on the size.
23145 PR middle-end/84309
23146 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
23147 of exps and logs in the use_exp2 case.
23149 2018-02-13 Jeff Law <law@redhat.com>
23151 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
23152 entry for "vector".
23154 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
23157 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
23161 * common.opt (gas-loc-support, gas-locview-support): New.
23162 (ginline-points, ginternal-reset-location-views): New.
23163 * doc/invoke.texi: Document them. Use @itemx where intended.
23164 (gvariable-location-views): Adjust.
23165 * target.def (reset_location_view): New.
23166 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
23167 (TARGET_RESET_LOCATION_VIEW): New.
23168 * doc/tm.texi: Rebuilt.
23169 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
23170 (dwarf2out_default_as_locview_support): New.
23171 (output_asm_line_debug_info): Use option variables.
23172 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
23173 (output_loc_list): Likewise.
23174 (add_high_low_attributes): Check option variables.
23175 Don't output entry view attribute in strict mode.
23176 (gen_inlined_subroutine_die): Check option variables.
23177 (dwarf2out_inline_entry): Likewise.
23178 (init_sections_and_labels): Likewise.
23179 (dwarf2out_early_finish): Likewise.
23180 (maybe_reset_location_view): New, from...
23181 (dwarf2out_var_location): ... here. Call it.
23182 * debug.h (dwarf2out_default_as_loc_support): Declare.
23183 (dwarf2out_default_as_locview_support): Declare.
23184 * hooks.c (hook_int_rtx_insn_0): New.
23185 * hooks.h (hook_int_rtx_insn_0): Declare.
23186 * toplev.c (process_options): Take -gas-loc-support and
23187 -gas-locview-support from dwarf2out. Enable
23188 -gvariable-location-views by default only with locview
23189 assembler support. Enable -ginternal-reset-location-views by
23190 default only if the target defines the corresponding hook.
23191 Enable -ginline-points by default if location views are
23192 enabled; force it disabled if statement frontiers are
23194 * tree-inline.c (expand_call_inline): Check option variables.
23195 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
23197 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
23199 PR tree-optimization/84321
23200 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
23201 handling. Also check whether the anti-range contains any values
23202 that satisfy the mask; switch to a VR_RANGE if not.
23204 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
23207 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
23209 2018-02-13 Martin Jambor <mjambor@suse.cz>
23212 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
23213 of call statements, also set location of a load to a temporary.
23215 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
23217 * config/rl78/rl78.c (add_vector_labels): New function.
23218 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
23219 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
23220 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
23221 which checks that no arguments are passed.
23222 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
23223 * doc/extend.texi: Documentation for the new attribute.
23225 2018-02-13 Andreas Schwab <schwab@suse.de>
23227 * config/riscv/linux.h (CPP_SPEC): Define.
23229 2018-02-13 Jakub Jelinek <jakub@redhat.com>
23232 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
23233 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
23234 OPTION_MASK_ISA_AES as first argument to def_builtin_const
23235 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
23236 instead of OPTION_MASK_ISA_PCLMUL as first argument to
23237 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
23238 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
23239 temporarily for AES and PCLMUL builtins.
23241 PR tree-optimization/84339
23242 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
23243 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
23246 PR middle-end/84309
23247 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
23248 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
23249 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
23251 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
23253 * omp-simd-clone.h: New file.
23254 * omp-simd-clone.c: Include omp-simd-clone.h.
23255 (expand_simd_clones): No longer static.
23256 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
23257 cgraph.h and omp-simd-clone.h.
23258 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
23259 (vect_recog_widen_shift_pattern): Formatting fix.
23260 (vect_pattern_recog_1): Don't check optab for calls.
23263 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
23264 operands[2] into a REG before using gen_lowpart on it.
23266 2018-02-12 Jeff Law <law@redhat.com>
23269 * config/sh/sh.c (find_barrier): Consider a sibling call
23272 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
23273 successfully back substituting a reg.
23275 2018-02-12 Richard Biener <rguenther@suse.de>
23277 PR tree-optimization/84037
23278 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
23279 parameter, move visited init to caller.
23280 (vect_slp_analyze_operations): Separate cost from validity
23281 check, initialize visited once for all instances.
23282 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
23284 * tree-vect-stmts.c (vect_model_simple_cost): Make early
23286 (vect_model_promotion_demotion_cost): Likewise.
23287 (vectorizable_bswap): Guard cost modeling with !slp_node
23288 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
23290 (vectorizable_call): Likewise.
23291 (vectorizable_conversion): Likewise.
23292 (vectorizable_assignment): Likewise.
23293 (vectorizable_shift): Likewise.
23294 (vectorizable_operation): Likewise.
23295 (vectorizable_store): Likewise.
23296 (vectorizable_load): Likewise.
23297 (vectorizable_condition): Likewise.
23298 (vectorizable_comparison): Likewise.
23300 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
23303 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
23304 (ASAN_MARK): Fix fnspec to account for return value, change pointer
23305 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
23307 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
23309 PR middle-end/83665
23310 * params.def (inline-min-speedup): Increase from 8 to 15.
23311 (max-inline-insns-auto): Decrease from 40 to 30.
23312 * ipa-split.c (consider_split): Add some buffer for function to
23313 be considered inlining candidate.
23314 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
23317 2018-02-12 Richard Biener <rguenther@suse.de>
23319 PR tree-optimization/84037
23320 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
23321 matched stmts if we cannot swap the non-matched ones.
23323 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
23325 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
23326 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
23327 _mm_maskz_scalef_round_ss): New intrinsics.
23328 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
23329 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
23330 __builtin_ia32_scalefss_round): Remove.
23331 (__builtin_ia32_scalefsd_mask_round,
23332 __builtin_ia32_scalefss_mask_round): New intrinsics.
23333 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
23334 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
23335 ((match_operand:VF_128 2 "<round_nimm_predicate>"
23336 "<round_constraint>")): Changed to ...
23337 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
23338 "<round_scalar_constraint>")): ... this.
23339 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
23340 %0, %1, %2<round_op3>}"): Changed to ...
23341 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
23342 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
23343 %2<round_scalar_mask_op3>}"): ... this.
23344 * config/i386/subst.md (round_scalar_nimm_predicate): New.
23346 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
23348 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
23349 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
23350 (_mm_maskz_sqrt_round_ss): New intrinsics.
23351 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
23352 (__builtin_ia32_sqrtsd_mask_round)
23353 (__builtin_ia32_sqrtss_mask_round): New builtins.
23354 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
23355 (__builtin_ia32_sqrtss_round): Remove.
23356 (__builtin_ia32_sqrtsd_mask_round)
23357 (__builtin_ia32_sqrtss_mask_round): New builtins.
23358 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
23359 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
23360 ((match_operand:VF_128 1 "vector_operand"
23361 "xBm,<round_constraint>")): Changed to ...
23362 ((match_operand:VF_128 1 "vector_operand"
23363 "xBm,<round_scalar_constraint>")): ... this.
23364 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
23365 %0, %2, %<iptr>1<round_op3>}): Changed to ...
23366 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
23367 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
23368 %<iptr>1<round_scalar_mask_op3>}): ... this.
23369 ((set_attr "prefix" "<round_prefix>")): Changed to ...
23370 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
23372 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
23375 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
23376 Cast vec_cmpeq result to correct type.
23377 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
23378 Cast vec_cmpgt result to correct type.
23380 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
23382 * final.c (final_scan_insn_1): Renamed from...
23383 (final_scan_insn): ... this. New wrapper, to recover
23384 seen from the outermost call in recursive ones.
23385 * config/sparc/sparc.c (output_return): Drop seen from call.
23386 (output_sibcall): Likewise.
23387 * config/visium/visium.c (output_branch): Likewise.
23389 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
23391 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
23394 2018-02-10 Alan Modra <amodra@gmail.com>
23397 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
23398 Specify LR as an input.
23400 2018-02-10 Jakub Jelinek <jakub@redhat.com>
23403 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
23404 remove_member_access_dummy_vars): New functions.
23405 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
23406 lower_omp_1, execute_lower_omp): Use them.
23408 PR rtl-optimization/84308
23409 * shrink-wrap.c (spread_components): Release todo vector.
23411 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
23413 PR rtl-optimization/57193
23414 * ira-color.c (struct allocno_color_data): Add member
23415 conflict_allocno_hard_prefs.
23416 (update_conflict_allocno_hard_prefs): New.
23417 (bucket_allocno_compare_func): Add a preference based on
23418 conflict_allocno_hard_prefs.
23419 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
23420 (color_allocnos): Remove a dead code. Initiate
23421 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
23423 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23426 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
23427 constraint from =wa to wa. Avoid a subreg on the output operand,
23428 instead use a pseudo and subreg it in a move.
23429 (p9_xxbrd_<mode>): Changed to ...
23430 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
23431 (p9_xxbrd_v2df): New expander.
23432 (p9_xxbrw_<mode>): Changed to ...
23433 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
23434 (p9_xxbrw_v4sf): New expander.
23436 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23438 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
23440 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
23443 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
23444 multiply in 32-bit mode.
23445 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
23446 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
23449 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
23451 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
23452 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
23453 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
23454 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
23456 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
23459 * dwarf2out.c (is_trivial_indirect_ref): New function.
23460 (dwarf2out_late_global_decl): Do not generate a location
23461 attribute for variables that have a non-trivial DECL_VALUE_EXPR
23462 and that are not defined in the current unit.
23464 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23466 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
23467 instead of a libcall for UNORDERED.
23469 2018-02-09 Tamar Christina <tamar.christina@arm.com>
23472 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
23473 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
23475 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23478 * config/s390/s390.c (s390_set_current_function): Invoke
23479 s390_indirect_branch_settings also if fndecl didn't change.
23481 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23483 * config/rs6000/rs6000.md (blockage): Set length to zero.
23485 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
23487 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
23489 2018-02-09 Jakub Jelinek <jakub@redhat.com>
23492 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
23493 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
23497 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
23498 PARALLEL incoming that failed vt_get_decl_and_offset check.
23500 PR middle-end/84237
23501 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
23502 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
23504 (get_variable_section): For decls in named .bss* sections pass true as
23505 second argument to bss_initializer_p.
23507 2018-02-09 Marek Polacek <polacek@redhat.com>
23508 Jakub Jelinek <jakub@redhat.com>
23511 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
23512 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
23513 Sync some changes from cxx_fold_indirect_ref.
23515 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
23517 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
23519 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
23520 (BLOCK_INLINE_ENTRY_LABEL): New.
23521 (dwarf2out_var_location): Disregard inline entry markers.
23522 (inline_entry_data): New struct.
23523 (inline_entry_data_hasher): New hashtable type.
23524 (inline_entry_data_hasher::hash): New.
23525 (inline_entry_data_hasher::equal): New.
23526 (inline_entry_data_table): New variable.
23527 (add_high_low_attributes): Add DW_AT_entry_pc and
23528 DW_AT_GNU_entry_view attributes if a pending entry is found
23529 in inline_entry_data_table. Add old entry_pc attribute only
23530 if debug nonbinding markers are disabled.
23531 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
23532 markers are enabled.
23533 (block_within_block_p, dwarf2out_inline_entry): New.
23534 (dwarf2out_finish): Check that no entries remained in
23535 inline_entry_data_table.
23536 * final.c (reemit_insn_block_notes): Handle inline entry notes.
23537 (final_scan_insn, notice_source_line): Likewise.
23538 (rest_of_clean_state): Skip inline entry markers.
23539 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
23541 * gimple.c (gimple_build_debug_inline_entry): New.
23542 * gimple.h (enum gimple_debug_subcode): Add
23543 GIMPLE_DEBUG_INLINE_ENTRY.
23544 (gimple_build_debug_inline_entry): Declare.
23545 (gimple_debug_inline_entry_p): New.
23546 (gimple_debug_nonbind_marker_p): Adjust.
23547 * insn-notes.def (INLINE_ENTRY): New.
23548 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
23549 inline entry marker notes.
23550 (print_insn): Likewise.
23551 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
23552 (INSN_DEBUG_MARKER_KIND): Likewise.
23553 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
23554 * tree-inline.c (expand_call_inline): Build and insert
23555 debug_inline_entry stmt.
23556 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
23557 inline entry blocks early, if nonbind markers are enabled.
23558 (dump_scope_block): Dump fragment info.
23559 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
23560 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
23561 (gimple_build_debug_inline_entry): New.
23562 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
23563 Enable/disable inline entry points too.
23564 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
23565 (DEBUG_INSN): Describe inline entry markers.
23567 * common.opt (gvariable-location-views): New.
23568 (gvariable-location-views=incompat5): New.
23569 * config.in: Rebuilt.
23570 * configure: Rebuilt.
23571 * configure.ac: Test assembler for view support.
23572 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
23573 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
23574 * dwarf2out.c (var_loc_view): New typedef.
23575 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
23576 (dwarf2out_locviews_in_attribute): New.
23577 (dwarf2out_locviews_in_loclist): New.
23578 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
23579 (enum dw_line_info_opcode): Add LI_adv_address.
23580 (struct dw_line_info_table): Add view.
23581 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
23582 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
23583 (zero_view_p): New variable.
23584 (ZERO_VIEW_P): New macro.
23585 (output_asm_line_debug_info): New.
23586 (struct var_loc_node): Add view.
23587 (add_AT_view_list, AT_loc_list): New.
23588 (add_var_loc_to_decl): Add view param. Test it against last.
23589 (new_loc_list): Add view params. Record them.
23590 (AT_loc_list_ptr): Handle loc and view lists.
23591 (view_list_to_loc_list_val_node): New.
23592 (print_dw_val): Handle dw_val_class_view_list.
23593 (size_of_die): Likewise.
23594 (value_format): Likewise.
23595 (loc_list_has_views): New.
23596 (gen_llsym): Set vl_symbol too.
23597 (maybe_gen_llsym, skip_loc_list_entry): New.
23598 (dwarf2out_maybe_output_loclist_view_pair): New.
23599 (output_loc_list): Output view list or entries too.
23600 (output_view_list_offset): New.
23601 (output_die): Handle dw_val_class_view_list.
23602 (output_dwarf_version): New.
23603 (output_compilation_unit_header): Use it.
23604 (output_skeleton_debug_sections): Likewise.
23605 (output_rnglists, output_line_info): Likewise.
23606 (output_pubnames, output_aranges): Update version comments.
23607 (output_one_line_info_table): Output view numbers in asm comments.
23608 (dw_loc_list): Determine current endview, pass it to new_loc_list.
23609 Call maybe_gen_llsym.
23610 (loc_list_from_tree_1): Adjust.
23611 (add_AT_location_description): Create view list attribute if
23612 needed, check it's absent otherwise.
23613 (convert_cfa_to_fb_loc_list): Adjust.
23614 (maybe_emit_file): Call output_asm_line_debug_info for test.
23615 (dwarf2out_var_location): Reset views as needed. Precompute
23616 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
23617 attribute. Set view.
23618 (new_line_info_table): Reset next view.
23619 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
23620 (dwarf2out_source_line): Likewise. Output view resets and labels to
23621 the assembler, or select appropriate line info opcodes.
23622 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
23623 (optimize_string_length): Catch it. Adjust.
23624 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
23625 dw_val_class_view_list, and remove it if no longer needed.
23626 (hash_loc_list): Hash view numbers.
23627 (loc_list_hasher::equal): Compare them.
23628 (optimize_location_lists): Check whether a view list symbol is
23629 needed, and whether the locview attribute is present, and
23630 whether they match. Remove the locview attribute if no longer
23632 (index_location_lists): Call skip_loc_list_entry for test.
23633 (dwarf2out_finish): Call output_asm_line_debug_info for test.
23634 Use output_dwarf_version.
23635 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
23636 (struct dw_val_node): Add val_view_list.
23637 * final.c (SEEN_NEXT_VIEW): New.
23638 (set_next_view_needed): New.
23639 (clear_next_view_needed): New.
23640 (maybe_output_next_view): New.
23641 (final_start_function): Rename to...
23642 (final_start_function_1): ... this. Take pointer to FIRST,
23643 add SEEN parameter. Emit param bindings in the initial view.
23644 (final_start_function): Reintroduce SEEN-less interface.
23645 (final): Rename to...
23646 (final_1): ... this. Take SEEN parameter. Output final pending
23647 next view at the end.
23648 (final): Reintroduce seen-less interface.
23649 (final_scan_insn): Output pending next view before switching
23650 sections or ending a block. Mark the next view as needed when
23651 outputting variable locations. Notify debug backend of section
23652 changes, and of location view changes.
23653 (rest_of_handle_final): Adjust.
23654 * toplev.c (process_options): Autodetect value for debug variable
23655 location views option. Warn on incompat5 without -gdwarf-5.
23656 * doc/invoke.texi (gvariable-location-views): New.
23657 (gvariable-location-views=incompat5): New.
23658 (gno-variable-location-views): New.
23660 2018-02-08 David Malcolm <dmalcolm@redhat.com>
23662 PR tree-optimization/84136
23663 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
23664 that the result of find_edge is non-NULL.
23666 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23669 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
23670 storing integer register in SImode. Fix cost of 256 and 512
23671 byte aligned SSE register store.
23673 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
23675 * config/i386/i386.c (ix86_multiplication_cost): Fix
23676 multiplication cost for TARGET_AVX512DQ.
23678 2018-02-08 Marek Polacek <polacek@redhat.com>
23680 PR tree-optimization/84238
23681 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
23684 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23686 PR tree-optimization/84265
23687 * tree-vect-stmts.c (vectorizable_store): Don't treat
23688 VMAT_CONTIGUOUS accesses as grouped.
23689 (vectorizable_load): Likewise.
23691 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23693 PR tree-optimization/81635
23694 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
23695 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
23696 (test_round_for_mask): New functions.
23697 (wide_int_cc_tests): Call test_round_for_mask.
23698 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
23699 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
23700 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
23701 range returned by get_range_info.
23703 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
23706 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
23707 * symtab.c: Include builtins.h
23708 (symtab_node::output_to_lto_symbol_table_p): Move here
23709 from lto-streamer-out.c:output_symbol_p.
23710 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
23711 (output_symbol_p): Move all logic to symtab.c
23712 (produce_symtab): Update.
23714 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23716 * config/s390/s390-opts.h (enum indirect_branch): Define.
23717 * config/s390/s390-protos.h (s390_return_addr_from_memory)
23718 (s390_indirect_branch_via_thunk)
23719 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
23720 (enum s390_indirect_branch_type): Define.
23721 * config/s390/s390.c (struct s390_frame_layout, struct
23722 machine_function): Remove.
23723 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
23724 (indirect_branch_table_label_no, indirect_branch_table_name):
23726 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
23727 (enum s390_indirect_branch_option): Define.
23728 (s390_return_addr_from_memory): New function.
23729 (s390_handle_string_attribute): New function.
23730 (s390_attribute_table): Add new attribute handler.
23731 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
23732 (s390_indirect_branch_via_thunk): New function.
23733 (s390_indirect_branch_via_inline_thunk): New function.
23734 (s390_function_ok_for_sibcall): When jumping via thunk disallow
23735 sibling call optimization for non z10 compiles.
23736 (s390_emit_call): Force indirect branch target to be a single
23737 register. Add r1 clobber for non-z10 compiles.
23738 (s390_emit_epilogue): Emit return jump via return_use expander.
23739 (s390_reorg): Handle JUMP_INSNs as execute targets.
23740 (s390_option_override_internal): Perform validity checks for the
23741 new command line options.
23742 (s390_indirect_branch_attrvalue): New function.
23743 (s390_indirect_branch_settings): New function.
23744 (s390_set_current_function): Invoke s390_indirect_branch_settings.
23745 (s390_output_indirect_thunk_function): New function.
23746 (s390_code_end): Implement target hook.
23747 (s390_case_values_threshold): Implement target hook.
23748 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
23750 * config/s390/s390.h (struct s390_frame_layout)
23751 (struct machine_function): Move here from s390.c.
23752 (TARGET_INDIRECT_BRANCH_NOBP_RET)
23753 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
23754 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
23755 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
23756 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
23757 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
23758 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
23759 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
23760 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
23761 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
23762 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
23763 (mnemonic attribute): Add values which aren't recognized
23765 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
23766 pattern for branch conversion. Fix mnemonic attribute.
23767 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
23768 indirect branch via thunk if requested.
23769 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
23770 ("*indirect_jump"): Disable for branch conversion using out of
23772 ("indirect_jump_via_thunk<mode>_z10")
23773 ("indirect_jump_via_thunk<mode>")
23774 ("indirect_jump_via_inlinethunk<mode>_z10")
23775 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
23776 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
23777 ("casesi_jump_via_inlinethunk<mode>_z10")
23778 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
23779 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
23780 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
23781 ("*indirect2_jump"): Disable for branch conversion.
23782 ("casesi_jump"): Turn into expander and expand patterns for branch
23784 ("return_use"): New expander.
23785 ("*return"): Emit return via thunk and rename it to ...
23786 ("*return<mode>"): ... this one.
23787 * config/s390/s390.opt: Add new options and and enum for the
23790 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
23792 * lra-constraints.c (match_reload): Unconditionally use
23793 gen_lowpart_SUBREG, rather than selecting between that
23794 and equivalent gen_rtx_SUBREG code.
23796 2018-02-08 Richard Biener <rguenther@suse.de>
23798 PR tree-optimization/84233
23799 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
23800 changed flag instead of boguously re-using phi_inserted.
23802 2018-02-08 Martin Jambor <mjambor@suse.cz>
23804 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
23805 static local variables.
23807 2018-02-08 Richard Biener <rguenther@suse.de>
23809 PR tree-optimization/84278
23810 * tree-vect-stmts.c (vectorizable_store): When looking for
23811 smaller vector types to perform grouped strided loads/stores
23812 make sure the mode is supported by the target.
23813 (vectorizable_load): Likewise.
23815 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23817 * config/aarch64/aarch64.c (aarch64_components_for_bb):
23818 Increase LDP/STP opportunities by adding adjacent callee-saves.
23820 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
23822 PR rtl-optimization/84068
23823 PR rtl-optimization/83459
23824 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
23826 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
23828 PR tree-optimization/84224
23829 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
23830 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
23831 non-zero arguments.
23833 2018-02-07 Iain Sandoe <iain@codesourcery.com>
23836 * config/rs6000/altivec.md (*restore_world): Remove LR use.
23837 * config/rs6000/predicates.md (restore_world_operation): Adjust op
23838 count, remove one USE.
23840 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
23842 * doc/install.texi (Configuration): Document the
23843 --with-long-double-format={ibm,ieee} PowerPC configuration
23847 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
23848 Convert from define_expand to be define_insn_and_split. Rework
23849 float/double/_Float128 conversions to QI/HI/SImode to work with
23850 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
23851 conversions to QI/HImode types did a store and then a load to
23852 truncate the value. For conversions to VSX registers, don't split
23853 the insn, instead emit the code directly. Use the code iterator
23854 any_fix to combine signed and unsigned conversions.
23855 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
23856 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
23857 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23858 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
23859 (fix_<mode>di2_hw): Likewise.
23860 (fixuns_<mode>di2_hw): Likewise.
23861 (fix_<mode>si2_hw): Likewise.
23862 (fixuns_<mode>si2_hw): Likewise.
23863 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
23864 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
23865 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
23866 fix<uns>_trunc<SFDF:mode>si2_p8.
23867 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
23868 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23869 (fix<uns>_<mode>_mem): Likewise.
23870 (fctiw<u>z_<mode>_mem): Likewise.
23871 (fix<uns>_<mode>_mem): Likewise.
23872 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
23873 the register allocator from doing a direct move to the GPRs to do
23874 a store, and instead use the ISA 3.0 store byte/half-word from
23875 vector register instruction. For IEEE 128-bit floating point,
23876 also optimize stores of 32-bit ints.
23877 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
23879 2018-02-07 Alan Hayward <alan.hayward@arm.com>
23881 * genextract.c (push_pathstr_operand): New function to support
23883 (walk_rtx): Call push_pathstr_operand.
23884 (print_path): Support [a-zA-Z].
23886 2018-02-07 Richard Biener <rguenther@suse.de>
23888 PR tree-optimization/84037
23889 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
23890 (cse_and_gimplify_to_preheader): Declare.
23891 (vect_get_place_in_interleaving_chain): Likewise.
23892 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
23894 (_loop_vec_info::~_loop_vec_info): Delete it.
23895 (cse_and_gimplify_to_preheader): New function.
23896 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
23897 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
23898 (vectorizable_load): Likewise. For grouped stores always base
23899 the IV on the first element.
23900 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
23901 condition before gimplifying.
23903 2018-02-07 Jakub Jelinek <jakub@redhat.com>
23905 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
23906 *DIV_EXPR and *MOD_EXPR.
23908 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
23911 * config/i386/i386.c (ix86_option_override_internal): Mask out
23912 the CF_SET bit when checking -fcf-protection.
23914 2018-02-07 Tom de Vries <tom@codesourcery.com>
23917 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
23920 2018-02-07 Richard Biener <rguenther@suse.de>
23922 PR tree-optimization/84204
23923 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
23926 PR tree-optimization/84205
23927 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
23928 special-case isl_ast_op_zdiv_r.
23930 PR tree-optimization/84223
23931 * graphite-scop-detection.c (gather_bbs::before_dom_children):
23932 Only add conditions from within the region.
23933 (gather_bbs::after_dom_children): Adjust.
23935 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
23938 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
23939 * config/avr/avr.md: Only post-reload split REG-REG moves if
23940 either register is GENERAL_REG_P.
23942 2018-02-07 Jakub Jelinek <jakub@redhat.com>
23944 PR tree-optimization/84235
23945 * tree-ssa-scopedtables.c
23946 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
23947 if the subtraction is performed in floating point type where NaNs are
23948 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
23949 build 1. Formatting fix.
23951 2018-02-06 Jakub Jelinek <jakub@redhat.com>
23954 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
23955 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
23956 and skip it regardless of bb boundaries. Use CALL_P macro,
23957 don't test INSN_P (insn) together with CALL_P or JUMP_P check
23958 unnecessarily, formatting fix.
23960 2018-02-06 Michael Collison <michael.collison@arm.com>
23962 * config/arm/thumb2.md:
23963 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
23964 (*thumb_mov_notscc): Ditto.
23966 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
23969 * config/rs6000/rs6000.md (su code attribute): Use "u" for
23970 unsigned_fix, not "s".
23972 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23974 * configure.ac (gcc_fn_eh_frame_ro): New function.
23975 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
23976 correct .eh_frame permissions.
23977 * configure: Regenerate.
23979 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
23981 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
23982 irrelevant options.
23984 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23986 * config/rs6000/rs6000.c (rs6000_option_override_internal):
23987 Display warning message for -mno-speculate-indirect-jumps.
23989 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
23991 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
23993 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
23995 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
23997 PR tree-optimization/84225
23998 * tree-eh.c (find_trapping_overflow): Only call
23999 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
24001 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
24004 * config/i386/i386.c: Reimplement the check of possible options
24005 -mibt/-mshstk conbination. Change error messages.
24006 * doc/invoke.texi: Fix a typo: remove extra '='.
24008 2018-02-06 Marek Polacek <polacek@redhat.com>
24010 PR tree-optimization/84228
24011 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
24013 2018-02-06 Tamar Christina <tamar.christina@arm.com>
24016 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
24017 emitted arch directives.
24018 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
24019 __ARM_FEATURE_COPROC before changing architectures.
24021 2018-02-06 Richard Biener <rguenther@suse.de>
24023 * config/i386/i386.c (print_reg): Fix typo.
24024 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
24026 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
24028 * configure: Regenerate.
24030 2018-02-05 Martin Sebor <msebor@redhat.com>
24032 PR tree-optimization/83369
24033 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
24036 2018-02-05 Martin Liska <mliska@suse.cz>
24038 * doc/invoke.texi: Cherry-pick upstream r323995.
24040 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
24042 * ira.c (ira_init_register_move_cost): Adjust comment.
24044 2018-02-05 Martin Liska <mliska@suse.cz>
24046 PR gcov-profile/84137
24047 * doc/gcov.texi: Fix typo in documentation.
24049 2018-02-05 Martin Liska <mliska@suse.cz>
24051 PR gcov-profile/83879
24052 * doc/gcov.texi: Document necessity of --dynamic-list-data when
24053 using dlopen functionality.
24055 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
24057 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
24058 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
24059 _mm_maskz_range_ss, _mm_mask_range_round_ss,
24060 _mm_maskz_range_round_ss): New intrinsics.
24061 (__builtin_ia32_rangesd128_round)
24062 (__builtin_ia32_rangess128_round): Remove.
24063 (__builtin_ia32_rangesd128_mask_round,
24064 __builtin_ia32_rangess128_mask_round): New builtins.
24065 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
24066 __builtin_ia32_rangess128_round): Remove.
24067 (__builtin_ia32_rangesd128_mask_round,
24068 __builtin_ia32_rangess128_mask_round): New builtins.
24069 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
24070 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
24071 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
24072 "<round_saeonly_constraint>")): Changed to ...
24073 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
24074 "<round_saeonly_scalar_constraint>")): ... this.
24075 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
24076 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
24077 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
24078 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
24079 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
24081 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
24083 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
24085 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
24086 Remove all values except native, 8540 and 8548.
24088 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
24090 * config/i386/i386.c (ix86_output_function_return): Pass
24091 INVALID_REGNUM, instead of -1, as invalid register number to
24092 indirect_thunk_name and output_indirect_thunk.
24094 2018-02-02 Julia Koval <julia.koval@intel.com>
24096 * config.gcc: Add -march=icelake.
24097 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
24098 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
24099 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
24100 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
24101 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
24102 (processor_target_table): Add icelake.
24103 (ix86_option_override_internal): Handle new PTAs.
24104 (get_builtin_code_for_version): Handle icelake.
24105 (M_INTEL_COREI7_ICELAKE): New.
24106 (fold_builtin_cpu): Handle icelake.
24107 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
24108 * doc/invoke.texi: Add -march=icelake.
24110 2018-02-02 Julia Koval <julia.koval@intel.com>
24112 * config/i386/i386.c (ix86_option_override_internal): Change flags type
24113 to wide_int_bitmask.
24114 * wide-int-bitmask.h: New.
24116 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
24119 * config/i386/i386.md: Replace Pmode with word_mode in
24120 builtin_setjmp_setup and builtin_longjmp to support x32.
24122 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
24126 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
24128 (rs6000_supported_cpu_names): New static variable.
24129 (linux_cpu_translation_table): Likewise.
24130 (elf_platform) <cpu>: Define new static variable and use it.
24131 Translate kernel AT_PLATFORM name to canonical name if needed.
24132 Error if platform name is unknown.
24134 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
24137 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
24139 2018-02-01 Jeff Law <law@redhat.com>
24142 * config/i386/i386.c (release_scratch_register_on_entry): Add new
24143 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
24144 the scratch if RELEASE_VIA_POP is false.
24145 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
24146 If we have to save a temporary register, decrement SIZE appropriately.
24147 Pass new arguments to release_scratch_register_on_entry.
24148 (ix86_adjust_stack_and_probe): Likewise.
24149 (ix86_emit_probe_stack_range): Pass new arguments to
24150 release_scratch_register_on_entry.
24152 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
24154 PR rtl-optimization/84157
24155 * combine.c (change_zero_ext): Use REG_P predicate in
24156 front of HARD_REGISTER_P predicate.
24158 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
24160 * config/avr/avr.c (avr_option_override): Move disabling of
24161 -fdelete-null-pointer-checks to...
24162 * common/config/avr/avr-common.c (avr_option_optimization_table):
24165 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24167 PR tree-optimization/81635
24168 * tree-data-ref.c (split_constant_offset_1): For types that
24169 wrap on overflow, try to use range info to prove that wrapping
24172 2018-02-01 Renlin Li <renlin.li@arm.com>
24175 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
24176 TAILCALL_ADDR_REGS.
24177 (aarch64_register_move_cost): Likewise.
24178 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
24179 TAILCALL_ADDR_REGS.
24180 (REG_CLASS_NAMES): Likewise.
24181 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
24182 TAILCALL_ADDR_REGS. Remove IP registers.
24183 * config/aarch64/aarch64.md (Ucs): Update register constraint.
24185 2018-02-01 Richard Biener <rguenther@suse.de>
24187 * domwalk.h (dom_walker::dom_walker): Add additional constructor
24188 for specifying RPO order and allow NULL for that.
24189 * domwalk.c (dom_walker::dom_walker): Likewise.
24190 (dom_walker::walk): Handle NULL RPO order.
24191 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
24193 (rewrite_update_dom_walker): Likewise.
24194 (mark_def_dom_walker): Likewise.
24196 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24198 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
24199 (aarch64_maybe_expand_sve_subreg_move): Declare.
24200 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
24201 * config/aarch64/predicates.md (aarch64_any_register_operand): New
24203 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
24204 that are semantically a reverse operation.
24205 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
24206 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
24207 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
24209 (aarch64_can_change_mode_class): For big-endian, forbid changes
24210 between two SVE modes if they have different element sizes.
24212 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24214 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
24215 the TImode handling for big-endian targets.
24217 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24219 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
24220 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
24222 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
24223 Remove BSWAP handing for big-endian targets and use the form of
24224 LD1RQ appropariate for the mode.
24226 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24228 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
24229 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
24230 duplicated element.
24232 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
24235 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
24236 check for operands that need to go through aarch64_sve_reload_be.
24238 2018-02-01 Jakub Jelinek <jakub@redhat.com>
24240 PR tree-optimization/81661
24241 PR tree-optimization/84117
24242 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
24243 * tree-eh.c: Include gimplify.h.
24244 (find_trapping_overflow, replace_trapping_overflow,
24245 rewrite_to_non_trapping_overflow): New functions.
24246 * tree-vect-loop.c: Include tree-eh.h.
24247 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
24248 * tree-data-ref.c: Include tree-eh.h.
24249 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
24251 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
24253 PR rtl-optimization/84123
24254 * combine.c (change_zero_ext): Check if hard register satisfies
24255 can_change_dest_mode before calling gen_lowpart_SUBREG.
24257 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
24260 * ira.c (ira_init_register_move_cost): Remove assert.
24262 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
24264 PR rtl-optimization/84071
24265 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
24266 * doc/tm.texi: Regenerate.
24268 2018-01-31 Richard Biener <rguenther@suse.de>
24270 PR tree-optimization/84132
24271 * tree-data-ref.c (analyze_miv_subscript): Properly
24272 check whether evolution_function_is_affine_multivariate_p
24273 before calling gcd_of_steps_may_divide_p.
24275 2018-01-31 Julia Koval <julia.koval@intel.com>
24278 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
24279 * config/i386/i386.md (rdpid_rex64) New.
24280 (rdpid): Make 32bit only.
24282 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
24285 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
24286 an IDENTIFIER_NODE for FUNCTION_TYPE's.
24288 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
24291 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
24293 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
24295 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
24297 PR rtl-optimization/84071
24298 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
24299 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
24301 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
24303 * config/arc/arc.c (arc_handle_aux_attribute): New function.
24304 (arc_attribute_table): Add 'aux' attribute.
24305 (arc_in_small_data_p): Consider aux like variables.
24306 (arc_is_aux_reg_p): New function.
24307 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
24308 (arc_get_aux_arg): New function.
24309 (prepare_move_operands): Handle aux-register access.
24310 (arc_handle_aux_attribute): New function.
24311 * doc/extend.texi (ARC Variable attributes): Add subsection.
24313 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
24315 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
24316 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
24317 (arc_attribute_table): Add 'uncached' attribute.
24318 (arc_print_operand): Print '.di' flag for uncached memory
24320 (arc_in_small_data_p): Do not consider for small data the uncached
24322 (arc_is_uncached_mem_p): New function.
24323 * config/arc/predicates.md (compact_store_memory_operand): Check
24324 for uncached memory accesses.
24325 (nonvol_nonimm_operand): Likewise.
24326 * doc/extend.texi (ARC Type Attribute): New subsection.
24328 2018-01-31 Jakub Jelinek <jakub@redhat.com>
24331 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
24332 falign-loops=): Add Optimization flag.
24334 2018-01-30 Jeff Law <law@redhat.com>
24337 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
24338 INT_REGISTERS_SAVED. Check it prior to calling
24339 get_scratch_register_on_entry.
24340 (ix86_adjust_stack_and_probe): Similarly.
24341 (ix86_emit_probe_stack_range): Similarly.
24342 (ix86_expand_prologue): Corresponding changes.
24344 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24347 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
24348 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
24350 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
24353 * lra-constraints.c (curr_insn_transform): Process AND in the
24356 2018-01-30 Jakub Jelinek <jakub@redhat.com>
24358 PR rtl-optimization/83986
24359 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
24360 dependence against last_pending_memory_flush in addition to
24361 pending_jump_insns.
24363 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
24365 PR tree-optimization/81611
24366 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
24369 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24372 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
24375 2018-01-30 Richard Biener <rguenther@suse.de>
24376 Jakub Jelinek <jakub@redhat.com>
24378 PR tree-optimization/84111
24379 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
24380 inner loops added during recursion, as they don't have up-to-date
24383 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
24386 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
24387 (can_inline_edge_by_limits_p): ... here.
24388 (can_early_inline_edge_p, check_callers,
24389 update_caller_keys, update_callee_keys, recursive_inlining,
24390 add_new_edges_to_heap, speculation_useful_p,
24391 inline_small_functions,
24392 inline_small_functions, flatten_function,
24393 inline_to_all_callers_1): Update.
24395 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
24397 * profile-count.c (profile_count::combine_with_ipa_count): Handle
24400 2018-01-30 Richard Biener <rguenther@suse.de>
24402 PR tree-optimization/83008
24403 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
24404 invariant and constant vector uses in stmts when they need
24405 more than one stmt.
24407 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24410 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
24411 * configure: Regenerate.
24413 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24415 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
24417 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
24418 Use gen_rtx_REG rather than gen_lowpart.
24420 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24422 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
24423 rather than 0 when creating partial subregs.
24425 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
24427 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
24430 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
24433 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
24434 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
24435 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
24436 flags. This restores the settings used before the 2017-07-24.
24437 Turning off pre increment/decrement/modify allows IVOPTS to
24438 optimize DF/SF loops where the index is an int.
24440 2018-01-29 Richard Biener <rguenther@suse.de>
24441 Kelvin Nilsen <kelvin@gcc.gnu.org>
24444 * tree-vect-stmts.c (vectorizable_call): Don't call
24445 targetm.vectorize_builtin_md_vectorized_function if callee is
24448 2018-01-22 Carl Love <cel@us.ibm.com>
24450 * doc/extend.tex: Fix typo in second arg in
24451 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
24453 2018-01-29 Richard Biener <rguenther@suse.de>
24455 PR tree-optimization/84086
24456 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
24457 (flush_ssaname_freelist): When SSA names were released reset
24458 the SCEV hash table.
24460 2018-01-29 Richard Biener <rguenther@suse.de>
24462 PR tree-optimization/84057
24463 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
24464 removed paths when removing edges.
24466 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
24468 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
24469 -mfunction-return=@var{choice}.
24471 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
24473 PR diagnostic/84034
24474 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
24475 Handle CR like TAB.
24476 (layout::print_source_line): Likewise.
24477 (test_get_line_width_without_trailing_whitespace): Add test cases.
24479 2018-01-27 Jakub Jelinek <jakub@redhat.com>
24481 PR middle-end/84040
24482 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
24485 2018-01-26 Jim Wilson <jimw@sifive.com>
24487 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
24489 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
24492 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24494 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
24495 and CMP + SUB-immediate -> SUBS.
24497 2018-01-26 Martin Sebor <msebor@redhat.com>
24499 PR tree-optimization/83896
24500 * tree-ssa-strlen.c (get_string_len): Rename...
24501 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
24502 Avoid assuming length is constant.
24503 (handle_char_store): Use HOST_WIDE_INT for string length.
24505 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
24508 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
24509 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
24511 2018-01-26 Richard Biener <rguenther@suse.de>
24513 PR rtl-optimization/84003
24514 * dse.c (record_store): Only record redundant stores when
24515 the earlier store aliases at least all accesses the later one does.
24517 2018-01-26 Jakub Jelinek <jakub@redhat.com>
24519 PR rtl-optimization/83985
24520 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
24521 REG_CFA_RESTORE insns.
24522 (delete_unmarked_insns): Don't ignore separate shrink wrapping
24523 REG_CFA_RESTORE insns here.
24526 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
24527 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
24529 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24531 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
24532 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
24533 (arc_init): Likewise.
24534 (arc_override_options): Likewise.
24535 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
24537 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
24539 * config/arc/arc.h (TARGET_DBNZ): Define.
24540 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
24541 properly set the tune attribute.
24542 (dbnz): Use TARGET_DBNZ guard.
24543 * config/arc/arc.opt (mtune): Add core3 option.
24545 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24547 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
24548 recognize new pic like addresses.
24549 (arc_delegitimize_address): Clean up.
24551 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24553 * config/arc/arc-arches.def: Option mrf16 valid for all
24555 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
24556 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
24557 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
24558 * config/arc/arc-tables.opt: Regenerate.
24559 * config/arc/arc.c (arc_conditional_register_usage): Handle
24560 reduced register file case.
24561 (arc_file_start): Set must have build attributes.
24562 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
24563 mrf16 option value.
24564 * config/arc/arc.opt (mrf16): Add new option.
24565 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
24566 * config/arc/genmultilib.awk: Handle new mrf16 option.
24567 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
24568 * config/arc/t-multilib: Regenerate.
24569 * doc/invoke.texi (ARC Options): Document mrf16 option.
24571 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24573 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
24574 * config/arc/arc.c (arc_handle_secure_attribute): New function.
24575 (arc_attribute_table): Add 'secure_call' attribute.
24576 (arc_print_operand): Print secure call operand.
24577 (arc_function_ok_for_sibcall): Don't optimize tail calls when
24579 (arc_is_secure_call_p): New function. * config/arc/arc.md
24580 (call_i): Add support for sjli instruction.
24581 (call_value_i): Likewise.
24582 * config/arc/constraints.md (Csc): New constraint.
24584 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
24585 John Eric Martin <John.Martin@emmicro-us.com>
24587 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
24588 * config/arc/arc.c (_arc_jli_section): New struct.
24589 (arc_jli_section): New type.
24590 (rc_jli_sections): New static variable.
24591 (arc_handle_jli_attribute): New function.
24592 (arc_attribute_table): Add jli_always and jli_fixed attribute.
24593 (arc_file_end): New function.
24594 (TARGET_ASM_FILE_END): Define.
24595 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
24596 (arc_add_jli_section): New function.
24597 (jli_call_scan): Likewise.
24598 (arc_reorg): Call jli_call_scan.
24599 (arc_output_addsi): Remove 'S' from printing asm operand.
24600 (arc_is_jli_call_p): New function.
24601 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
24603 (movhi_insn): Likewise.
24604 (movsi_insn): Likewise.
24605 (movsi_set_cc_insn): Likewise.
24606 (loadqi_update): Likewise.
24607 (load_zeroextendqisi_update): Likewise.
24608 (load_signextendqisi_update): Likewise.
24609 (loadhi_update): Likewise.
24610 (load_zeroextendhisi_update): Likewise.
24611 (load_signextendhisi_update): Likewise.
24612 (loadsi_update): Likewise.
24613 (loadsf_update): Likewise.
24614 (movsicc_insn): Likewise.
24615 (bset_insn): Likewise.
24616 (bxor_insn): Likewise.
24617 (bclr_insn): Likewise.
24618 (bmsk_insn): Likewise.
24619 (bicsi3_insn): Likewise.
24620 (cmpsi_cc_c_insn): Likewise.
24621 (movsi_ne): Likewise.
24622 (movsi_cond_exec): Likewise.
24623 (clrsbsi2): Likewise.
24624 (norm_f): Likewise.
24634 (call_i): Remove 'S' asm letter, add jli instruction.
24635 (call_value_i): Likewise.
24636 * config/arc/arc.op (mjli-always): New option.
24637 * config/arc/constraints.md (Cji): New constraint.
24638 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
24640 (subsf3_fpx): Likewise.
24641 (mulsf3_fpx): Likewise.
24642 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
24644 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
24645 function attrbutes.
24646 * doc/invoke.texi (ARC): Document mjli-always option.
24648 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
24650 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
24651 avoid addition with 0 and use incw and decw where possible.
24653 2018-01-26 Richard Biener <rguenther@suse.de>
24655 PR tree-optimization/81082
24656 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
24657 association if it requires casting to unsigned.
24658 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
24659 from fold_plusminus_mult_expr to catch important cases late when
24660 range info is available.
24662 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24664 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
24665 * configure.ac (hidden_linkonce): New test.
24666 * configure: Regenerate.
24667 * config.in: Regenerate.
24669 2018-01-26 Julia Koval <julia.koval@intel.com>
24671 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
24672 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
24673 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
24674 _mm_mask_bitshuffle_epi64_mask): Fix type.
24675 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
24676 USI_FTYPE_V4DI_V4DI_USI): Remove.
24677 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
24678 __builtin_ia32_vpshufbitqmb256_mask,
24679 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
24680 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
24681 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
24683 2018-01-26 Alan Modra <amodra@gmail.com>
24686 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
24687 UNSPEC_VBPERMQ. Sort other unspecs.
24689 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
24691 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
24693 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
24695 PR middle-end/83055
24696 * predict.c (drop_profile): Do not push/pop cfun; update also
24698 (handle_missing_profiles): Fix logic looking for zero profiles.
24700 2018-01-25 Jakub Jelinek <jakub@redhat.com>
24702 PR middle-end/83977
24703 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
24704 on functions with #pragma omp declare simd or functions with simd
24706 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
24707 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
24708 Remove trailing \n from warning_at calls.
24710 2018-01-25 Tom de Vries <tom@codesourcery.com>
24713 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24714 for neutered workers.
24716 2018-01-24 Joseph Myers <joseph@codesourcery.com>
24719 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
24720 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
24722 2018-01-24 Jeff Law <law@redhat.com>
24725 * i386.c (get_probe_interval): Move to earlier point.
24726 (ix86_compute_frame_layout): If -fstack-clash-protection and
24727 the frame is larger than the probe interval, then use pushes
24728 to save registers rather than reg->mem moves.
24729 (ix86_expand_prologue): Remove conditional for int_registers_saved
24732 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
24735 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
24736 min/max for never referenced object.
24738 2018-01-24 Jakub Jelinek <jakub@redhat.com>
24740 PR middle-end/83977
24741 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
24743 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
24744 attributes from DECL_ATTRIBUTES (decl) without affecting
24745 DECL_ATTRIBUTES (current_function_decl).
24746 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
24747 functions with non-NULL DECL_ABSTRACT_ORIGIN.
24749 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
24751 PR tree-optimization/83979
24752 * fold-const.c (fold_comparison): Use constant_boolean_node
24753 instead of boolean_{true,false}_node.
24755 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
24757 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
24760 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24762 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
24763 Simplify the clause that sets the length attribute.
24764 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
24765 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
24766 clause that sets the length attribute.
24767 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
24769 2018-01-24 Tom de Vries <tom@codesourcery.com>
24772 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
24773 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
24774 Add strict parameter.
24775 (prevent_branch_around_nothing): Insert dummy insn between branch to
24776 label and label with no ptx insn inbetween.
24777 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
24779 2018-01-24 Tom de Vries <tom@codesourcery.com>
24782 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
24783 for neutered threads in warp.
24784 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
24786 2018-01-24 Richard Biener <rguenther@suse.de>
24788 PR tree-optimization/83176
24789 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
24792 2018-01-24 Richard Biener <rguenther@suse.de>
24794 PR tree-optimization/82819
24795 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
24796 code generating pluses that are no-ops in the target precision.
24798 2018-01-24 Richard Biener <rguenther@suse.de>
24800 PR middle-end/84000
24801 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
24803 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24805 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
24806 to merge probabilities.
24807 * predict.c (probably_never_executed): Also mark as cold functions
24808 with global 0 profile and guessed local profile.
24809 * profile-count.c (profile_probability::combine_with_count): New
24811 * profile-count.h (profile_probability::operator*,
24812 profile_probability::operator*=, profile_probability::operator/,
24813 profile_probability::operator/=): Reduce precision to adjusted
24814 and set value to guessed on contradictory divisions.
24815 (profile_probability::combine_with_freq): Remove.
24816 (profile_probability::combine_wiht_count): Declare.
24817 (profile_count::force_nonzero):: Set to adjusted.
24818 (profile_count::probability_in):: Set quality to adjusted.
24819 * tree-ssa-tail-merge.c (replace_block_by): Use
24820 combine_with_count.
24822 2018-01-23 Andrew Waterman <andrew@sifive.com>
24823 Jim Wilson <jimw@sifive.com>
24825 * config/riscv/riscv.c (riscv_stack_boundary): New.
24826 (riscv_option_override): Set riscv_stack_boundary. Handle
24827 riscv_preferred_stack_boundary_arg.
24828 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
24829 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
24830 (STACK_BOUNDARY): Set to riscv_stack_boundary.
24831 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
24832 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
24833 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
24835 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
24838 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
24839 of struct ix86_frame.
24840 (ix86_expand_epilogue): Likewise. Add a local variable for
24841 the reg_save_offset field in struct ix86_frame.
24843 2018-01-23 Bin Cheng <bin.cheng@arm.com>
24845 PR tree-optimization/82604
24846 * tree-loop-distribution.c (enum partition_kind): New enum item
24847 PKIND_PARTIAL_MEMSET.
24848 (partition_builtin_p): Support above new enum item.
24849 (generate_code_for_partition): Ditto.
24850 (compute_access_range): Differentiate cases that equality can be
24851 proven at all loops, the innermost loops or no loops.
24852 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
24853 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
24854 (finalize_partitions, distribute_loop): Don't fuse partition of
24855 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
24856 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
24857 parloop is enabled.
24859 2018-01-23 Martin Liska <mliska@suse.cz>
24861 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
24862 order to ignore the predictor.
24863 (PRED_POLYMORPHIC_CALL): Likewise.
24864 (PRED_RECURSIVE_CALL): Likewise.
24866 2018-01-23 Martin Liska <mliska@suse.cz>
24868 * tree-profile.c (tree_profiling): Print function header to
24869 aware reader which function we are working on.
24870 * value-prof.c (gimple_find_values_to_profile): Do not print
24871 not interesting value histograms.
24873 2018-01-23 Martin Liska <mliska@suse.cz>
24875 * profile-count.h (enum profile_quality): Add
24876 profile_uninitialized as the first value. Do not number values
24877 as they are zero based.
24878 (profile_count::verify): Update sanity check.
24879 (profile_probability::verify): Likewise.
24881 2018-01-23 Nathan Sidwell <nathan@acm.org>
24883 * doc/invoke.texi (ffor-scope): Deprecate.
24885 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24887 PR tree-optimization/83510
24888 * domwalk.c (set_all_edges_as_executable): New function.
24889 (dom_walker::dom_walker): Convert bool param
24890 "skip_unreachable_blocks" to enum reachability. Move setup of
24891 edge flags to set_all_edges_as_executable and only do it when
24892 reachability is REACHABLE_BLOCKS.
24893 * domwalk.h (enum dom_walker::reachability): New enum.
24894 (dom_walker::dom_walker): Convert bool param
24895 "skip_unreachable_blocks" to enum reachability.
24896 (set_all_edges_as_executable): New decl.
24897 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
24898 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
24900 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
24901 but converting true to REACHABLE_BLOCKS.
24902 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
24904 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
24905 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
24906 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
24908 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
24909 if check_all_array_refs will be called.
24911 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24913 * tree.c (selftest::test_location_wrappers): Add more test
24916 2018-01-23 David Malcolm <dmalcolm@redhat.com>
24918 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
24919 (selftest::test_bit_in_range): Likewise.
24921 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
24924 * doc/sourcebuild.texi (vect_float): Say that the selector
24925 only describes the situation when -funsafe-math-optimizations is on.
24926 (vect_float_strict): Document.
24928 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
24930 PR tree-optimization/83965
24931 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
24932 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
24933 instead of checking only for a reduction.
24934 (vect_recog_widen_sum_pattern): Likewise.
24936 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24938 * predict.c (probably_never_executed): Only use precise profile info.
24939 (compute_function_frequency): Skip after inlining hack since we now
24940 have quality checking.
24942 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
24944 * profile-count.h (profile_probability::very_unlikely,
24945 profile_probability::unlikely, profile_probability::even): Set
24946 precision to guessed.
24948 2018-01-23 Richard Biener <rguenther@suse.de>
24950 PR tree-optimization/83963
24951 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
24952 Properly terminate dominator walk when crossing the exit edge not
24953 when visiting its source block.
24955 2018-01-23 Jakub Jelinek <jakub@redhat.com>
24958 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
24959 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
24961 2018-01-22 Jakub Jelinek <jakub@redhat.com>
24963 PR tree-optimization/83957
24964 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
24965 semicolon after for body surrounded by braces.
24967 PR tree-optimization/83081
24968 * profile-count.h (profile_probability::split): New method.
24969 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
24970 Use profile_probability::split.
24971 (do_compare_rtx_and_jump): Fix adjustment of probabilities
24972 when splitting a single conditional jump into 2.
24974 2018-01-22 David Malcolm <dmalcolm@redhat.com>
24976 PR tree-optimization/69452
24977 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
24980 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24982 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
24983 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
24984 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
24986 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
24988 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
24990 * config/rl78/rl78.md (movdi): New define_expand.
24991 * config/rl78/rl78.c (rl78_split_movdi): New function.
24993 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
24996 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
24998 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
24999 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
25000 128-bit to produce an UNSPEC move to get the double word with the
25001 signbit and then a shift directly to do signbit.
25002 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
25003 implementation with a new version that just does either a direct
25004 move or a regular move. Move memory interface to separate insns.
25005 Move insns so they are next to the expander.
25006 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
25007 with signbit move. Split big and little endian case.
25008 (signbit<mode>2_dm_mem_le): Likewise.
25009 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
25010 (signbit<mode>2_dm2): Likewise.
25012 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25014 * config/rl78/rl78.md (anddi3): New define_expand.
25016 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25018 * config/rl78/rl78.md (umindi3): New define_expand.
25020 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25022 * config/rl78/rl78.md (smindi3): New define_expand.
25024 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25026 * config/rl78/rl78.md (smaxdi3): New define_expand.
25028 2018-01-22 Carl Love <cel@us.ibm.com>
25030 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
25031 LVX_V1TI): Add macro expansion.
25032 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
25033 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
25034 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
25035 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
25036 Change check to determine if the instruction is a byte reversing
25037 entry. Fix typo in comment.
25038 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
25039 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
25040 Add def_builtin calls for new builtins.
25041 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
25042 Add define_insn expansion.
25044 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25046 * config/rl78/rl78.md (umaxdi3): New define_expand.
25048 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
25050 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
25051 for non-QImode registers.
25053 2018-01-22 Richard Biener <rguenther@suse.de>
25055 PR tree-optimization/83963
25056 * graphite-scop-detection.c (scop_detection::get_sese): Delay
25057 including the loop exit block.
25058 (scop_detection::merge_sese): Likewise.
25059 (scop_detection::add_scop): Do it here instead.
25061 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25063 * doc/sourcebuild.texi (arm_softfloat): Document.
25065 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
25068 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
25069 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
25070 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
25072 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25073 David Edelsohn <dje.gcc@gmail.com>
25076 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
25077 Change "crset eq" to "crset 2".
25078 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
25079 (*call_indirect_aix<mode>_nospec): Likewise.
25080 (*call_value_indirect_aix<mode>_nospec): Likewise.
25081 (*call_indirect_elfv2<mode>_nospec): Likewise.
25082 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
25083 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
25084 change assembly output from . to $.
25085 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
25086 (indirect_jump<mode>_nospec): Change assembly output from . to $.
25087 (*tablejump<mode>_internal1_nospec): Likewise.
25089 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
25092 * config/sh/sh_optimize_sett_clrt.cc:
25093 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
25095 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
25097 PR tree-optimization/83940
25098 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
25099 offset_dt to vect_constant_def rather than vect_unknown_def_type.
25100 (vect_check_load_store_mask): Add a mask_dt_out parameter and
25101 use it to pass back the definition type.
25102 (vect_check_store_rhs): Likewise rhs_dt_out.
25103 (vect_build_gather_load_calls): Add a mask_dt argument and use
25104 it instead of a call to vect_is_simple_use.
25105 (vectorizable_store): Update calls to vect_check_load_store_mask
25106 and vect_check_store_rhs. Use the dt returned by the latter instead
25107 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
25108 instead of calls to vect_is_simple_use. Pass the scalar rather
25109 than the vector operand to vect_is_simple_use when handling
25110 second and subsequent copies of an rhs value.
25111 (vectorizable_load): Update calls to vect_check_load_store_mask
25112 and vect_build_gather_load_calls. Use the cached mask_dt and
25113 gs_info.offset_dt instead of calls to vect_is_simple_use.
25115 2018-01-20 Jakub Jelinek <jakub@redhat.com>
25117 PR middle-end/83945
25118 * tree-emutls.c: Include gimplify.h.
25119 (lower_emutls_2): New function.
25120 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
25121 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
25122 it before further processing.
25125 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
25126 UINTVAL (trueop1) instead of INTVAL (op1).
25128 2018-01-19 Jakub Jelinek <jakub@redhat.com>
25132 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
25133 INCOMING_FRAME_SP_OFFSET if not defined.
25134 (scan_trace): Add ENTRY argument. If true and
25135 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
25136 emit a note to adjust the CFA offset.
25137 (create_cfi_notes): Adjust scan_trace callers.
25138 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
25139 INCOMING_FRAME_SP_OFFSET in the CIE.
25140 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
25141 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
25143 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
25144 * doc/tm.texi: Regenerated.
25146 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25148 PR rtl-optimization/83147
25149 * lra-constraints.c (remove_inheritance_pseudos): Use
25150 lra_substitute_pseudo_within_insn.
25152 2018-01-19 Tom de Vries <tom@codesourcery.com>
25153 Cesar Philippidis <cesar@codesourcery.com>
25156 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
25158 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
25161 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
25162 spaces for function labels.
25164 2018-01-19 Martin Liska <mliska@suse.cz>
25166 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
25167 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
25168 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
25169 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
25170 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
25171 (PRED_CONST_RETURN): Change from 69 to 65.
25172 (PRED_NULL_RETURN): Change from 91 to 71.
25173 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
25174 (PRED_LOOP_GUARD): Change from 66 to 73.
25176 2018-01-19 Martin Liska <mliska@suse.cz>
25178 * predict.c (predict_insn_def): Add new assert.
25179 (struct branch_predictor): Change type to signed integer.
25180 (test_prediction_value_range): Amend test to cover
25181 PROB_UNINITIALIZED.
25182 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
25183 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
25184 (PRED_LOOP_ITERATIONS_MAX): Likewise.
25185 (PRED_LOOP_IV_COMPARE): Likewise.
25186 * predict.h (PROB_UNINITIALIZED): Define new constant.
25188 2018-01-19 Martin Liska <mliska@suse.cz>
25190 * predict.c (dump_prediction): Add new format for
25191 analyze_brprob.py script which is enabled with -details
25193 * profile-count.h (precise_p): New function.
25195 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
25197 PR tree-optimization/83922
25198 * tree-vect-loop.c (vect_verify_full_masking): Return false if
25199 there are no statements that need masking.
25200 (vect_active_double_reduction_p): New function.
25201 (vect_analyze_loop_operations): Use it when handling phis that
25202 are not in the loop header.
25204 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
25206 PR tree-optimization/83914
25207 * tree-vect-loop.c (vectorizable_induction): Don't convert
25208 init_expr or apply the peeling adjustment for inductions
25209 that are nested within the vectorized loop.
25211 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25213 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
25216 2018-01-18 Jakub Jelinek <jakub@redhat.com>
25220 * function.h (gimplify_parameters): Add gimple_seq * argument.
25221 * function.c: Include gimple.h and options.h.
25222 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
25223 for the added local temporaries if needed.
25224 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
25225 if there are any parameter cleanups, wrap whole body into a
25226 try/finally with the cleanups.
25228 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
25231 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
25232 Use GET_MODE_CLASS for scalar floating point.
25234 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
25238 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
25239 Fix call of call_cgraph_insertion_hooks.
25241 2018-01-18 Martin Sebor <msebor@redhat.com>
25243 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
25245 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
25248 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
25251 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
25254 * common.opt: (-ffile-prefix-map): New option.
25255 * opts.c (common_handle_option): Defer it.
25256 * opts-global.c (handle_common_deferred_options): Handle it.
25257 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
25258 * file-prefix-map.h: New file.
25259 (remap_debug_filename, add_debug_prefix_map): ...here.
25260 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
25261 * final.c (debug_prefix_map, add_debug_prefix_map
25262 remap_debug_filename): Move to...
25263 * file-prefix-map.c: New file.
25264 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
25265 generalize, get rid of alloca(), use strrchr() instead of strchr().
25266 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
25267 Implement in terms of add_prefix_map().
25268 (remap_macro_filename, remap_debug_filename): Implement in term of
25270 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
25271 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
25272 * dbxout.c: Include file-prefix-map.h.
25273 * varasm.c: Likewise.
25274 * vmsdbgout.c: Likewise.
25275 * xcoffout.c: Likewise.
25276 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
25277 * doc/cppopts.texi (-fmacro-prefix-map): Document.
25278 * doc/invoke.texi (-ffile-prefix-map): Document.
25279 (-fdebug-prefix-map): Update description.
25281 2018-01-18 Martin Liska <mliska@suse.cz>
25283 * config/i386/i386.c (indirect_thunk_name): Document that also
25285 (output_indirect_thunk): Document why both instructions
25286 (pause and lfence) are generated.
25288 2018-01-18 Richard Biener <rguenther@suse.de>
25290 PR tree-optimization/83887
25291 * graphite-scop-detection.c
25292 (scop_detection::get_nearest_dom_with_single_entry): Remove.
25293 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
25294 (scop_detection::merge_sese): Re-implement with a flood-fill
25295 algorithm that properly finds a SESE region if it exists.
25297 2018-01-18 Jakub Jelinek <jakub@redhat.com>
25300 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
25301 pointer_diff optimizations use view_convert instead of convert.
25303 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25305 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
25306 Generate different code for -mno-speculate-indirect-jumps.
25307 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
25308 (*call_indirect_aix<mode>): Disable for
25309 -mno-speculate-indirect-jumps.
25310 (*call_indirect_aix<mode>_nospec): New define_insn.
25311 (*call_value_indirect_aix<mode>): Disable for
25312 -mno-speculate-indirect-jumps.
25313 (*call_value_indirect_aix<mode>_nospec): New define_insn.
25314 (*sibcall_nonlocal_sysv<mode>): Generate different code for
25315 -mno-speculate-indirect-jumps.
25316 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
25318 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
25320 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
25321 long double type, set the flags for noting the default long double
25322 type, even if we don't pass or return a long double type.
25324 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
25327 * ipa-inline.c (flatten_function): Do not overwrite final inlining
25330 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
25332 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
25333 support for merge[hl].
25334 (fold_mergehl_helper): New helper function.
25335 (tree-vector-builder.h): New #include for tree_vector_builder usage.
25336 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
25337 (altivec_vmrglw_direct): Add xxmrglw insn.
25339 2018-01-17 Andrew Waterman <andrew@sifive.com>
25341 * config/riscv/riscv.c (riscv_conditional_register_usage): If
25342 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
25344 2018-01-17 David Malcolm <dmalcolm@redhat.com>
25347 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
25348 call the lto_location_cache before reading the
25349 DECL_SOURCE_LOCATION of the types.
25351 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
25352 Richard Sandiford <richard.sandiford@linaro.org>
25354 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
25355 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
25356 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
25357 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
25358 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
25360 * config/aarch64/constraints.md (aarch64_movti_operand):
25362 * config/aarch64/predicates.md (Uti): Add new constraint.
25364 2018-01-17 Carl Love <cel@us.ibm.com>
25366 * config/rs6000/vsx.md (define_expand xl_len_r,
25367 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
25368 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
25370 (define_expand, define_insn): Move the shift left from the
25371 define_insn to the define_expand for lxvl and stxvl instructions.
25372 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
25373 and XL_LEN_R definitions to PURE.
25375 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
25377 * config/i386/i386.c (indirect_thunk_name): Declare regno
25378 as unsigned int. Compare regno with INVALID_REGNUM.
25379 (output_indirect_thunk): Ditto.
25380 (output_indirect_thunk_function): Ditto.
25381 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
25382 in the call to output_indirect_thunk_function.
25384 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
25386 PR middle-end/83884
25387 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
25388 rather than the size of inner_type to determine the stack slot size
25389 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
25391 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
25394 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
25397 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
25399 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
25400 endian Linux systems to optionally enable multilibs for selecting
25401 the long double type if the user configured an explicit type.
25402 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
25403 have no long double multilibs if not defined.
25404 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
25405 warn if the user used -mabi={ieee,ibm}longdouble and we built
25406 multilibs for long double.
25407 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
25408 appropriate multilib option.
25409 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
25411 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
25412 for building long double multilibs.
25413 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
25415 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
25417 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
25420 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
25422 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
25425 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
25428 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
25431 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25433 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
25434 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
25436 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
25438 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
25439 different rtl trees depending on TARGET_64BIT.
25440 (rs6000_gen_lvx): Likewise.
25442 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
25444 * config/visium/visium.md (nop): Tweak comment.
25445 (hazard_nop): Likewise.
25447 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25449 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
25450 -mspeculate-indirect-jumps.
25451 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
25452 for -mno-speculate-indirect-jumps.
25453 (*call_indirect_elfv2<mode>_nospec): New define_insn.
25454 (*call_value_indirect_elfv2<mode>): Disable for
25455 -mno-speculate-indirect-jumps.
25456 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
25457 (indirect_jump): Emit different RTL for
25458 -mno-speculate-indirect-jumps.
25459 (*indirect_jump<mode>): Disable for
25460 -mno-speculate-indirect-jumps.
25461 (*indirect_jump<mode>_nospec): New define_insn.
25462 (tablejump): Emit different RTL for
25463 -mno-speculate-indirect-jumps.
25464 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
25465 (tablejumpsi_nospec): New define_expand.
25466 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
25467 (tablejumpdi_nospec): New define_expand.
25468 (*tablejump<mode>_internal1): Disable for
25469 -mno-speculate-indirect-jumps.
25470 (*tablejump<mode>_internal1_nospec): New define_insn.
25471 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
25474 2018-01-16 Artyom Skrobov tyomitch@gmail.com
25476 * caller-save.c (insert_save): Drop unnecessary parameter. All
25479 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25480 Richard Biener <rguenth@suse.de>
25483 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
25484 return early, inline manually is_gimple_sizepos. Make sure if we
25485 call gimplify_expr we don't end up with a gimple constant.
25486 * tree.c (variably_modified_type_p): Don't return true for
25487 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
25488 * gimplify.h (is_gimple_sizepos): Remove.
25490 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25492 PR tree-optimization/83857
25493 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
25494 vectorizable_live_operation for pure SLP statements.
25495 (vectorizable_live_operation): Handle PHIs.
25497 2018-01-16 Richard Biener <rguenther@suse.de>
25499 PR tree-optimization/83867
25500 * tree-vect-stmts.c (vect_transform_stmt): Precompute
25501 nested_in_vect_loop_p since the scalar stmt may get invalidated.
25503 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25506 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
25507 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
25508 If off is not INTEGER_CST, issue a may not be aligned warning
25509 rather than isn't aligned. Use isn%'t rather than isn't.
25510 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
25512 <case MULT_EXPR>: Improve the case when bottom and one of the
25513 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
25514 operand, in that case check if the other operand is multiple of
25515 bottom divided by the INTEGER_CST operand.
25517 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25520 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
25521 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
25522 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
25523 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
25524 * config/pa/pa.c (pa_function_arg_advance): Likewise.
25525 (pa_function_arg, pa_arg_partial_bytes): Likewise.
25526 (pa_function_arg_size): New function.
25528 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25530 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
25531 in a separate statement.
25533 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
25535 PR tree-optimization/83847
25536 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
25537 group gathers and scatters.
25539 2018-01-16 Jakub Jelinek <jakub@redhat.com>
25541 PR rtl-optimization/86620
25542 * params.def (max-sched-ready-insns): Bump minimum value to 1.
25544 PR rtl-optimization/83213
25545 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
25546 to last if both are JUMP_INSNs.
25548 PR tree-optimization/83843
25549 * gimple-ssa-store-merging.c
25550 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
25551 store_immediate_info for bswap/nop orig_stores.
25553 2018-01-15 Andrew Waterman <andrew@sifive.com>
25555 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
25557 <UDIV>: Increase cost if !TARGET_DIV.
25559 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
25561 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
25562 (define_attr "cr_logical_3op"): New.
25563 (cceq_ior_compare): Adjust.
25564 (cceq_ior_compare_complement): Adjust.
25565 (*cceq_rev_compare): Adjust.
25566 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
25567 (is_cracked_insn): Adjust.
25568 (insn_must_be_first_in_group): Adjust.
25569 * config/rs6000/40x.md: Adjust.
25570 * config/rs6000/440.md: Adjust.
25571 * config/rs6000/476.md: Adjust.
25572 * config/rs6000/601.md: Adjust.
25573 * config/rs6000/603.md: Adjust.
25574 * config/rs6000/6xx.md: Adjust.
25575 * config/rs6000/7450.md: Adjust.
25576 * config/rs6000/7xx.md: Adjust.
25577 * config/rs6000/8540.md: Adjust.
25578 * config/rs6000/cell.md: Adjust.
25579 * config/rs6000/e300c2c3.md: Adjust.
25580 * config/rs6000/e500mc.md: Adjust.
25581 * config/rs6000/e500mc64.md: Adjust.
25582 * config/rs6000/e5500.md: Adjust.
25583 * config/rs6000/e6500.md: Adjust.
25584 * config/rs6000/mpc.md: Adjust.
25585 * config/rs6000/power4.md: Adjust.
25586 * config/rs6000/power5.md: Adjust.
25587 * config/rs6000/power6.md: Adjust.
25588 * config/rs6000/power7.md: Adjust.
25589 * config/rs6000/power8.md: Adjust.
25590 * config/rs6000/power9.md: Adjust.
25591 * config/rs6000/rs64.md: Adjust.
25592 * config/rs6000/titan.md: Adjust.
25594 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25596 * config/i386/predicates.md (indirect_branch_operand): Rewrite
25597 ix86_indirect_branch_register logic.
25599 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25601 * config/i386/constraints.md (Bs): Update
25602 ix86_indirect_branch_register check. Don't check
25603 ix86_indirect_branch_register with GOT_memory_operand.
25605 * config/i386/predicates.md (GOT_memory_operand): Don't check
25606 ix86_indirect_branch_register here.
25607 (GOT32_symbol_operand): Likewise.
25609 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25611 * config/i386/predicates.md (constant_call_address_operand):
25612 Rewrite ix86_indirect_branch_register logic.
25613 (sibcall_insn_operand): Likewise.
25615 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25617 * config/i386/constraints.md (Bs): Replace
25618 ix86_indirect_branch_thunk_register with
25619 ix86_indirect_branch_register.
25621 * config/i386/i386.md (indirect_jump): Likewise.
25622 (tablejump): Likewise.
25623 (*sibcall_memory): Likewise.
25624 (*sibcall_value_memory): Likewise.
25625 Peepholes of indirect call and jump via memory: Likewise.
25626 * config/i386/i386.opt: Likewise.
25627 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25628 (GOT_memory_operand): Likewise.
25629 (call_insn_operand): Likewise.
25630 (sibcall_insn_operand): Likewise.
25631 (GOT32_symbol_operand): Likewise.
25633 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25635 PR middle-end/83837
25636 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
25637 type rather than type addr's type points to.
25638 (expand_omp_atomic_mutex): Likewise.
25639 (expand_omp_atomic): Likewise.
25641 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
25644 * config/i386/i386.c (output_indirect_thunk_function): Use
25645 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
25646 for __x86_return_thunk.
25648 2018-01-15 Richard Biener <rguenther@suse.de>
25650 PR middle-end/83850
25651 * expmed.c (extract_bit_field_1): Fix typo.
25653 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25656 * config/arm/iterators.md (VF): New mode iterator.
25657 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
25658 Remove integer-related logic from pattern.
25659 (neon_vabd<mode>_3): Likewise.
25661 2018-01-15 Jakub Jelinek <jakub@redhat.com>
25663 PR middle-end/82694
25664 * common.opt (fstrict-overflow): No longer an alias.
25665 (fwrapv-pointer): New option.
25666 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
25667 also for pointer types based on flag_wrapv_pointer.
25668 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
25669 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
25670 opts->x_flag_wrapv got set.
25671 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
25672 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
25673 POINTER_TYPE_OVERFLOW_UNDEFINED.
25674 * match.pd: Likewise in address comparison pattern.
25675 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
25677 2018-01-15 Richard Biener <rguenther@suse.de>
25680 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
25681 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
25682 Reset type names to their identifier if their TYPE_DECL doesn't
25683 have linkage (and thus is used for ODR and devirt).
25684 (save_debug_info_for_decl): Remove.
25685 (save_debug_info_for_type): Likewise.
25686 (add_tree_to_fld_list): Adjust.
25687 * tree-pretty-print.c (dump_generic_node): Make dumping of
25688 type names more robust.
25690 2018-01-15 Richard Biener <rguenther@suse.de>
25692 * BASE-VER: Bump to 8.0.1.
25694 2018-01-14 Martin Sebor <msebor@redhat.com>
25697 * builtins.c (check_access): Avoid warning when the no-warning bit
25700 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
25702 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
25703 * ira-color (allocno_hard_regs_compare): Likewise.
25705 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
25708 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
25709 Use .pushsection/.popsection.
25711 2018-01-14 Martin Sebor <msebor@redhat.com>
25714 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
25716 2018-01-14 Jakub Jelinek <jakub@redhat.com>
25718 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
25719 entry from extra_headers.
25720 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
25721 extra_headers, make the list bitwise identical to the i?86-*-* one.
25723 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25725 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
25726 -mcmodel=large with -mindirect-branch=thunk,
25727 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
25728 -mfunction-return=thunk-extern.
25729 * doc/invoke.texi: Document -mcmodel=large is incompatible with
25730 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
25731 -mfunction-return=thunk and -mfunction-return=thunk-extern.
25733 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25735 * config/i386/i386.c (print_reg): Print the name of the full
25736 integer register without '%'.
25737 (ix86_print_operand): Handle 'V'.
25738 * doc/extend.texi: Document 'V' modifier.
25740 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25742 * config/i386/constraints.md (Bs): Disallow memory operand for
25743 -mindirect-branch-register.
25745 * config/i386/predicates.md (indirect_branch_operand): Likewise.
25746 (GOT_memory_operand): Likewise.
25747 (call_insn_operand): Likewise.
25748 (sibcall_insn_operand): Likewise.
25749 (GOT32_symbol_operand): Likewise.
25750 * config/i386/i386.md (indirect_jump): Call convert_memory_address
25751 for -mindirect-branch-register.
25752 (tablejump): Likewise.
25753 (*sibcall_memory): Likewise.
25754 (*sibcall_value_memory): Likewise.
25755 Disallow peepholes of indirect call and jump via memory for
25756 -mindirect-branch-register.
25757 (*call_pop): Replace m with Bw.
25758 (*call_value_pop): Likewise.
25759 (*sibcall_pop_memory): Replace m with Bs.
25760 * config/i386/i386.opt (mindirect-branch-register): New option.
25761 * doc/invoke.texi: Document -mindirect-branch-register option.
25763 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25765 * config/i386/i386-protos.h (ix86_output_function_return): New.
25766 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
25767 set function_return_type.
25768 (indirect_thunk_name): Add ret_p to indicate thunk for function
25770 (output_indirect_thunk_function): Pass false to
25771 indirect_thunk_name.
25772 (ix86_output_indirect_branch_via_reg): Likewise.
25773 (ix86_output_indirect_branch_via_push): Likewise.
25774 (output_indirect_thunk_function): Create alias for function
25775 return thunk if regno < 0.
25776 (ix86_output_function_return): New function.
25777 (ix86_handle_fndecl_attribute): Handle function_return.
25778 (ix86_attribute_table): Add function_return.
25779 * config/i386/i386.h (machine_function): Add
25780 function_return_type.
25781 * config/i386/i386.md (simple_return_internal): Use
25782 ix86_output_function_return.
25783 (simple_return_internal_long): Likewise.
25784 * config/i386/i386.opt (mfunction-return=): New option.
25785 (indirect_branch): Mention -mfunction-return=.
25786 * doc/extend.texi: Document function_return function attribute.
25787 * doc/invoke.texi: Document -mfunction-return= option.
25789 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
25791 * config/i386/i386-opts.h (indirect_branch): New.
25792 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
25793 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
25794 with local indirect jump when converting indirect call and jump.
25795 (ix86_set_indirect_branch_type): New.
25796 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
25797 (indirectlabelno): New.
25798 (indirect_thunk_needed): Likewise.
25799 (indirect_thunk_bnd_needed): Likewise.
25800 (indirect_thunks_used): Likewise.
25801 (indirect_thunks_bnd_used): Likewise.
25802 (INDIRECT_LABEL): Likewise.
25803 (indirect_thunk_name): Likewise.
25804 (output_indirect_thunk): Likewise.
25805 (output_indirect_thunk_function): Likewise.
25806 (ix86_output_indirect_branch_via_reg): Likewise.
25807 (ix86_output_indirect_branch_via_push): Likewise.
25808 (ix86_output_indirect_branch): Likewise.
25809 (ix86_output_indirect_jmp): Likewise.
25810 (ix86_code_end): Call output_indirect_thunk_function if needed.
25811 (ix86_output_call_insn): Call ix86_output_indirect_branch if
25813 (ix86_handle_fndecl_attribute): Handle indirect_branch.
25814 (ix86_attribute_table): Add indirect_branch.
25815 * config/i386/i386.h (machine_function): Add indirect_branch_type
25816 and has_local_indirect_jump.
25817 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
25819 (tablejump): Likewise.
25820 (*indirect_jump): Use ix86_output_indirect_jmp.
25821 (*tablejump_1): Likewise.
25822 (simple_return_indirect_internal): Likewise.
25823 * config/i386/i386.opt (mindirect-branch=): New option.
25824 (indirect_branch): New.
25827 (thunk-inline): Likewise.
25828 (thunk-extern): Likewise.
25829 * doc/extend.texi: Document indirect_branch function attribute.
25830 * doc/invoke.texi: Document -mindirect-branch= option.
25832 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
25835 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
25837 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
25839 * ipa-inline.c (want_inline_small_function_p): Return false if
25840 inlining has already failed with CIF_FINAL_ERROR.
25841 (update_caller_keys): Call want_inline_small_function_p before
25843 (update_callee_keys): Likewise.
25845 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25847 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
25849 (rs6000_quadword_masked_address_p): Likewise.
25850 (quad_aligned_load_p): Likewise.
25851 (quad_aligned_store_p): Likewise.
25852 (const_load_sequence_p): Add comment to describe the outer-most loop.
25853 (mimic_memory_attributes_and_flags): New function.
25854 (rs6000_gen_stvx): Likewise.
25855 (replace_swapped_aligned_store): Likewise.
25856 (rs6000_gen_lvx): Likewise.
25857 (replace_swapped_aligned_load): Likewise.
25858 (replace_swapped_load_constant): Capitalize argument name in
25859 comment describing this function.
25860 (rs6000_analyze_swaps): Add a third pass to search for vector loads
25861 and stores that access quad-word aligned addresses and replace
25862 with stvx or lvx instructions when appropriate.
25863 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
25864 New function prototype.
25865 (rs6000_quadword_masked_address_p): Likewise.
25866 (rs6000_gen_lvx): Likewise.
25867 (rs6000_gen_stvx): Likewise.
25868 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
25869 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
25870 when memory address is aligned.
25871 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
25872 this split to select lvx instruction when memory address is aligned.
25873 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
25874 instruction when memory address is aligned.
25875 (*vsx_le_perm_load_v16qi): Likewise.
25876 (four unnamed splitters): Modify to select the stvx instruction
25877 when memory is aligned.
25879 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
25881 * predict.c (determine_unlikely_bbs): Handle correctly BBs
25882 which appears in the queue multiple times.
25884 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25885 Alan Hayward <alan.hayward@arm.com>
25886 David Sherwood <david.sherwood@arm.com>
25888 * tree-vectorizer.h (vec_lower_bound): New structure.
25889 (_loop_vec_info): Add check_nonzero and lower_bounds.
25890 (LOOP_VINFO_CHECK_NONZERO): New macro.
25891 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
25892 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
25893 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
25894 fields. Make seg_len the distance travelled, not including the
25896 (dr_direction_indicator): Declare.
25897 (dr_zero_step_indicator): Likewise.
25898 (dr_known_forward_stride_p): Likewise.
25899 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
25901 (runtime_alias_check_p): Allow runtime alias checks with
25903 (operator ==): Compare access_size and align.
25904 (prune_runtime_alias_test_list): Rework for new distinction between
25905 the access_size and seg_len.
25906 (create_intersect_range_checks_index): Likewise. Cope with polynomial
25908 (get_segment_min_max): New function.
25909 (create_intersect_range_checks): Use it.
25910 (dr_step_indicator): New function.
25911 (dr_direction_indicator): Likewise.
25912 (dr_zero_step_indicator): Likewise.
25913 (dr_known_forward_stride_p): Likewise.
25914 * tree-loop-distribution.c (data_ref_segment_size): Return
25915 DR_STEP * (niters - 1).
25916 (compute_alias_check_pairs): Update call to the dr_with_seg_len
25918 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
25919 (vect_preserves_scalar_order_p): New function, split out from...
25920 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
25921 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
25922 (vect_vfa_access_size): New function.
25923 (vect_vfa_align): Likewise.
25924 (vect_compile_time_alias): Take access_size_a and access_b arguments.
25925 (dump_lower_bound): New function.
25926 (vect_check_lower_bound): Likewise.
25927 (vect_small_gap_p): Likewise.
25928 (vectorizable_with_step_bound_p): Likewise.
25929 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
25930 depencies if the vectorization factor is 1. Convert the checks
25931 for nonzero steps into checks on the bounds of DR_STEP. Try using
25932 a bunds check for variable steps if the minimum required step is
25933 relatively small. Update calls to the dr_with_seg_len
25934 constructor and to vect_compile_time_alias.
25935 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
25937 (vect_loop_versioning): Call it.
25938 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
25940 (vect_estimate_min_profitable_iters): Account for any bounds checks.
25942 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25943 Alan Hayward <alan.hayward@arm.com>
25944 David Sherwood <david.sherwood@arm.com>
25946 * doc/sourcebuild.texi (vect_scatter_store): Document.
25947 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
25949 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
25951 * genopinit.c (main): Add supports_vec_scatter_store and
25952 supports_vec_scatter_store_cached to target_optabs.
25953 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
25954 IFN_MASK_SCATTER_STORE.
25955 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
25957 * internal-fn.h (internal_store_fn_p): Declare.
25958 (internal_fn_stored_value_index): Likewise.
25959 * internal-fn.c (scatter_store_direct): New macro.
25960 (expand_scatter_store_optab_fn): New function.
25961 (direct_scatter_store_optab_supported_p): New macro.
25962 (internal_store_fn_p): New function.
25963 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
25964 IFN_MASK_SCATTER_STORE.
25965 (internal_fn_mask_index): Likewise.
25966 (internal_fn_stored_value_index): New function.
25967 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
25968 for scatter stores.
25969 * optabs-query.h (supports_vec_scatter_store_p): Declare.
25970 * optabs-query.c (supports_vec_scatter_store_p): New function.
25971 * tree-vectorizer.h (vect_get_store_rhs): Declare.
25972 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
25973 true for scatter stores.
25974 (vect_gather_scatter_fn_p): Handle scatter stores too.
25975 (vect_check_gather_scatter): Consider using scatter stores if
25976 supports_vec_scatter_store_p.
25977 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
25978 scatter stores too.
25979 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
25980 internal_fn_stored_value_index.
25981 (check_load_store_masking): Handle scatter stores too.
25982 (vect_get_store_rhs): Make public.
25983 (vectorizable_call): Use internal_store_fn_p.
25984 (vectorizable_store): Handle scatter store internal functions.
25985 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
25986 when deciding whether the end of the group has been reached.
25987 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
25988 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
25989 (mask_scatter_store<mode>): New insns.
25991 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
25992 Alan Hayward <alan.hayward@arm.com>
25993 David Sherwood <david.sherwood@arm.com>
25995 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
25996 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
25997 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
25999 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
26000 Use vect_truncate_gather_scatter_offset if we can't treat the
26001 operation as a normal gather load or scatter store.
26002 (get_group_load_store_type): Take the gather_scatter_info
26003 as argument. Try using a gather load or scatter store for
26004 single-element groups.
26005 (get_load_store_type): Update calls to get_group_load_store_type
26006 and vect_use_strided_gather_scatters_p.
26008 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26009 Alan Hayward <alan.hayward@arm.com>
26010 David Sherwood <david.sherwood@arm.com>
26012 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
26013 optional tree argument.
26014 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
26016 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
26017 but continue to use the current value as a fallback.
26018 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
26019 to compare the updates.
26020 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
26021 (get_load_store_type): Use it when handling a strided access.
26022 (vect_get_strided_load_store_ops): New function.
26023 (vect_get_data_ptr_increment): Likewise.
26024 (vectorizable_load): Handle strided gather loads. Always pass
26025 a step to vect_create_data_ref_ptr and bump_vector_ptr.
26027 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26028 Alan Hayward <alan.hayward@arm.com>
26029 David Sherwood <david.sherwood@arm.com>
26031 * doc/md.texi (gather_load@var{m}): Document.
26032 (mask_gather_load@var{m}): Likewise.
26033 * genopinit.c (main): Add supports_vec_gather_load and
26034 supports_vec_gather_load_cached to target_optabs.
26035 * optabs-tree.c (init_tree_optimization_optabs): Use
26036 ggc_cleared_alloc to allocate target_optabs.
26037 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
26038 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
26040 * internal-fn.h (internal_load_fn_p): Declare.
26041 (internal_gather_scatter_fn_p): Likewise.
26042 (internal_fn_mask_index): Likewise.
26043 (internal_gather_scatter_fn_supported_p): Likewise.
26044 * internal-fn.c (gather_load_direct): New macro.
26045 (expand_gather_load_optab_fn): New function.
26046 (direct_gather_load_optab_supported_p): New macro.
26047 (direct_internal_fn_optab): New function.
26048 (internal_load_fn_p): Likewise.
26049 (internal_gather_scatter_fn_p): Likewise.
26050 (internal_fn_mask_index): Likewise.
26051 (internal_gather_scatter_fn_supported_p): Likewise.
26052 * optabs-query.c (supports_at_least_one_mode_p): New function.
26053 (supports_vec_gather_load_p): Likewise.
26054 * optabs-query.h (supports_vec_gather_load_p): Declare.
26055 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
26056 and memory_type field.
26057 (NUM_PATTERNS): Bump to 15.
26058 * tree-vect-data-refs.c: Include internal-fn.h.
26059 (vect_gather_scatter_fn_p): New function.
26060 (vect_describe_gather_scatter_call): Likewise.
26061 (vect_check_gather_scatter): Try using internal functions for
26062 gather loads. Recognize existing calls to a gather load function.
26063 (vect_analyze_data_refs): Consider using gather loads if
26064 supports_vec_gather_load_p.
26065 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
26066 (vect_get_gather_scatter_offset_type): Likewise.
26067 (vect_convert_mask_for_vectype): Likewise.
26068 (vect_add_conversion_to_patterm): Likewise.
26069 (vect_try_gather_scatter_pattern): Likewise.
26070 (vect_recog_gather_scatter_pattern): New pattern recognizer.
26071 (vect_vect_recog_func_ptrs): Add it.
26072 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
26073 internal_fn_mask_index and internal_gather_scatter_fn_p.
26074 (check_load_store_masking): Take the gather_scatter_info as an
26075 argument and handle gather loads.
26076 (vect_get_gather_scatter_ops): New function.
26077 (vectorizable_call): Check internal_load_fn_p.
26078 (vectorizable_load): Likewise. Handle gather load internal
26080 (vectorizable_store): Update call to check_load_store_masking.
26081 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
26082 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
26083 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
26084 (aarch64_gather_scale_operand_d): New predicates.
26085 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
26086 (mask_gather_load<mode>): New insns.
26088 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26089 Alan Hayward <alan.hayward@arm.com>
26090 David Sherwood <david.sherwood@arm.com>
26092 * optabs.def (fold_left_plus_optab): New optab.
26093 * doc/md.texi (fold_left_plus_@var{m}): Document.
26094 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
26095 * internal-fn.c (fold_left_direct): Define.
26096 (expand_fold_left_optab_fn): Likewise.
26097 (direct_fold_left_optab_supported_p): Likewise.
26098 * fold-const-call.c (fold_const_fold_left): New function.
26099 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
26100 * tree-parloops.c (valid_reduction_p): New function.
26101 (gather_scalar_reductions): Use it.
26102 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
26103 (vect_finish_replace_stmt): Declare.
26104 * tree-vect-loop.c (fold_left_reduction_fn): New function.
26105 (needs_fold_left_reduction_p): New function, split out from...
26106 (vect_is_simple_reduction): ...here. Accept reductions that
26107 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
26108 (vect_force_simple_reduction): Also store the reduction type in
26109 the assignment's STMT_VINFO_REDUC_TYPE.
26110 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
26111 (merge_with_identity): New function.
26112 (vect_expand_fold_left): Likewise.
26113 (vectorize_fold_left_reduction): Likewise.
26114 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
26115 scalar phi in place for it. Check for target support and reject
26116 cases that would reassociate the operation. Defer the transform
26117 phase to vectorize_fold_left_reduction.
26118 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
26119 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
26120 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
26122 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26124 * tree-if-conv.c (predicate_mem_writes): Remove redundant
26125 call to ifc_temp_var.
26127 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26128 Alan Hayward <alan.hayward@arm.com>
26129 David Sherwood <david.sherwood@arm.com>
26131 * target.def (legitimize_address_displacement): Take the original
26132 offset as a poly_int.
26133 * targhooks.h (default_legitimize_address_displacement): Update
26135 * targhooks.c (default_legitimize_address_displacement): Likewise.
26136 * doc/tm.texi: Regenerate.
26137 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
26138 as an argument, moving assert of ad->disp == ad->disp_term to...
26139 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
26140 Try calling targetm.legitimize_address_displacement before expanding
26141 the address rather than afterwards, and adjust for the new interface.
26142 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
26143 Match the new hook interface. Handle SVE addresses.
26144 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
26145 new hook interface.
26147 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26149 * Makefile.in (OBJS): Add early-remat.o.
26150 * target.def (select_early_remat_modes): New hook.
26151 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
26152 * doc/tm.texi: Regenerate.
26153 * targhooks.h (default_select_early_remat_modes): Declare.
26154 * targhooks.c (default_select_early_remat_modes): New function.
26155 * timevar.def (TV_EARLY_REMAT): New timevar.
26156 * passes.def (pass_early_remat): New pass.
26157 * tree-pass.h (make_pass_early_remat): Declare.
26158 * early-remat.c: New file.
26159 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
26161 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
26163 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26164 Alan Hayward <alan.hayward@arm.com>
26165 David Sherwood <david.sherwood@arm.com>
26167 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
26168 vfm1 with a bound_epilog parameter.
26169 (vect_do_peeling): Update calls accordingly, and move the prologue
26170 call earlier in the function. Treat the base bound_epilog as 0 for
26171 fully-masked loops and retain vf - 1 for other loops. Add 1 to
26172 this base when peeling for gaps.
26173 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
26174 with fully-masked loops.
26175 (vect_estimate_min_profitable_iters): Handle the single peeled
26176 iteration in that case.
26178 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26179 Alan Hayward <alan.hayward@arm.com>
26180 David Sherwood <david.sherwood@arm.com>
26182 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
26183 single-element interleaving even if the size is not a power of 2.
26184 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
26185 accesses for single-element interleaving if the group size is
26188 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26189 Alan Hayward <alan.hayward@arm.com>
26190 David Sherwood <david.sherwood@arm.com>
26192 * doc/md.texi (fold_extract_last_@var{m}): Document.
26193 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
26194 * optabs.def (fold_extract_last_optab): New optab.
26195 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
26196 * internal-fn.c (fold_extract_direct): New macro.
26197 (expand_fold_extract_optab_fn): Likewise.
26198 (direct_fold_extract_optab_supported_p): Likewise.
26199 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
26200 * tree-vect-loop.c (vect_model_reduction_cost): Handle
26201 EXTRACT_LAST_REDUCTION.
26202 (get_initial_def_for_reduction): Do not create an initial vector
26203 for EXTRACT_LAST_REDUCTION reductions.
26204 (vectorizable_reduction): Leave the scalar phi in place for
26205 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
26206 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
26207 epilogue code for EXTRACT_LAST_REDUCTION and defer the
26208 transform phase to vectorizable_condition.
26209 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
26211 (vect_finish_stmt_generation): ...here.
26212 (vect_finish_replace_stmt): New function.
26213 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
26214 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
26216 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
26218 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26219 Alan Hayward <alan.hayward@arm.com>
26220 David Sherwood <david.sherwood@arm.com>
26222 * doc/md.texi (extract_last_@var{m}): Document.
26223 * optabs.def (extract_last_optab): New optab.
26224 * internal-fn.def (EXTRACT_LAST): New internal function.
26225 * internal-fn.c (cond_unary_direct): New macro.
26226 (expand_cond_unary_optab_fn): Likewise.
26227 (direct_cond_unary_optab_supported_p): Likewise.
26228 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
26229 loops using EXTRACT_LAST.
26230 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
26231 (extract_last_<mode>): ...this optab.
26232 (vec_extract<mode><Vel>): Update accordingly.
26234 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26235 Alan Hayward <alan.hayward@arm.com>
26236 David Sherwood <david.sherwood@arm.com>
26238 * target.def (empty_mask_is_expensive): New hook.
26239 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
26240 * doc/tm.texi: Regenerate.
26241 * targhooks.h (default_empty_mask_is_expensive): Declare.
26242 * targhooks.c (default_empty_mask_is_expensive): New function.
26243 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
26244 if the target says that empty masks are expensive.
26245 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
26247 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
26249 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26250 Alan Hayward <alan.hayward@arm.com>
26251 David Sherwood <david.sherwood@arm.com>
26253 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
26254 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
26255 (vect_use_loop_mask_for_alignment_p): New function.
26256 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
26257 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
26258 niters_skip argument. Make sure that the first niters_skip elements
26259 of the first iteration are inactive.
26260 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
26261 Update call to vect_set_loop_masks_directly.
26262 (get_misalign_in_elems): New function, split out from...
26263 (vect_gen_prolog_loop_niters): ...here.
26264 (vect_update_init_of_dr): Take a code argument that specifies whether
26265 the adjustment should be added or subtracted.
26266 (vect_update_init_of_drs): Likewise.
26267 (vect_prepare_for_masked_peels): New function.
26268 (vect_do_peeling): Skip prologue peeling if we're using a mask
26269 instead. Update call to vect_update_inits_of_drs.
26270 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26272 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
26273 alignment. Do not include the number of peeled iterations in
26274 the minimum threshold in that case.
26275 (vectorizable_induction): Adjust the start value down by
26276 LOOP_VINFO_MASK_SKIP_NITERS iterations.
26277 (vect_transform_loop): Call vect_prepare_for_masked_peels.
26278 Take the number of skipped iterations into account when calculating
26280 * tree-vect-stmts.c (vect_gen_while_not): New function.
26282 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26283 Alan Hayward <alan.hayward@arm.com>
26284 David Sherwood <david.sherwood@arm.com>
26286 * doc/sourcebuild.texi (vect_fully_masked): Document.
26287 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
26288 default value to 0.
26289 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
26291 (vect_analyze_loop_2): ...here. Don't check the vectorization
26292 factor against the number of loop iterations if the loop is
26295 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26296 Alan Hayward <alan.hayward@arm.com>
26297 David Sherwood <david.sherwood@arm.com>
26299 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
26300 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
26301 (dump_groups): Update accordingly.
26302 (iv_use::mem_type): New member variable.
26303 (address_p): New function.
26304 (record_use): Add a mem_type argument and initialize the new
26306 (record_group_use): Add a mem_type argument. Use address_p.
26307 Remove obsolete null checks of base_object. Update call to record_use.
26308 (find_interesting_uses_op): Update call to record_group_use.
26309 (find_interesting_uses_cond): Likewise.
26310 (find_interesting_uses_address): Likewise.
26311 (get_mem_type_for_internal_fn): New function.
26312 (find_address_like_use): Likewise.
26313 (find_interesting_uses_stmt): Try find_address_like_use before
26314 calling find_interesting_uses_op.
26315 (addr_offset_valid_p): Use the iv mem_type field as the type
26316 of the addressed memory.
26317 (add_autoinc_candidates): Likewise.
26318 (get_address_cost): Likewise.
26319 (split_small_address_groups_p): Use address_p.
26320 (split_address_groups): Likewise.
26321 (add_iv_candidate_for_use): Likewise.
26322 (autoinc_possible_for_pair): Likewise.
26323 (rewrite_groups): Likewise.
26324 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
26325 (determine_group_iv_cost): Update after split of USE_ADDRESS.
26326 (get_alias_ptr_type_for_ptr_address): New function.
26327 (rewrite_use_address): Rewrite address uses in calls that were
26328 identified by find_address_like_use.
26330 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26331 Alan Hayward <alan.hayward@arm.com>
26332 David Sherwood <david.sherwood@arm.com>
26334 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
26336 * gimple-expr.h (is_gimple_addressable: Likewise.
26337 * gimple-expr.c (is_gimple_address): Likewise.
26338 * internal-fn.c (expand_call_mem_ref): New function.
26339 (expand_mask_load_optab_fn): Use it.
26340 (expand_mask_store_optab_fn): Likewise.
26342 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26343 Alan Hayward <alan.hayward@arm.com>
26344 David Sherwood <david.sherwood@arm.com>
26346 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
26347 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
26348 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
26349 (cond_umax@var{mode}): Document.
26350 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
26351 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
26352 (cond_umin_optab, cond_umax_optab): New optabs.
26353 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
26354 (COND_IOR, COND_XOR): New internal functions.
26355 * internal-fn.h (get_conditional_internal_fn): Declare.
26356 * internal-fn.c (cond_binary_direct): New macro.
26357 (expand_cond_binary_optab_fn): Likewise.
26358 (direct_cond_binary_optab_supported_p): Likewise.
26359 (get_conditional_internal_fn): New function.
26360 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
26361 Cope with reduction statements that are vectorized as calls rather
26363 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
26364 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
26365 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
26366 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
26367 (UNSPEC_COND_EOR): New unspecs.
26368 (optab): Add mappings for them.
26369 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
26370 (sve_int_op, sve_fp_op): New int attributes.
26372 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26373 Alan Hayward <alan.hayward@arm.com>
26374 David Sherwood <david.sherwood@arm.com>
26376 * optabs.def (while_ult_optab): New optab.
26377 * doc/md.texi (while_ult@var{m}@var{n}): Document.
26378 * internal-fn.def (WHILE_ULT): New internal function.
26379 * internal-fn.h (direct_internal_fn_supported_p): New override
26380 that takes two types as argument.
26381 * internal-fn.c (while_direct): New macro.
26382 (expand_while_optab_fn): New function.
26383 (convert_optab_supported_p): Likewise.
26384 (direct_while_optab_supported_p): New macro.
26385 * wide-int.h (wi::udiv_ceil): New function.
26386 * tree-vectorizer.h (rgroup_masks): New structure.
26387 (vec_loop_masks): New typedef.
26388 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
26389 and fully_masked_p.
26390 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
26391 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
26392 (vect_max_vf): New function.
26393 (slpeel_make_loop_iterate_ntimes): Delete.
26394 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
26395 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
26396 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
26397 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
26398 internal-fn.h, stor-layout.h and optabs-query.h.
26399 (vect_set_loop_mask): New function.
26400 (add_preheader_seq): Likewise.
26401 (add_header_seq): Likewise.
26402 (interleave_supported_p): Likewise.
26403 (vect_maybe_permute_loop_masks): Likewise.
26404 (vect_set_loop_masks_directly): Likewise.
26405 (vect_set_loop_condition_masked): Likewise.
26406 (vect_set_loop_condition_unmasked): New function, split out from
26407 slpeel_make_loop_iterate_ntimes.
26408 (slpeel_make_loop_iterate_ntimes): Rename to..
26409 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
26410 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
26411 (vect_do_peeling): Update call accordingly.
26412 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
26414 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
26415 mask_compare_type, can_fully_mask_p and fully_masked_p.
26416 (release_vec_loop_masks): New function.
26417 (_loop_vec_info): Use it to free the loop masks.
26418 (can_produce_all_loop_masks_p): New function.
26419 (vect_get_max_nscalars_per_iter): Likewise.
26420 (vect_verify_full_masking): Likewise.
26421 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
26422 retries, and free the mask rgroups before retrying. Check loop-wide
26423 reasons for disallowing fully-masked loops. Make the final decision
26424 about whether use a fully-masked loop or not.
26425 (vect_estimate_min_profitable_iters): Do not assume that peeling
26426 for the number of iterations will be needed for fully-masked loops.
26427 (vectorizable_reduction): Disable fully-masked loops.
26428 (vectorizable_live_operation): Likewise.
26429 (vect_halve_mask_nunits): New function.
26430 (vect_double_mask_nunits): Likewise.
26431 (vect_record_loop_mask): Likewise.
26432 (vect_get_loop_mask): Likewise.
26433 (vect_transform_loop): Handle the case in which the final loop
26434 iteration might handle a partial vector. Call vect_set_loop_condition
26435 instead of slpeel_make_loop_iterate_ntimes.
26436 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
26437 (check_load_store_masking): New function.
26438 (prepare_load_store_mask): Likewise.
26439 (vectorizable_store): Handle fully-masked loops.
26440 (vectorizable_load): Likewise.
26441 (supportable_widening_operation): Use vect_halve_mask_nunits for
26443 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
26444 (vect_gen_while): New function.
26445 * config/aarch64/aarch64.md (umax<mode>3): New expander.
26446 (aarch64_uqdec<mode>): New insn.
26448 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26449 Alan Hayward <alan.hayward@arm.com>
26450 David Sherwood <david.sherwood@arm.com>
26452 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
26453 (reduc_xor_scal_optab): New optabs.
26454 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
26455 (reduc_xor_scal_@var{m}): Document.
26456 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
26457 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
26458 internal functions.
26459 * fold-const-call.c (fold_const_call): Handle them.
26460 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
26461 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
26462 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
26463 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
26464 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
26465 (UNSPEC_XORV): New unspecs.
26466 (optab): Add entries for them.
26467 (BITWISEV): New int iterator.
26468 (bit_reduc_op): New int attributes.
26470 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26471 Alan Hayward <alan.hayward@arm.com>
26472 David Sherwood <david.sherwood@arm.com>
26474 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
26475 * internal-fn.def (VEC_SHL_INSERT): New internal function.
26476 * optabs.def (vec_shl_insert_optab): New optab.
26477 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
26478 (duplicate_and_interleave): Likewise.
26479 * tree-vect-loop.c: Include internal-fn.h.
26480 (neutral_op_for_slp_reduction): New function, split out from
26481 get_initial_defs_for_reduction.
26482 (get_initial_def_for_reduction): Handle option 2 for variable-length
26483 vectors by loading the neutral value into a vector and then shifting
26484 the initial value into element 0.
26485 (get_initial_defs_for_reduction): Replace the code argument with
26486 the neutral value calculated by neutral_op_for_slp_reduction.
26487 Use gimple_build_vector for constant-length vectors.
26488 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
26489 but the first group_size elements have a neutral value.
26490 Use duplicate_and_interleave otherwise.
26491 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
26492 Update call to get_initial_defs_for_reduction. Handle SLP
26493 reductions for variable-length vectors by creating one vector
26494 result for each scalar result, with the elements associated
26495 with other scalar results stubbed out with the neutral value.
26496 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
26497 Require IFN_VEC_SHL_INSERT for double reductions on
26498 variable-length vectors, or SLP reductions that have
26499 a neutral value. Require can_duplicate_and_interleave_p
26500 support for variable-length unchained SLP reductions if there
26501 is no neutral value, such as for MIN/MAX reductions. Also require
26502 the number of vector elements to be a multiple of the number of
26503 SLP statements when doing variable-length unchained SLP reductions.
26504 Update call to vect_create_epilog_for_reduction.
26505 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
26506 and remove initial values.
26507 (duplicate_and_interleave): Make public.
26508 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
26509 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
26511 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26512 Alan Hayward <alan.hayward@arm.com>
26513 David Sherwood <david.sherwood@arm.com>
26515 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
26516 (can_duplicate_and_interleave_p): New function.
26517 (vect_get_and_check_slp_defs): Take the vector of statements
26518 rather than just the current one. Remove excess parentheses.
26519 Restriction rejectinon of vect_constant_def and vect_external_def
26520 for variable-length vectors to boolean types, or types for which
26521 can_duplicate_and_interleave_p is false.
26522 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
26523 (duplicate_and_interleave): New function.
26524 (vect_get_constant_vectors): Use gimple_build_vector for
26525 constant-length vectors and suitable variable-length constant
26526 vectors. Use duplicate_and_interleave for other variable-length
26527 vectors. Don't defer the update when inserting new statements.
26529 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26530 Alan Hayward <alan.hayward@arm.com>
26531 David Sherwood <david.sherwood@arm.com>
26533 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
26534 min_profitable_iters doesn't go negative.
26536 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26537 Alan Hayward <alan.hayward@arm.com>
26538 David Sherwood <david.sherwood@arm.com>
26540 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
26541 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
26542 * optabs.def (vec_mask_load_lanes_optab): New optab.
26543 (vec_mask_store_lanes_optab): Likewise.
26544 * internal-fn.def (MASK_LOAD_LANES): New internal function.
26545 (MASK_STORE_LANES): Likewise.
26546 * internal-fn.c (mask_load_lanes_direct): New macro.
26547 (mask_store_lanes_direct): Likewise.
26548 (expand_mask_load_optab_fn): Handle masked operations.
26549 (expand_mask_load_lanes_optab_fn): New macro.
26550 (expand_mask_store_optab_fn): Handle masked operations.
26551 (expand_mask_store_lanes_optab_fn): New macro.
26552 (direct_mask_load_lanes_optab_supported_p): Likewise.
26553 (direct_mask_store_lanes_optab_supported_p): Likewise.
26554 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
26556 (vect_load_lanes_supported): Likewise.
26557 * tree-vect-data-refs.c (strip_conversion): New function.
26558 (can_group_stmts_p): Likewise.
26559 (vect_analyze_data_ref_accesses): Use it instead of checking
26560 for a pair of assignments.
26561 (vect_store_lanes_supported): Take a masked_p parameter.
26562 (vect_load_lanes_supported): Likewise.
26563 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
26564 vect_store_lanes_supported and vect_load_lanes_supported.
26565 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
26566 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
26567 parameter. Don't allow gaps for masked accesses.
26568 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
26569 and vect_load_lanes_supported.
26570 (get_load_store_type): Take a masked_p parameter and update
26571 call to get_group_load_store_type.
26572 (vectorizable_store): Update call to get_load_store_type.
26573 Handle IFN_MASK_STORE_LANES.
26574 (vectorizable_load): Update call to get_load_store_type.
26575 Handle IFN_MASK_LOAD_LANES.
26577 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26578 Alan Hayward <alan.hayward@arm.com>
26579 David Sherwood <david.sherwood@arm.com>
26581 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
26583 * config/aarch64/aarch64-protos.h
26584 (aarch64_sve_struct_memory_operand_p): Declare.
26585 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
26586 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
26587 (VPRED, vpred): Handle SVE structure modes.
26588 * config/aarch64/constraints.md (Utx): New constraint.
26589 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
26590 (aarch64_sve_struct_nonimmediate_operand): New predicates.
26591 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
26592 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
26593 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
26594 structure modes. Split into pieces after RA.
26595 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
26596 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
26598 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
26599 SVE structure modes.
26600 (aarch64_classify_address): Likewise.
26601 (sizetochar): Move earlier in file.
26602 (aarch64_print_operand): Handle SVE register lists.
26603 (aarch64_array_mode): New function.
26604 (aarch64_sve_struct_memory_operand_p): Likewise.
26605 (TARGET_ARRAY_MODE): Redefine.
26607 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26608 Alan Hayward <alan.hayward@arm.com>
26609 David Sherwood <david.sherwood@arm.com>
26611 * target.def (array_mode): New target hook.
26612 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
26613 * doc/tm.texi: Regenerate.
26614 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
26615 * hooks.c (hook_optmode_mode_uhwi_none): New function.
26616 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
26617 targetm.array_mode.
26618 * stor-layout.c (mode_for_array): Likewise. Support polynomial
26621 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26622 Alan Hayward <alan.hayward@arm.com>
26623 David Sherwood <david.sherwood@arm.com>
26625 * fold-const.c (fold_binary_loc): Check the argument types
26626 rather than the result type when testing for a vector operation.
26628 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26630 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
26631 * doc/tm.texi: Regenerate.
26633 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26634 Alan Hayward <alan.hayward@arm.com>
26635 David Sherwood <david.sherwood@arm.com>
26637 * doc/invoke.texi (-msve-vector-bits=): Document new option.
26638 (sve): Document new AArch64 extension.
26639 * doc/md.texi (w): Extend the description of the AArch64
26640 constraint to include SVE vectors.
26641 (Upl, Upa): Document new AArch64 predicate constraints.
26642 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
26644 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
26645 (msve-vector-bits=): New option.
26646 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
26647 SVE when these are disabled.
26648 (sve): New extension.
26649 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
26650 modes. Adjust their number of units based on aarch64_sve_vg.
26651 (MAX_BITSIZE_MODE_ANY_MODE): Define.
26652 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
26653 aarch64_addr_query_type.
26654 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
26655 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
26656 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
26657 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
26658 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
26659 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
26660 (aarch64_simd_imm_zero_p): Delete.
26661 (aarch64_check_zero_based_sve_index_immediate): Declare.
26662 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26663 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26664 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26665 (aarch64_sve_float_mul_immediate_p): Likewise.
26666 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26667 rather than an rtx.
26668 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
26669 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
26670 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
26671 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
26672 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
26673 (aarch64_regmode_natural_size): Likewise.
26674 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
26675 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
26677 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
26678 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
26679 for VG and the SVE predicate registers.
26680 (V_ALIASES): Add a "z"-prefixed alias.
26681 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
26682 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
26683 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
26684 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
26685 (REG_CLASS_NAMES): Add entries for them.
26686 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
26687 and the predicate registers.
26688 (aarch64_sve_vg): Declare.
26689 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
26690 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
26691 (REGMODE_NATURAL_SIZE): Define.
26692 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
26694 * config/aarch64/aarch64.c: Include cfgrtl.h.
26695 (simd_immediate_info): Add a constructor for series vectors,
26696 and an associated step field.
26697 (aarch64_sve_vg): New variable.
26698 (aarch64_dbx_register_number): Handle VG and the predicate registers.
26699 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
26700 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
26701 (VEC_ANY_DATA, VEC_STRUCT): New constants.
26702 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
26703 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
26704 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
26705 (aarch64_get_mask_mode): New functions.
26706 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
26707 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
26708 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
26709 predicate modes and predicate registers. Explicitly restrict
26710 GPRs to modes of 16 bytes or smaller. Only allow FP registers
26711 to store a vector mode if it is recognized by
26712 aarch64_classify_vector_mode.
26713 (aarch64_regmode_natural_size): New function.
26714 (aarch64_hard_regno_caller_save_mode): Return the original mode
26716 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
26717 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
26718 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
26719 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
26721 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
26722 does not overlap dest if the function is frame-related. Handle
26724 (aarch64_split_add_offset): New function.
26725 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
26726 them aarch64_add_offset.
26727 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
26728 and update call to aarch64_sub_sp.
26729 (aarch64_add_cfa_expression): New function.
26730 (aarch64_expand_prologue): Pass extra temporary registers to the
26731 functions above. Handle the case in which we need to emit new
26732 DW_CFA_expressions for registers that were originally saved
26733 relative to the stack pointer, but now have to be expressed
26734 relative to the frame pointer.
26735 (aarch64_output_mi_thunk): Pass extra temporary registers to the
26737 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
26738 IP0 and IP1 values for SVE frames.
26739 (aarch64_expand_vec_series): New function.
26740 (aarch64_expand_sve_widened_duplicate): Likewise.
26741 (aarch64_expand_sve_const_vector): Likewise.
26742 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
26743 Handle SVE constants. Use emit_move_insn to move a force_const_mem
26744 into the register, rather than emitting a SET directly.
26745 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
26746 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
26747 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
26748 (offset_9bit_signed_scaled_p): New functions.
26749 (aarch64_replicate_bitmask_imm): New function.
26750 (aarch64_bitmask_imm): Use it.
26751 (aarch64_cannot_force_const_mem): Reject expressions involving
26752 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
26753 (aarch64_classify_index): Handle SVE indices, by requiring
26754 a plain register index with a scale that matches the element size.
26755 (aarch64_classify_address): Handle SVE addresses. Assert that
26756 the mode of the address is VOIDmode or an integer mode.
26757 Update call to aarch64_classify_symbol.
26758 (aarch64_classify_symbolic_expression): Update call to
26759 aarch64_classify_symbol.
26760 (aarch64_const_vec_all_in_range_p): New function.
26761 (aarch64_print_vector_float_operand): Likewise.
26762 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
26763 "vN" for FP registers with SVE modes. Handle (const ...) vectors
26764 and the FP immediates 1.0 and 0.5.
26765 (aarch64_print_address_internal): Handle SVE addresses.
26766 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
26767 (aarch64_regno_regclass): Handle predicate registers.
26768 (aarch64_secondary_reload): Handle big-endian reloads of SVE
26770 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
26771 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
26772 (aarch64_convert_sve_vector_bits): New function.
26773 (aarch64_override_options): Use it to handle -msve-vector-bits=.
26774 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
26775 rather than an rtx.
26776 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
26777 Handle SVE vector and predicate modes. Accept VL-based constants
26778 that need only one temporary register, and VL offsets that require
26779 no temporary registers.
26780 (aarch64_conditional_register_usage): Mark the predicate registers
26781 as fixed if SVE isn't available.
26782 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
26783 Return true for SVE vector and predicate modes.
26784 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
26785 rather than an unsigned int. Handle SVE modes.
26786 (aarch64_preferred_simd_mode): Update call accordingly. Handle
26788 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
26790 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
26791 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
26792 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
26793 (aarch64_sve_float_mul_immediate_p): New functions.
26794 (aarch64_sve_valid_immediate): New function.
26795 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
26796 Explicitly reject structure modes. Check for INDEX constants.
26797 Handle PTRUE and PFALSE constants.
26798 (aarch64_check_zero_based_sve_index_immediate): New function.
26799 (aarch64_simd_imm_zero_p): Delete.
26800 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
26801 vector modes. Accept constants in the range of CNT[BHWD].
26802 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
26803 ask for an Advanced SIMD mode.
26804 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
26805 (aarch64_simd_vector_alignment): Handle SVE predicates.
26806 (aarch64_vectorize_preferred_vector_alignment): New function.
26807 (aarch64_simd_vector_alignment_reachable): Use it instead of
26809 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
26810 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
26812 (MAX_VECT_LEN): Delete.
26813 (expand_vec_perm_d): Add a vec_flags field.
26814 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
26815 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
26816 (aarch64_evpc_ext): Don't apply a big-endian lane correction
26818 (aarch64_evpc_rev): Rename to...
26819 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
26820 (aarch64_evpc_rev_global): New function.
26821 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
26822 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
26824 (aarch64_evpc_sve_tbl): New function.
26825 (aarch64_expand_vec_perm_const_1): Update after rename of
26826 aarch64_evpc_rev. Handle SVE permutes too, trying
26827 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
26828 than aarch64_evpc_tbl.
26829 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
26830 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
26831 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
26832 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
26833 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
26834 (aarch64_expand_sve_vcond): New functions.
26835 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
26836 of aarch64_vector_mode_p.
26837 (aarch64_dwarf_poly_indeterminate_value): New function.
26838 (aarch64_compute_pressure_classes): Likewise.
26839 (aarch64_can_change_mode_class): Likewise.
26840 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
26841 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
26842 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
26843 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
26844 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
26845 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
26846 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
26847 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
26849 (Dn, Dl, Dr): Accept const as well as const_vector.
26850 (Dz): Likewise. Compare against CONST0_RTX.
26851 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
26852 of "vector" where appropriate.
26853 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
26854 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
26855 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
26856 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
26857 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
26858 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
26859 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
26860 (v_int_equiv): Extend to SVE modes.
26861 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
26863 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
26864 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
26865 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
26866 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
26867 (SVE_COND_FP_CMP): New int iterators.
26868 (perm_hilo): Handle the new unpack unspecs.
26869 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
26871 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
26872 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
26873 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
26874 (aarch64_equality_operator, aarch64_constant_vector_operand)
26875 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
26876 (aarch64_sve_nonimmediate_operand): Likewise.
26877 (aarch64_sve_general_operand): Likewise.
26878 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
26879 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
26880 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
26881 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
26882 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
26883 (aarch64_sve_float_arith_immediate): Likewise.
26884 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
26885 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
26886 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
26887 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
26888 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
26889 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
26890 (aarch64_sve_float_arith_operand): Likewise.
26891 (aarch64_sve_float_arith_with_sub_operand): Likewise.
26892 (aarch64_sve_float_mul_operand): Likewise.
26893 (aarch64_sve_vec_perm_operand): Likewise.
26894 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
26895 (aarch64_mov_operand): Accept const_poly_int and const_vector.
26896 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
26897 as well as const_vector.
26898 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
26899 in file. Use CONST0_RTX and CONSTM1_RTX.
26900 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
26901 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
26902 Use aarch64_simd_imm_zero.
26903 * config/aarch64/aarch64-sve.md: New file.
26904 * config/aarch64/aarch64.md: Include it.
26905 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
26906 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
26907 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
26908 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
26909 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
26910 (sve): New attribute.
26911 (enabled): Disable instructions with the sve attribute unless
26913 (movqi, movhi): Pass CONST_POLY_INT operaneds through
26914 aarch64_expand_mov_immediate.
26915 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
26916 CNT[BHSD] immediates.
26917 (movti): Split CONST_POLY_INT moves into two halves.
26918 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
26919 Split additions that need a temporary here if the destination
26920 is the stack pointer.
26921 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
26922 (*add<mode>3_poly_1): New instruction.
26923 (set_clobber_cc): New expander.
26925 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26927 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
26928 parameter and use it instead of GET_MODE_SIZE (innermode). Use
26929 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
26930 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
26931 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
26932 Change innermode from fixed_mode_size to machine_mode.
26933 (simplify_subreg): Update call accordingly. Handle a constant-sized
26934 subreg of a variable-length CONST_VECTOR.
26936 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
26937 Alan Hayward <alan.hayward@arm.com>
26938 David Sherwood <david.sherwood@arm.com>
26940 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
26941 (add_offset_to_base): New function, split out from...
26942 (create_mem_ref): ...here. When handling a scale other than 1,
26943 check first whether the address is valid without the offset.
26944 Add it into the base if so, leaving the index and scale as-is.
26946 2018-01-12 Jakub Jelinek <jakub@redhat.com>
26949 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
26950 fold_for_warn before checking if arg2 is INTEGER_CST.
26952 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
26954 * config/rs6000/predicates.md (load_multiple_operation): Delete.
26955 (store_multiple_operation): Delete.
26956 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
26957 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
26958 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
26959 guarded by TARGET_STRING.
26960 (rs6000_output_load_multiple): Delete.
26961 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
26962 OPTION_MASK_STRING / TARGET_STRING handling.
26963 (print_operand) <'N', 'O'>: Add comment that these are unused now.
26964 (const rs6000_opt_masks) <"string">: Change mask to 0.
26965 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
26966 (MASK_STRING): Delete.
26967 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
26969 (load_multiple): Delete.
26976 (store_multiple): Delete.
26983 (movmemsi_8reg): Delete.
26984 (corresponding unnamed define_insn): Delete.
26985 (movmemsi_6reg): Delete.
26986 (corresponding unnamed define_insn): Delete.
26987 (movmemsi_4reg): Delete.
26988 (corresponding unnamed define_insn): Delete.
26989 (movmemsi_2reg): Delete.
26990 (corresponding unnamed define_insn): Delete.
26991 (movmemsi_1reg): Delete.
26992 (corresponding unnamed define_insn): Delete.
26993 * config/rs6000/rs6000.opt (mno-string): New.
26994 (mstring): Replace by deprecation warning stub.
26995 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
26997 2018-01-12 Jakub Jelinek <jakub@redhat.com>
26999 * regrename.c (regrename_do_replace): If replacing the same
27000 reg multiple times, try to reuse last created gen_raw_REG.
27003 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
27004 main to workaround a bug in GDB.
27006 2018-01-12 Tom de Vries <tom@codesourcery.com>
27009 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
27011 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
27013 PR rtl-optimization/80481
27014 * ira-color.c (get_cap_member): New function.
27015 (allocnos_conflict_by_live_ranges_p): Use it.
27016 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
27017 (setup_slot_coalesced_allocno_live_ranges): Ditto.
27019 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
27022 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
27023 (*saddl_se_1): Ditto.
27024 (*ssubsi_1): Ditto.
27025 (*ssubl_se_1): Ditto.
27027 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
27029 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
27030 rather than wi::to_widest for DR_INITs.
27031 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
27032 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
27033 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
27035 (vect_analyze_group_access_1): Note that here.
27037 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
27039 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
27040 polynomial type sizes.
27042 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
27044 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
27045 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
27046 (gimple_add_tmp_var): Likewise.
27048 2018-01-12 Martin Liska <mliska@suse.cz>
27050 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
27051 (gimple_alloc_sizes): Likewise.
27052 (dump_gimple_statistics): Use PRIu64 in printf format.
27053 * gimple.h: Change uint64_t to int.
27055 2018-01-12 Martin Liska <mliska@suse.cz>
27057 * tree-core.h: Use uint64_t instead of int.
27058 * tree.c (tree_node_counts): Likewise.
27059 (tree_node_sizes): Likewise.
27060 (dump_tree_statistics): Use PRIu64 in printf format.
27062 2018-01-12 Martin Liska <mliska@suse.cz>
27064 * Makefile.in: As qsort_chk is implemented in vec.c, add
27065 vec.o to linkage of gencfn-macros.
27066 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
27067 passing the info to record_node_allocation_statistics.
27068 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
27070 * ggc-common.c (struct ggc_usage): Add operator== and use
27071 it in operator< and compare function.
27072 * mem-stats.h (struct mem_usage): Likewise.
27073 * vec.c (struct vec_usage): Remove operator< and compare
27074 function. Can be simply inherited.
27076 2018-01-12 Martin Jambor <mjambor@suse.cz>
27079 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
27080 * tree-ssa-math-opts.c: Include domwalk.h.
27081 (convert_mult_to_fma_1): New function.
27082 (fma_transformation_info): New type.
27083 (fma_deferring_state): Likewise.
27084 (cancel_fma_deferring): New function.
27085 (result_of_phi): Likewise.
27086 (last_fma_candidate_feeds_initial_phi): Likewise.
27087 (convert_mult_to_fma): Added deferring logic, split actual
27088 transformation to convert_mult_to_fma_1.
27089 (math_opts_dom_walker): New type.
27090 (math_opts_dom_walker::after_dom_children): New method, body moved
27091 here from pass_optimize_widening_mul::execute, added deferring logic
27093 (pass_optimize_widening_mul::execute): Moved most of code to
27094 math_opts_dom_walker::after_dom_children.
27095 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
27096 * config/i386/i386.c (ix86_option_override_internal): Added
27097 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
27099 2018-01-12 Richard Biener <rguenther@suse.de>
27102 * dwarf2out.c (gen_variable_die): Do not reset old_die for
27103 inline instance vars.
27105 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
27108 * config/rx/rx.c (rx_is_restricted_memory_address):
27109 Handle SUBREG case.
27111 2018-01-12 Richard Biener <rguenther@suse.de>
27113 PR tree-optimization/80846
27114 * target.def (split_reduction): New target hook.
27115 * targhooks.c (default_split_reduction): New function.
27116 * targhooks.h (default_split_reduction): Declare.
27117 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
27118 target requests first reduce vectors by combining low and high
27120 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
27121 (get_vectype_for_scalar_type_and_size): Export.
27122 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
27123 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
27124 * doc/tm.texi: Regenerate.
27125 * config/i386/i386.c (ix86_split_reduction): Implement
27126 TARGET_VECTORIZE_SPLIT_REDUCTION.
27128 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
27131 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
27132 in PIC mode except for TARGET_VXWORKS_RTP.
27133 * config/sparc/sparc.c: Include cfgrtl.h.
27134 (TARGET_INIT_PIC_REG): Define.
27135 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
27136 (sparc_pic_register_p): New predicate.
27137 (sparc_legitimate_address_p): Use it.
27138 (sparc_legitimize_pic_address): Likewise.
27139 (sparc_delegitimize_address): Likewise.
27140 (sparc_mode_dependent_address_p): Likewise.
27141 (gen_load_pcrel_sym): Remove 4th parameter.
27142 (load_got_register): Adjust call to above. Remove obsolete stuff.
27143 (sparc_expand_prologue): Do not call load_got_register here.
27144 (sparc_flat_expand_prologue): Likewise.
27145 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
27146 (sparc_use_pseudo_pic_reg): New function.
27147 (sparc_init_pic_reg): Likewise.
27148 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
27149 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
27151 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
27153 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27154 Add item for branch_cost.
27156 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
27158 PR rtl-optimization/83565
27159 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
27160 not extend the result to a larger mode for rotate operations.
27161 (num_sign_bit_copies1): Likewise.
27163 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27166 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
27168 Use values-Xc.o for -pedantic.
27169 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
27171 2018-01-12 Martin Liska <mliska@suse.cz>
27174 * ipa-devirt.c (final_warning_record::grow_type_warnings):
27176 (possible_polymorphic_call_targets): Use it.
27177 (ipa_devirt): Likewise.
27179 2018-01-12 Martin Liska <mliska@suse.cz>
27181 * profile-count.h (enum profile_quality): Use 0 as invalid
27182 enum value of profile_quality.
27184 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
27186 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
27187 -mext-string options.
27189 2018-01-12 Richard Biener <rguenther@suse.de>
27191 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
27192 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
27193 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
27195 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
27197 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
27199 * configure.ac (--with-long-double-format): Add support for the
27200 configuration option to change the default long double format on
27202 * config.gcc (powerpc*-linux*-*): Likewise.
27203 * configure: Regenerate.
27204 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
27205 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
27206 used without modification.
27208 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27210 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
27211 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
27212 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
27213 MISC_BUILTIN_SPEC_BARRIER.
27214 (rs6000_init_builtins): Likewise.
27215 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
27217 (speculation_barrier): New define_insn.
27218 * doc/extend.texi: Document __builtin_speculation_barrier.
27220 2018-01-11 Jakub Jelinek <jakub@redhat.com>
27223 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
27224 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
27225 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
27227 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
27228 integral modes instead of "ss" and "sd".
27229 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
27230 vectors with 32-bit and 64-bit elements.
27231 (vecdupssescalarmodesuffix): New mode attribute.
27232 (vec_dup<mode>): Use it.
27234 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
27237 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
27238 frame if argument is passed on stack.
27240 2018-01-11 Jakub Jelinek <jakub@redhat.com>
27243 * ree.c (combine_reaching_defs): Optimize also
27244 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
27245 reg2=any_extend(exp); reg1=reg2;, formatting fix.
27247 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
27249 PR middle-end/83189
27250 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
27252 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
27254 PR middle-end/83718
27255 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
27256 after they are computed.
27258 2018-01-11 Bin Cheng <bin.cheng@arm.com>
27260 PR tree-optimization/83695
27261 * gimple-loop-linterchange.cc
27262 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
27263 reset cached scev information after interchange.
27264 (pass_linterchange::execute): Remove call to scev_reset_htab.
27266 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27268 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
27269 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
27270 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
27271 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
27272 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
27273 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
27274 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
27275 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
27276 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
27277 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
27278 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
27279 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
27280 (V_lane_reg): Likewise.
27281 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
27283 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
27284 (vfmal_lane_low<mode>_intrinsic,
27285 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
27286 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
27287 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
27288 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
27289 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
27290 vfmsl_lane_high<mode>_intrinsic): New define_insns.
27292 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27294 * config/arm/arm-cpus.in (fp16fml): New feature.
27295 (ALL_SIMD): Add fp16fml.
27296 (armv8.2-a): Add fp16fml as an option.
27297 (armv8.3-a): Likewise.
27298 (armv8.4-a): Add fp16fml as part of fp16.
27299 * config/arm/arm.h (TARGET_FP16FML): Define.
27300 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
27302 * config/arm/arm-modes.def (V2HF): Define.
27303 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
27304 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
27305 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
27306 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
27307 vfmsl_low, vfmsl_high): New set of builtins.
27308 * config/arm/iterators.md (PLUSMINUS): New code iterator.
27309 (vfml_op): New code attribute.
27310 (VFMLHALVES): New int iterator.
27311 (VFML, VFMLSEL): New mode attributes.
27312 (V_reg): Define mapping for V2HF.
27313 (V_hi, V_lo): New mode attributes.
27314 (VF_constraint): Likewise.
27315 (vfml_half, vfml_half_selector): New int attributes.
27316 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
27318 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
27319 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
27321 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
27322 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
27323 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
27324 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
27326 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
27327 Document new effective target and option set.
27329 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27331 * config/arm/arm-cpus.in (armv8_4): New feature.
27332 (ARMv8_4a): New fgroup.
27333 (armv8.4-a): New arch.
27334 * config/arm/arm-tables.opt: Regenerate.
27335 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
27336 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
27337 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
27338 Add matching rules for -march=armv8.4-a and extensions.
27339 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
27341 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
27344 * config/rx/rx.md (BW): New mode attribute.
27345 (sync_lock_test_and_setsi): Add mode suffix to insn output.
27347 2018-01-11 Richard Biener <rguenther@suse.de>
27349 PR tree-optimization/83435
27350 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
27351 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
27352 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
27354 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27355 Alan Hayward <alan.hayward@arm.com>
27356 David Sherwood <david.sherwood@arm.com>
27358 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
27360 (aarch64_classify_address): Initialize it. Track polynomial offsets.
27361 (aarch64_print_address_internal): Use it to check for a zero offset.
27363 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27364 Alan Hayward <alan.hayward@arm.com>
27365 David Sherwood <david.sherwood@arm.com>
27367 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
27368 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
27369 Return a poly_int64 rather than a HOST_WIDE_INT.
27370 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
27371 rather than a HOST_WIDE_INT.
27372 * config/aarch64/aarch64.h (aarch64_frame): Protect with
27373 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
27374 hard_fp_offset, frame_size, initial_adjust, callee_offset and
27375 final_offset from HOST_WIDE_INT to poly_int64.
27376 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
27377 to_constant when getting the number of units in an Advanced SIMD
27379 (aarch64_builtin_vectorized_function): Check for a constant number
27381 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
27383 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
27384 attribute instead of GET_MODE_NUNITS.
27385 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
27386 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
27387 GET_MODE_SIZE for fixed-size registers.
27388 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
27389 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
27390 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
27391 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
27392 (aarch64_print_operand, aarch64_print_address_internal)
27393 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
27394 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
27395 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
27396 Handle polynomial GET_MODE_SIZE.
27397 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
27398 wider than SImode without modification.
27399 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
27400 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
27401 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
27402 passing and returning SVE modes.
27403 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
27404 rather than GEN_INT.
27405 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
27406 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
27407 (aarch64_allocate_and_probe_stack_space): Likewise.
27408 (aarch64_layout_frame): Cope with polynomial offsets.
27409 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
27410 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
27411 polynomial offsets.
27412 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
27413 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
27414 poly_int64 rather than a HOST_WIDE_INT.
27415 (aarch64_get_separate_components, aarch64_process_components)
27416 (aarch64_expand_prologue, aarch64_expand_epilogue)
27417 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
27418 (aarch64_anchor_offset): New function, split out from...
27419 (aarch64_legitimize_address): ...here.
27420 (aarch64_builtin_vectorization_cost): Handle polynomial
27421 TYPE_VECTOR_SUBPARTS.
27422 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
27424 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
27425 number of elements from the PARALLEL rather than the mode.
27426 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
27427 rather than GET_MODE_BITSIZE.
27428 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
27429 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
27430 (aarch64_expand_vec_perm_const_1): Handle polynomial
27431 d->perm.length () and d->perm elements.
27432 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
27433 Apply to_constant to d->perm elements.
27434 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
27435 polynomial CONST_VECTOR_NUNITS.
27436 (aarch64_move_pointer): Take amount as a poly_int64 rather
27438 (aarch64_progress_pointer): Avoid temporary variable.
27439 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
27440 the mode attribute instead of GET_MODE.
27442 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27443 Alan Hayward <alan.hayward@arm.com>
27444 David Sherwood <david.sherwood@arm.com>
27446 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
27447 x exists before using it.
27448 (aarch64_add_constant_internal): Rename to...
27449 (aarch64_add_offset_1): ...this. Replace regnum with separate
27450 src and dest rtxes. Handle the case in which they're different,
27451 including when the offset is zero. Replace scratchreg with an rtx.
27452 Use 2 additions if there is no spare register into which we can
27453 move a 16-bit constant.
27454 (aarch64_add_constant): Delete.
27455 (aarch64_add_offset): Replace reg with separate src and dest
27456 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
27457 Use aarch64_add_offset_1.
27458 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
27459 an rtx rather than an int. Take the delta as a poly_int64
27460 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
27461 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
27462 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
27463 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
27464 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
27465 and aarch64_add_sp.
27466 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
27467 aarch64_add_constant.
27469 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27471 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
27472 Use scalar_float_mode.
27474 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27476 * config/aarch64/aarch64-simd.md
27477 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
27478 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
27479 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
27480 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
27481 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
27482 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
27483 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
27484 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
27485 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
27486 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
27488 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27491 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
27492 targ_options->x_arm_arch_string is non NULL.
27494 2018-01-11 Tamar Christina <tamar.christina@arm.com>
27496 * config/aarch64/aarch64.h
27497 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
27499 2018-01-11 Sudakshina Das <sudi.das@arm.com>
27502 * expmed.c (emit_store_flag_force): Swap if const op0
27503 and change VOIDmode to mode of op0.
27505 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
27507 PR rtl-optimization/83761
27508 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
27509 than bytes to mode_for_size.
27511 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27513 PR middle-end/83189
27514 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
27515 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
27518 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27520 PR middle-end/83575
27521 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
27522 when in layout mode.
27523 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
27524 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
27527 2018-01-10 Michael Collison <michael.collison@arm.com>
27529 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
27530 * config/aarch64/aarch64-option-extension.def: Add
27531 AARCH64_OPT_EXTENSION of 'fp16fml'.
27532 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27533 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
27534 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
27535 * config/aarch64/constraints.md (Ui7): New constraint.
27536 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
27537 (VFMLA_SEL_W): Ditto.
27540 (VFMLA16_LOW): New int iterator.
27541 (VFMLA16_HIGH): Ditto.
27542 (UNSPEC_FMLAL): New unspec.
27543 (UNSPEC_FMLSL): Ditto.
27544 (UNSPEC_FMLAL2): Ditto.
27545 (UNSPEC_FMLSL2): Ditto.
27546 (f16mac): New code attribute.
27547 * config/aarch64/aarch64-simd-builtins.def
27548 (aarch64_fmlal_lowv2sf): Ditto.
27549 (aarch64_fmlsl_lowv2sf): Ditto.
27550 (aarch64_fmlalq_lowv4sf): Ditto.
27551 (aarch64_fmlslq_lowv4sf): Ditto.
27552 (aarch64_fmlal_highv2sf): Ditto.
27553 (aarch64_fmlsl_highv2sf): Ditto.
27554 (aarch64_fmlalq_highv4sf): Ditto.
27555 (aarch64_fmlslq_highv4sf): Ditto.
27556 (aarch64_fmlal_lane_lowv2sf): Ditto.
27557 (aarch64_fmlsl_lane_lowv2sf): Ditto.
27558 (aarch64_fmlal_laneq_lowv2sf): Ditto.
27559 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
27560 (aarch64_fmlalq_lane_lowv4sf): Ditto.
27561 (aarch64_fmlsl_lane_lowv4sf): Ditto.
27562 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
27563 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
27564 (aarch64_fmlal_lane_highv2sf): Ditto.
27565 (aarch64_fmlsl_lane_highv2sf): Ditto.
27566 (aarch64_fmlal_laneq_highv2sf): Ditto.
27567 (aarch64_fmlsl_laneq_highv2sf): Ditto.
27568 (aarch64_fmlalq_lane_highv4sf): Ditto.
27569 (aarch64_fmlsl_lane_highv4sf): Ditto.
27570 (aarch64_fmlalq_laneq_highv4sf): Ditto.
27571 (aarch64_fmlsl_laneq_highv4sf): Ditto.
27572 * config/aarch64/aarch64-simd.md:
27573 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
27574 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27575 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
27576 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
27577 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
27578 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
27579 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
27580 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
27581 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
27582 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
27583 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
27584 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
27585 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
27586 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
27587 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
27588 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
27589 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
27590 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
27591 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
27592 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
27593 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
27594 (vfmlsl_low_u32): Ditto.
27595 (vfmlalq_low_u32): Ditto.
27596 (vfmlslq_low_u32): Ditto.
27597 (vfmlal_high_u32): Ditto.
27598 (vfmlsl_high_u32): Ditto.
27599 (vfmlalq_high_u32): Ditto.
27600 (vfmlslq_high_u32): Ditto.
27601 (vfmlal_lane_low_u32): Ditto.
27602 (vfmlsl_lane_low_u32): Ditto.
27603 (vfmlal_laneq_low_u32): Ditto.
27604 (vfmlsl_laneq_low_u32): Ditto.
27605 (vfmlalq_lane_low_u32): Ditto.
27606 (vfmlslq_lane_low_u32): Ditto.
27607 (vfmlalq_laneq_low_u32): Ditto.
27608 (vfmlslq_laneq_low_u32): Ditto.
27609 (vfmlal_lane_high_u32): Ditto.
27610 (vfmlsl_lane_high_u32): Ditto.
27611 (vfmlal_laneq_high_u32): Ditto.
27612 (vfmlsl_laneq_high_u32): Ditto.
27613 (vfmlalq_lane_high_u32): Ditto.
27614 (vfmlslq_lane_high_u32): Ditto.
27615 (vfmlalq_laneq_high_u32): Ditto.
27616 (vfmlslq_laneq_high_u32): Ditto.
27617 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
27618 (AARCH64_FL_FOR_ARCH8_4): New.
27619 (AARCH64_ISA_F16FML): New ISA flag.
27620 (TARGET_F16FML): New feature flag for fp16fml.
27621 (doc/invoke.texi): Document new fp16fml option.
27623 2018-01-10 Michael Collison <michael.collison@arm.com>
27625 * config/aarch64/aarch64-builtins.c:
27626 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
27627 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27628 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
27629 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
27630 (AARCH64_ISA_SHA3): New ISA flag.
27631 (TARGET_SHA3): New feature flag for sha3.
27632 * config/aarch64/iterators.md (sha512_op): New int attribute.
27633 (CRYPTO_SHA512): New int iterator.
27634 (UNSPEC_SHA512H): New unspec.
27635 (UNSPEC_SHA512H2): Ditto.
27636 (UNSPEC_SHA512SU0): Ditto.
27637 (UNSPEC_SHA512SU1): Ditto.
27638 * config/aarch64/aarch64-simd-builtins.def
27639 (aarch64_crypto_sha512hqv2di): New builtin.
27640 (aarch64_crypto_sha512h2qv2di): Ditto.
27641 (aarch64_crypto_sha512su0qv2di): Ditto.
27642 (aarch64_crypto_sha512su1qv2di): Ditto.
27643 (aarch64_eor3qv8hi): Ditto.
27644 (aarch64_rax1qv2di): Ditto.
27645 (aarch64_xarqv2di): Ditto.
27646 (aarch64_bcaxqv8hi): Ditto.
27647 * config/aarch64/aarch64-simd.md:
27648 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
27649 (aarch64_crypto_sha512su0qv2di): Ditto.
27650 (aarch64_crypto_sha512su1qv2di): Ditto.
27651 (aarch64_eor3qv8hi): Ditto.
27652 (aarch64_rax1qv2di): Ditto.
27653 (aarch64_xarqv2di): Ditto.
27654 (aarch64_bcaxqv8hi): Ditto.
27655 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
27656 (vsha512h2q_u64): Ditto.
27657 (vsha512su0q_u64): Ditto.
27658 (vsha512su1q_u64): Ditto.
27659 (veor3q_u16): Ditto.
27660 (vrax1q_u64): Ditto.
27661 (vxarq_u64): Ditto.
27662 (vbcaxq_u16): Ditto.
27663 * config/arm/types.md (crypto_sha512): New type attribute.
27664 (crypto_sha3): Ditto.
27665 (doc/invoke.texi): Document new sha3 option.
27667 2018-01-10 Michael Collison <michael.collison@arm.com>
27669 * config/aarch64/aarch64-builtins.c:
27670 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
27671 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27672 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
27673 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
27674 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
27675 (AARCH64_ISA_SM4): New ISA flag.
27676 (TARGET_SM4): New feature flag for sm4.
27677 * config/aarch64/aarch64-simd-builtins.def
27678 (aarch64_sm3ss1qv4si): Ditto.
27679 (aarch64_sm3tt1aq4si): Ditto.
27680 (aarch64_sm3tt1bq4si): Ditto.
27681 (aarch64_sm3tt2aq4si): Ditto.
27682 (aarch64_sm3tt2bq4si): Ditto.
27683 (aarch64_sm3partw1qv4si): Ditto.
27684 (aarch64_sm3partw2qv4si): Ditto.
27685 (aarch64_sm4eqv4si): Ditto.
27686 (aarch64_sm4ekeyqv4si): Ditto.
27687 * config/aarch64/aarch64-simd.md:
27688 (aarch64_sm3ss1qv4si): Ditto.
27689 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
27690 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
27691 (aarch64_sm4eqv4si): Ditto.
27692 (aarch64_sm4ekeyqv4si): Ditto.
27693 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
27694 (sm3part_op): Ditto.
27695 (CRYPTO_SM3TT): Ditto.
27696 (CRYPTO_SM3PART): Ditto.
27697 (UNSPEC_SM3SS1): New unspec.
27698 (UNSPEC_SM3TT1A): Ditto.
27699 (UNSPEC_SM3TT1B): Ditto.
27700 (UNSPEC_SM3TT2A): Ditto.
27701 (UNSPEC_SM3TT2B): Ditto.
27702 (UNSPEC_SM3PARTW1): Ditto.
27703 (UNSPEC_SM3PARTW2): Ditto.
27704 (UNSPEC_SM4E): Ditto.
27705 (UNSPEC_SM4EKEY): Ditto.
27706 * config/aarch64/constraints.md (Ui2): New constraint.
27707 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
27708 * config/arm/types.md (crypto_sm3): New type attribute.
27709 (crypto_sm4): Ditto.
27710 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
27711 (vsm3tt1aq_u32): Ditto.
27712 (vsm3tt1bq_u32): Ditto.
27713 (vsm3tt2aq_u32): Ditto.
27714 (vsm3tt2bq_u32): Ditto.
27715 (vsm3partw1q_u32): Ditto.
27716 (vsm3partw2q_u32): Ditto.
27717 (vsm4eq_u32): Ditto.
27718 (vsm4ekeyq_u32): Ditto.
27719 (doc/invoke.texi): Document new sm4 option.
27721 2018-01-10 Michael Collison <michael.collison@arm.com>
27723 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
27724 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
27725 (AARCH64_FL_FOR_ARCH8_4): New.
27726 (AARCH64_FL_V8_4): New flag.
27727 (doc/invoke.texi): Document new armv8.4-a option.
27729 2018-01-10 Michael Collison <michael.collison@arm.com>
27731 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
27732 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
27733 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
27734 * config/aarch64/aarch64-option-extension.def: Add
27735 AARCH64_OPT_EXTENSION of 'sha2'.
27736 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
27737 (crypto): Disable sha2 and aes if crypto disabled.
27738 (crypto): Enable aes and sha2 if enabled.
27739 (simd): Disable sha2 and aes if simd disabled.
27740 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
27742 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
27743 (TARGET_SHA2): New feature flag for sha2.
27744 (TARGET_AES): New feature flag for aes.
27745 * config/aarch64/aarch64-simd.md:
27746 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
27747 conditional on TARGET_AES.
27748 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
27749 (aarch64_crypto_sha1hsi): Make pattern conditional
27751 (aarch64_crypto_sha1hv4si): Ditto.
27752 (aarch64_be_crypto_sha1hv4si): Ditto.
27753 (aarch64_crypto_sha1su1v4si): Ditto.
27754 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
27755 (aarch64_crypto_sha1su0v4si): Ditto.
27756 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
27757 (aarch64_crypto_sha256su0v4si): Ditto.
27758 (aarch64_crypto_sha256su1v4si): Ditto.
27759 (doc/invoke.texi): Document new aes and sha2 options.
27761 2018-01-10 Martin Sebor <msebor@redhat.com>
27763 PR tree-optimization/83781
27764 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
27767 2018-01-11 Martin Sebor <msebor@gmail.com>
27768 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27770 PR tree-optimization/83501
27771 PR tree-optimization/81703
27773 * tree-ssa-strlen.c (get_string_cst): Rename...
27774 (get_string_len): ...to this. Handle global constants.
27775 (handle_char_store): Adjust.
27777 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
27778 Jim Wilson <jimw@sifive.com>
27780 * config/riscv/riscv-protos.h (riscv_output_return): New.
27781 * config/riscv/riscv.c (struct machine_function): New naked_p field.
27782 (riscv_attribute_table, riscv_output_return),
27783 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
27784 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
27785 (riscv_compute_frame_info): Only compute frame->mask if not a naked
27787 (riscv_expand_prologue): Add early return for naked function.
27788 (riscv_expand_epilogue): Likewise.
27789 (riscv_function_ok_for_sibcall): Return false for naked function.
27790 (riscv_set_current_function): New.
27791 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
27792 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
27793 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
27794 * doc/extend.texi (RISC-V Function Attributes): New.
27796 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
27798 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
27799 check for 128-bit long double before checking TCmode.
27800 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
27801 128-bit long doubles before checking TFmode or TCmode.
27802 (FLOAT128_IBM_P): Likewise.
27804 2018-01-10 Martin Sebor <msebor@redhat.com>
27806 PR tree-optimization/83671
27807 * builtins.c (c_strlen): Unconditionally return zero for the empty
27809 Use -Warray-bounds for warnings.
27810 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
27811 for non-constant array indices with COMPONENT_REF, arrays of
27812 arrays, and pointers to arrays.
27813 (gimple_fold_builtin_strlen): Determine and set length range for
27814 non-constant character arrays.
27816 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
27818 PR middle-end/81897
27819 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
27822 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
27824 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
27826 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27829 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
27830 VECTOR_MEM_ALTIVEC_OR_VSX_P.
27831 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
27832 indexed_or_indirect_operand predicate.
27833 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
27834 (*vsx_le_perm_load_v8hi): Likewise.
27835 (*vsx_le_perm_load_v16qi): Likewise.
27836 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
27837 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
27838 (*vsx_le_perm_store_v8hi): Likewise.
27839 (*vsx_le_perm_store_v16qi): Likewise.
27840 (eight unnamed splitters): Likewise.
27842 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
27844 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
27845 * config/rs6000/emmintrin.h: Likewise.
27846 * config/rs6000/mmintrin.h: Likewise.
27847 * config/rs6000/xmmintrin.h: Likewise.
27849 2018-01-10 David Malcolm <dmalcolm@redhat.com>
27852 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
27854 * tree.c (tree_nop_conversion): Return true for location wrapper
27856 (maybe_wrap_with_location): New function.
27857 (selftest::check_strip_nops): New function.
27858 (selftest::test_location_wrappers): New function.
27859 (selftest::tree_c_tests): Call it.
27860 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
27861 (maybe_wrap_with_location): New decl.
27862 (EXPR_LOCATION_WRAPPER_P): New macro.
27863 (location_wrapper_p): New inline function.
27864 (tree_strip_any_location_wrapper): New inline function.
27866 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
27869 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
27870 stack_realign_offset for the largest alignment of stack slot
27872 (ix86_find_max_used_stack_alignment): New function.
27873 (ix86_finalize_stack_frame_flags): Use it. Set
27874 max_used_stack_alignment if we don't realign stack.
27875 * config/i386/i386.h (machine_function): Add
27876 max_used_stack_alignment.
27878 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
27880 * config/arm/arm.opt (-mbranch-cost): New option.
27881 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
27884 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
27887 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
27888 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
27890 2018-01-10 Richard Biener <rguenther@suse.de>
27893 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
27894 early out so it also covers the case where we have a non-NULL
27897 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
27899 PR tree-optimization/83753
27900 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
27901 for non-strided grouped accesses if the number of elements is 1.
27903 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
27906 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
27907 * i386.h (TARGET_USE_GATHER): Define.
27908 * x86-tune.def (X86_TUNE_USE_GATHER): New.
27910 2018-01-10 Martin Liska <mliska@suse.cz>
27913 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
27914 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
27916 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
27917 CLEANUP_NO_PARTITIONING is not set.
27919 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
27921 * doc/rtl.texi: Remove documentation of (const ...) wrappers
27922 for vectors, as a partial revert of r254296.
27923 * rtl.h (const_vec_p): Delete.
27924 (const_vec_duplicate_p): Don't test for vector CONSTs.
27925 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
27926 * expmed.c (make_tree): Likewise.
27929 * common.md (E, F): Use CONSTANT_P instead of checking for
27931 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
27932 checking for CONST_VECTOR.
27934 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
27936 PR middle-end/83575
27937 * predict.c (force_edge_cold): Handle in more sane way edges
27938 with no prediction.
27940 2018-01-09 Carl Love <cel@us.ibm.com>
27942 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
27944 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
27945 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
27946 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
27947 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
27948 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
27949 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
27950 * config/rs6000/rs6000-protos.h: Add extern defition for
27951 rs6000_generate_float2_double_code.
27952 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
27954 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
27955 (float2_v2df): Add define_expand.
27957 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
27960 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
27961 op_mode in the force_to_mode call.
27963 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27965 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
27966 instead of checking each element individually.
27967 (aarch64_evpc_uzp): Likewise.
27968 (aarch64_evpc_zip): Likewise.
27969 (aarch64_evpc_ext): Likewise.
27970 (aarch64_evpc_rev): Likewise.
27971 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
27972 instead of checking each element individually. Return true without
27974 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
27975 whether all selected elements come from the same input, instead of
27976 checking each element individually. Remove calls to gen_rtx_REG,
27977 start_sequence and end_sequence and instead assert that no rtl is
27980 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27982 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
27983 order of HIGH and CONST checks.
27985 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
27987 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
27988 if the destination isn't an SSA_NAME.
27990 2018-01-09 Richard Biener <rguenther@suse.de>
27992 PR tree-optimization/83668
27993 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
27995 (canonicalize_loop_form): ... here, renamed from ...
27996 (canonicalize_loop_closed_ssa_form): ... this and amended to
27997 swap successor edges for loop exit blocks to make us use
27998 the RPO order we need for initial schedule generation.
28000 2018-01-09 Joseph Myers <joseph@codesourcery.com>
28002 PR tree-optimization/64811
28003 * match.pd: When optimizing comparisons with Inf, avoid
28004 introducing or losing exceptions from comparisons with NaN.
28006 2018-01-09 Martin Liska <mliska@suse.cz>
28009 * asan.c (shadow_mem_size): Add gcc_assert.
28011 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
28013 Don't save registers in main().
28016 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
28017 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
28018 * config/avr/avr.c (avr_set_current_function): Don't error if
28019 naked, OS_task or OS_main are specified at the same time.
28020 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
28022 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
28024 * common/config/avr/avr-common.c (avr_option_optimization_table):
28025 Switch on -mmain-is-OS_task for optimizing compilations.
28027 2018-01-09 Richard Biener <rguenther@suse.de>
28029 PR tree-optimization/83572
28030 * graphite.c: Include cfganal.h.
28031 (graphite_transform_loops): Connect infinite loops to exit
28032 and remove fake edges at the end.
28034 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
28036 * ipa-inline.c (edge_badness): Revert accidental checkin.
28038 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
28041 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
28042 symbols; not inline clones.
28044 2018-01-09 Jakub Jelinek <jakub@redhat.com>
28047 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
28048 hard registers. Formatting fixes.
28050 PR preprocessor/83722
28051 * gcc.c (try_generate_repro): Pass
28052 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
28053 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
28056 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
28057 Kito Cheng <kito.cheng@gmail.com>
28059 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
28060 (riscv_leaf_function_p): Delete.
28061 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
28063 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28065 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
28067 (do_ifelse): New function.
28068 (do_isel): New function.
28069 (do_sub3): New function.
28070 (do_add3): New function.
28071 (do_load_mask_compare): New function.
28072 (do_overlap_load_compare): New function.
28073 (expand_compare_loop): New function.
28074 (expand_block_compare): Call expand_compare_loop() when appropriate.
28075 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
28076 option description.
28077 (-mblock-compare-inline-loop-limit): New option.
28079 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28082 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
28083 Reverse order of second and third operands in first alternative.
28084 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
28085 of first and second elements in UNSPEC_VPERMR vector.
28086 (altivec_expand_vec_perm_le): Likewise.
28088 2018-01-08 Jeff Law <law@redhat.com>
28090 PR rtl-optimizatin/81308
28091 * tree-switch-conversion.c (cfg_altered): New file scoped static.
28092 (process_switch): If group_case_labels makes a change, then set
28094 (pass_convert_switch::execute): If a switch is converted, then
28095 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
28097 PR rtl-optimization/81308
28098 * recog.c (split_all_insns): Conditionally cleanup the CFG after
28101 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
28103 PR target/83663 - Revert r255946
28104 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
28105 generation for cases where splatting a value is not useful.
28106 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
28107 across a vec_duplicate and a paradoxical subreg forming a vector
28108 mode to a vec_concat.
28110 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28112 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
28113 -march=armv8.3-a variants.
28114 * config/arm/t-multilib: Likewise.
28115 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
28117 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28119 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
28121 (cceq_ior_compare_complement): Give it a name so I can use it, and
28122 change boolean_or_operator predicate to boolean_operator so it can
28123 be used to generate a crand.
28124 (eqne): New code iterator.
28125 (bd/bd_neg): New code_attrs.
28126 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
28127 a single define_insn.
28128 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
28129 decrement (bdnzt/bdnzf/bdzt/bdzf).
28130 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
28131 with the new names of the branch decrement patterns, and added the
28132 names of the branch decrement conditional patterns.
28134 2018-01-08 Richard Biener <rguenther@suse.de>
28136 PR tree-optimization/83563
28137 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
28140 2018-01-08 Richard Biener <rguenther@suse.de>
28142 PR middle-end/83713
28143 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
28145 2018-01-08 Richard Biener <rguenther@suse.de>
28147 PR tree-optimization/83685
28148 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
28149 references to abnormals.
28151 2018-01-08 Richard Biener <rguenther@suse.de>
28154 * dwarf2out.c (output_indirect_strings): Handle empty
28155 skeleton_debug_str_hash.
28156 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
28158 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
28160 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
28161 (emit_store_direct): Likewise.
28162 (arc_trampoline_adjust_address): Likewise.
28163 (arc_asm_trampoline_template): New function.
28164 (arc_initialize_trampoline): Use asm_trampoline_template.
28165 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
28166 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
28167 * config/arc/arc.md (flush_icache): Delete pattern.
28169 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
28171 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
28172 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
28175 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
28178 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
28179 by not USED_FOR_TARGET.
28180 (make_pass_resolve_sw_modes): Likewise.
28182 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
28184 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
28187 2018-01-08 Richard Biener <rguenther@suse.de>
28189 PR middle-end/83580
28190 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
28192 2018-01-08 Richard Biener <rguenther@suse.de>
28194 PR middle-end/83517
28195 * match.pd ((t * 2) / 2) -> t): Add missing :c.
28197 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
28199 PR middle-end/81897
28200 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
28201 basic blocks with a small number of successors.
28202 (convert_control_dep_chain_into_preds): Improve handling of
28204 (dump_predicates): Split apart into...
28205 (dump_pred_chain): ...here...
28206 (dump_pred_info): ...and here.
28207 (can_one_predicate_be_invalidated_p): Add debugging printfs.
28208 (can_chain_union_be_invalidated_p): Improve check for invalidation
28210 (uninit_uses_cannot_happen): Avoid unnecessary if
28211 convert_control_dep_chain_into_preds yielded nothing.
28213 2018-01-06 Martin Sebor <msebor@redhat.com>
28215 PR tree-optimization/83640
28216 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
28217 subtracting negative offset from size.
28218 (builtin_access::overlap): Adjust offset bounds of the access to fall
28219 within the size of the object if possible.
28221 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
28223 PR rtl-optimization/83699
28224 * expmed.c (extract_bit_field_1): Restrict the vector usage of
28225 extract_bit_field_as_subreg to cases in which the extracted
28226 value is also a vector.
28228 * lra-constraints.c (process_alt_operands): Test for the equivalence
28229 substitutions when detecting a possible reload cycle.
28231 2018-01-06 Jakub Jelinek <jakub@redhat.com>
28234 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
28235 by default if flag_selective_schedling{,2}. Formatting fixes.
28237 PR rtl-optimization/83682
28238 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
28239 if it has non-VECTOR_MODE element mode.
28240 (vec_duplicate_p): Likewise.
28242 PR middle-end/83694
28243 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
28244 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
28246 2018-01-05 Jakub Jelinek <jakub@redhat.com>
28249 * config/i386/i386-builtin.def
28250 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
28251 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
28252 Require also OPTION_MASK_ISA_AVX512F in addition to
28253 OPTION_MASK_ISA_GFNI.
28254 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
28255 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
28256 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
28257 to OPTION_MASK_ISA_GFNI.
28258 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
28259 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
28260 OPTION_MASK_ISA_AVX512BW.
28261 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
28262 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
28263 addition to OPTION_MASK_ISA_GFNI.
28264 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
28265 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
28266 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
28267 to OPTION_MASK_ISA_GFNI.
28268 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
28269 a requirement for all ISAs rather than any of them with a few
28271 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
28273 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
28274 bitmasks to be enabled with 3 exceptions, instead of requiring any
28275 enabled ISA with lots of exceptions.
28276 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
28277 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
28278 Change avx512bw in isa attribute to avx512f.
28279 * config/i386/sgxintrin.h: Add license boilerplate.
28280 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
28281 to __AVX512F__ and __AVX512VL to __AVX512VL__.
28282 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
28283 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
28285 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
28286 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
28287 temporarily sse2 rather than sse if not enabled already.
28290 * config/i386/sse.md (VI248_VLBW): Rename to ...
28291 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
28292 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
28293 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
28294 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
28295 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
28296 mode iterator instead of VI248_VLBW.
28298 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
28300 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
28301 (record_modified): Skip clobbers; add debug output.
28302 (param_change_prob): Use sreal frequencies.
28304 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
28306 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
28307 punt for user-aligned variables.
28309 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
28311 * tree-chrec.c (chrec_contains_symbols): Return true for
28314 2018-01-05 Sudakshina Das <sudi.das@arm.com>
28317 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
28318 of (x|y) == x for BICS pattern.
28320 2018-01-05 Jakub Jelinek <jakub@redhat.com>
28322 PR tree-optimization/83605
28323 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
28324 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
28327 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
28329 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
28330 * config/epiphany/rtems.h: New file.
28332 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28333 Uros Bizjak <ubizjak@gmail.com>
28336 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
28337 QIreg_operand instead of register_operand predicate.
28338 * config/i386/i386.c (ix86_rop_should_change_byte_p,
28339 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
28340 comments instead of -fmitigate[-_]rop.
28342 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28345 * cgraphunit.c (symbol_table::compile): Switch to text_section
28346 before calling assembly_start debug hook.
28347 * run-rtl-passes.c (run_rtl_passes): Likewise.
28350 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28352 * tree-vrp.c (extract_range_from_binary_expr_1): Check
28353 range_int_cst_p rather than !symbolic_range_p before calling
28354 extract_range_from_multiplicative_op_1.
28356 2018-01-04 Jeff Law <law@redhat.com>
28358 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
28359 redundant test in assertion.
28361 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28363 * doc/rtl.texi: Document machine_mode wrapper classes.
28365 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28367 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
28368 using tree_to_uhwi.
28370 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28372 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
28373 the VEC_PERM_EXPR fold to fail.
28375 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28378 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
28379 to switched_sections.
28381 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28384 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
28385 test for d.testing.
28387 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
28390 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
28391 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
28393 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28396 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
28397 is BLKmode and bitpos not zero or mode change is needed.
28399 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
28402 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
28405 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
28408 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
28409 instead of MULT rtx. Update all corresponding splitters.
28410 (*saddl_se): Ditto.
28411 (*ssub<modesuffix>): Ditto.
28412 (*ssubl_se): Ditto.
28413 (*cmp_sadd_di): Update split patterns.
28414 (*cmp_sadd_si): Ditto.
28415 (*cmp_sadd_sidi): Ditto.
28416 (*cmp_ssub_di): Ditto.
28417 (*cmp_ssub_si): Ditto.
28418 (*cmp_ssub_sidi): Ditto.
28419 * config/alpha/predicates.md (const23_operand): New predicate.
28420 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
28421 Look for ASHIFT, not MULT inner operand.
28422 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
28424 2018-01-04 Martin Liska <mliska@suse.cz>
28426 PR gcov-profile/83669
28427 * gcov.c (output_intermediate_file): Add version to intermediate
28429 * doc/gcov.texi: Document new field 'version' in intermediate
28430 file format. Fix location of '-k' option of gcov command.
28432 2018-01-04 Martin Liska <mliska@suse.cz>
28435 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
28437 2018-01-04 Jakub Jelinek <jakub@redhat.com>
28439 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
28441 2018-01-03 Martin Sebor <msebor@redhat.com>
28443 PR tree-optimization/83655
28444 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
28445 checking calls with invalid arguments.
28447 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28449 * tree-vect-stmts.c (vect_get_store_rhs): New function.
28450 (vectorizable_mask_load_store): Delete.
28451 (vectorizable_call): Return false for masked loads and stores.
28452 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
28453 instead of gimple_assign_rhs1.
28454 (vectorizable_load): Handle IFN_MASK_LOAD.
28455 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
28457 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28459 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
28461 (vectorizable_mask_load_store): ...here.
28462 (vectorizable_load): ...and here.
28464 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28466 * tree-vect-stmts.c (vect_build_all_ones_mask)
28467 (vect_build_zero_merge_argument): New functions, split out from...
28468 (vectorizable_load): ...here.
28470 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28472 * tree-vect-stmts.c (vect_check_store_rhs): New function,
28474 (vectorizable_mask_load_store): ...here.
28475 (vectorizable_store): ...and here.
28477 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28479 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
28481 (vectorizable_mask_load_store): ...here.
28483 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28485 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
28486 (vect_model_store_cost): Take a vec_load_store_type instead of a
28488 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
28489 (vect_model_store_cost): Take a vec_load_store_type instead of a
28491 (vectorizable_mask_load_store): Update accordingly.
28492 (vectorizable_store): Likewise.
28493 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
28495 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28497 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
28498 IFN_MASK_LOAD calls here rather than...
28499 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
28501 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28502 Alan Hayward <alan.hayward@arm.com>
28503 David Sherwood <david.sherwood@arm.com>
28505 * expmed.c (extract_bit_field_1): For vector extracts,
28506 fall back to extract_bit_field_as_subreg if vec_extract
28509 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28510 Alan Hayward <alan.hayward@arm.com>
28511 David Sherwood <david.sherwood@arm.com>
28513 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
28514 they are variable or constant sized.
28515 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
28516 slots for constant-sized data.
28518 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28519 Alan Hayward <alan.hayward@arm.com>
28520 David Sherwood <david.sherwood@arm.com>
28522 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
28523 handling COND_EXPRs with boolean comparisons, try to find a better
28524 basis for the mask type than the boolean itself.
28526 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28528 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
28529 is calculated and how it can be overridden.
28530 * genmodes.c (max_bitsize_mode_any_mode): New variable.
28531 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
28533 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
28536 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28537 Alan Hayward <alan.hayward@arm.com>
28538 David Sherwood <david.sherwood@arm.com>
28540 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
28541 Remove the mode argument.
28542 (aarch64_simd_valid_immediate): Remove the mode and inverse
28544 * config/aarch64/iterators.md (bitsize): New iterator.
28545 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
28546 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
28547 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
28548 aarch64_simd_valid_immediate.
28549 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
28550 (aarch64_reg_or_bic_imm): Likewise.
28551 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
28552 with an insn_type enum and msl with a modifier_type enum.
28553 Replace element_width with a scalar_mode. Change the shift
28554 to unsigned int. Add constructors for scalar_float_mode and
28555 scalar_int_mode elements.
28556 (aarch64_vect_float_const_representable_p): Delete.
28557 (aarch64_can_const_movi_rtx_p)
28558 (aarch64_simd_scalar_immediate_valid_for_move)
28559 (aarch64_simd_make_constant): Update call to
28560 aarch64_simd_valid_immediate.
28561 (aarch64_advsimd_valid_immediate_hs): New function.
28562 (aarch64_advsimd_valid_immediate): Likewise.
28563 (aarch64_simd_valid_immediate): Remove mode and inverse
28564 arguments. Rewrite to use the above. Use const_vec_duplicate_p
28565 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
28566 and aarch64_float_const_representable_p on the result.
28567 (aarch64_output_simd_mov_immediate): Remove mode argument.
28568 Update call to aarch64_simd_valid_immediate and use of
28569 simd_immediate_info.
28570 (aarch64_output_scalar_simd_mov_immediate): Update call
28573 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28574 Alan Hayward <alan.hayward@arm.com>
28575 David Sherwood <david.sherwood@arm.com>
28577 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
28578 (mode_nunits): Likewise CONST_MODE_NUNITS.
28579 * machmode.def (ADJUST_NUNITS): Document.
28580 * genmodes.c (mode_data::need_nunits_adj): New field.
28581 (blank_mode): Update accordingly.
28582 (adj_nunits): New variable.
28583 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
28585 (emit_mode_size_inline): Set need_bytesize_adj for all modes
28586 listed in adj_nunits.
28587 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
28588 listed in adj_nunits. Don't emit case statements for such modes.
28589 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
28590 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
28591 nothing if adj_nunits is nonnull.
28592 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
28593 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
28594 (emit_mode_fbit): Update use of print_maybe_const_decl.
28595 (emit_move_size): Likewise. Treat the array as non-const
28597 (emit_mode_adjustments): Handle adj_nunits.
28599 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28601 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
28602 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
28603 (VECTOR_MODES): Use it.
28604 (make_vector_modes): Take the prefix as an argument.
28606 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28607 Alan Hayward <alan.hayward@arm.com>
28608 David Sherwood <david.sherwood@arm.com>
28610 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
28611 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
28612 for MODE_VECTOR_BOOL.
28613 * machmode.def (VECTOR_BOOL_MODE): Document.
28614 * genmodes.c (VECTOR_BOOL_MODE): New macro.
28615 (make_vector_bool_mode): New function.
28616 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
28618 * lto-streamer-in.c (lto_input_mode_table): Likewise.
28619 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
28621 * stor-layout.c (int_mode_for_mode): Likewise.
28622 * tree.c (build_vector_type_for_mode): Likewise.
28623 * varasm.c (output_constant_pool_2): Likewise.
28624 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
28625 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
28626 for MODE_VECTOR_BOOL.
28627 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
28628 of mode class checks.
28629 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
28630 instead of a list of mode class checks.
28631 (expand_vector_scalar_condition): Likewise.
28632 (type_for_widest_vector_mode): Handle BImode as an inner mode.
28634 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28635 Alan Hayward <alan.hayward@arm.com>
28636 David Sherwood <david.sherwood@arm.com>
28638 * machmode.h (mode_size): Change from unsigned short to
28640 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
28641 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28642 or if measurement_type is not polynomial.
28643 (fixed_size_mode::includes_p): Check for constant-sized modes.
28644 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
28645 return a poly_uint16 rather than an unsigned short.
28646 (emit_mode_size): Change the type of mode_size from unsigned short
28647 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
28648 (emit_mode_adjustments): Cope with polynomial vector sizes.
28649 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28651 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28653 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
28654 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
28655 * caller-save.c (setup_save_areas): Likewise.
28656 (replace_reg_with_saved_mem): Likewise.
28657 * calls.c (emit_library_call_value_1): Likewise.
28658 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
28659 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
28660 (gen_lowpart_for_combine): Likewise.
28661 * convert.c (convert_to_integer_1): Likewise.
28662 * cse.c (equiv_constant, cse_insn): Likewise.
28663 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
28664 (cselib_subst_to_values): Likewise.
28665 * dce.c (word_dce_process_block): Likewise.
28666 * df-problems.c (df_word_lr_mark_ref): Likewise.
28667 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
28668 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
28669 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
28670 (rtl_for_decl_location): Likewise.
28671 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
28672 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
28673 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
28674 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
28675 (expand_expr_real_1): Likewise.
28676 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
28677 (pad_below): Likewise.
28678 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28679 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
28680 * ira.c (get_subreg_tracking_sizes): Likewise.
28681 * ira-build.c (ira_create_allocno_objects): Likewise.
28682 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
28683 (ira_sort_regnos_for_alter_reg): Likewise.
28684 * ira-costs.c (record_operand_costs): Likewise.
28685 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
28686 (resolve_simple_move): Likewise.
28687 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
28688 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
28689 (lra_constraints): Likewise.
28690 (CONST_POOL_OK_P): Reject variable-sized modes.
28691 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
28692 (add_pseudo_to_slot, lra_spill): Likewise.
28693 * omp-low.c (omp_clause_aligned_alignment): Likewise.
28694 * optabs-query.c (get_best_extraction_insn): Likewise.
28695 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
28696 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
28697 (expand_mult_highpart, valid_multiword_target_p): Likewise.
28698 * recog.c (offsettable_address_addr_space_p): Likewise.
28699 * regcprop.c (maybe_mode_change): Likewise.
28700 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
28701 * regrename.c (build_def_use): Likewise.
28702 * regstat.c (dump_reg_info): Likewise.
28703 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
28704 (find_reloads, find_reloads_subreg_address): Likewise.
28705 * reload1.c (eliminate_regs_1): Likewise.
28706 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
28707 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
28708 (simplify_binary_operation_1, simplify_subreg): Likewise.
28709 * targhooks.c (default_function_arg_padding): Likewise.
28710 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
28711 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
28712 (verify_gimple_assign_ternary): Likewise.
28713 * tree-inline.c (estimate_move_cost): Likewise.
28714 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28715 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
28716 (get_address_cost_ainc): Likewise.
28717 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28718 (vect_supportable_dr_alignment): Likewise.
28719 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28720 (vectorizable_reduction): Likewise.
28721 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
28722 (vectorizable_operation, vectorizable_load): Likewise.
28723 * tree.c (build_same_sized_truth_vector_type): Likewise.
28724 * valtrack.c (cleanup_auto_inc_dec): Likewise.
28725 * var-tracking.c (emit_note_insn_var_location): Likewise.
28726 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
28727 (ADDR_VEC_ALIGN): Likewise.
28729 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28730 Alan Hayward <alan.hayward@arm.com>
28731 David Sherwood <david.sherwood@arm.com>
28733 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
28735 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
28736 or if measurement_type is polynomial.
28737 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
28738 * combine.c (make_extraction): Likewise.
28739 * dse.c (find_shift_sequence): Likewise.
28740 * dwarf2out.c (mem_loc_descriptor): Likewise.
28741 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
28742 (extract_bit_field, extract_low_bits): Likewise.
28743 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
28744 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
28745 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
28746 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
28747 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
28748 * reload.c (find_reloads): Likewise.
28749 * reload1.c (alter_reg): Likewise.
28750 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
28751 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
28752 * tree-if-conv.c (predicate_mem_writes): Likewise.
28753 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
28754 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
28755 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
28756 * valtrack.c (dead_debug_insert_temp): Likewise.
28757 * varasm.c (mergeable_constant_section): Likewise.
28758 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
28760 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28761 Alan Hayward <alan.hayward@arm.com>
28762 David Sherwood <david.sherwood@arm.com>
28764 * expr.c (expand_assignment): Cope with polynomial mode sizes
28765 when assigning to a CONCAT.
28767 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28768 Alan Hayward <alan.hayward@arm.com>
28769 David Sherwood <david.sherwood@arm.com>
28771 * machmode.h (mode_precision): Change from unsigned short to
28773 (mode_to_precision): Return a poly_uint16 rather than an unsigned
28775 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
28776 or if measurement_type is not polynomial.
28777 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
28778 in which the mode is already known to be a scalar_int_mode.
28779 * genmodes.c (emit_mode_precision): Change the type of mode_precision
28780 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
28782 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28783 for GET_MODE_PRECISION.
28784 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28785 for GET_MODE_PRECISION.
28786 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
28788 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
28789 (expand_field_assignment, make_extraction): Likewise.
28790 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
28791 (get_last_value): Likewise.
28792 * convert.c (convert_to_integer_1): Likewise.
28793 * cse.c (cse_insn): Likewise.
28794 * expr.c (expand_expr_real_1): Likewise.
28795 * lra-constraints.c (simplify_operand_subreg): Likewise.
28796 * optabs-query.c (can_atomic_load_p): Likewise.
28797 * optabs.c (expand_atomic_load): Likewise.
28798 (expand_atomic_store): Likewise.
28799 * ree.c (combine_reaching_defs): Likewise.
28800 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
28801 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
28802 * tree.h (type_has_mode_precision_p): Likewise.
28803 * ubsan.c (instrument_si_overflow): Likewise.
28805 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28806 Alan Hayward <alan.hayward@arm.com>
28807 David Sherwood <david.sherwood@arm.com>
28809 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
28810 polynomial numbers of units.
28811 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
28812 (valid_vector_subparts_p): New function.
28813 (build_vector_type): Remove temporary shim and take the number
28814 of units as a poly_uint64 rather than an int.
28815 (build_opaque_vector_type): Take the number of units as a
28816 poly_uint64 rather than an int.
28817 * tree.c (build_vector_from_ctor): Handle polynomial
28818 TYPE_VECTOR_SUBPARTS.
28819 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
28820 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
28821 (build_vector_from_val): If the number of units is variable,
28822 use build_vec_duplicate_cst for constant operands and
28823 VEC_DUPLICATE_EXPR otherwise.
28824 (make_vector_type): Remove temporary is_constant ().
28825 (build_vector_type, build_opaque_vector_type): Take the number of
28826 units as a poly_uint64 rather than an int.
28827 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
28829 * cfgexpand.c (expand_debug_expr): Likewise.
28830 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
28831 (store_constructor, expand_expr_real_1): Likewise.
28832 (const_scalar_mask_from_tree): Likewise.
28833 * fold-const-call.c (fold_const_reduction): Likewise.
28834 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
28835 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
28836 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
28837 (fold_relational_const): Likewise.
28838 (native_interpret_vector): Likewise. Change the size from an
28839 int to an unsigned int.
28840 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
28841 TYPE_VECTOR_SUBPARTS.
28842 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
28843 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
28844 duplicating a non-constant operand into a variable-length vector.
28845 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
28846 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
28847 * ipa-icf.c (sem_variable::equals): Likewise.
28848 * match.pd: Likewise.
28849 * omp-simd-clone.c (simd_clone_subparts): Likewise.
28850 * print-tree.c (print_node): Likewise.
28851 * stor-layout.c (layout_type): Likewise.
28852 * targhooks.c (default_builtin_vectorization_cost): Likewise.
28853 * tree-cfg.c (verify_gimple_comparison): Likewise.
28854 (verify_gimple_assign_binary): Likewise.
28855 (verify_gimple_assign_ternary): Likewise.
28856 (verify_gimple_assign_single): Likewise.
28857 * tree-pretty-print.c (dump_generic_node): Likewise.
28858 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
28859 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
28860 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
28861 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
28862 (vect_shift_permute_load_chain): Likewise.
28863 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
28864 (expand_vector_condition, optimize_vector_constructor): Likewise.
28865 (lower_vec_perm, get_compute_type): Likewise.
28866 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
28867 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
28868 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
28869 (vect_recog_mask_conversion_pattern): Likewise.
28870 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
28871 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
28872 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
28873 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
28874 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
28875 (vectorizable_shift, vectorizable_operation, vectorizable_store)
28876 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
28877 (supportable_widening_operation): Likewise.
28878 (supportable_narrowing_operation): Likewise.
28879 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
28881 * varasm.c (output_constant): Likewise.
28883 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28884 Alan Hayward <alan.hayward@arm.com>
28885 David Sherwood <david.sherwood@arm.com>
28887 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
28888 so that both the length == 3 and length != 3 cases set up their
28889 own permute vectors. Add comments explaining why we know the
28890 number of elements is constant.
28891 (vect_permute_load_chain): Likewise.
28893 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28894 Alan Hayward <alan.hayward@arm.com>
28895 David Sherwood <david.sherwood@arm.com>
28897 * machmode.h (mode_nunits): Change from unsigned char to
28899 (ONLY_FIXED_SIZE_MODES): New macro.
28900 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
28901 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
28902 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
28904 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
28905 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
28906 or if measurement_type is not polynomial.
28907 * genmodes.c (ZERO_COEFFS): New macro.
28908 (emit_mode_nunits_inline): Make mode_nunits_inline return a
28910 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
28911 Use ZERO_COEFFS when emitting initializers.
28912 * data-streamer.h (bp_pack_poly_value): New function.
28913 (bp_unpack_poly_value): Likewise.
28914 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
28915 for GET_MODE_NUNITS.
28916 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
28917 for GET_MODE_NUNITS.
28918 * tree.c (make_vector_type): Remove temporary shim and make
28919 the real function take the number of units as a poly_uint64
28920 rather than an int.
28921 (build_vector_type_for_mode): Handle polynomial nunits.
28922 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
28923 * emit-rtl.c (const_vec_series_p_1): Likewise.
28924 (gen_rtx_CONST_VECTOR): Likewise.
28925 * fold-const.c (test_vec_duplicate_folding): Likewise.
28926 * genrecog.c (validate_pattern): Likewise.
28927 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
28928 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
28929 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
28930 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
28931 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
28932 * rtlanal.c (subreg_get_info): Likewise.
28933 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
28934 (vect_grouped_load_supported): Likewise.
28935 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
28936 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
28937 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
28938 (simplify_const_unary_operation, simplify_binary_operation_1)
28939 (simplify_const_binary_operation, simplify_ternary_operation)
28940 (test_vector_ops_duplicate, test_vector_ops): Likewise.
28941 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
28942 instead of CONST_VECTOR_NUNITS.
28943 * varasm.c (output_constant_pool_2): Likewise.
28944 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
28945 explicit-encoded elements in the XVEC for variable-length vectors.
28947 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28949 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
28951 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28952 Alan Hayward <alan.hayward@arm.com>
28953 David Sherwood <david.sherwood@arm.com>
28955 * coretypes.h (fixed_size_mode): Declare.
28956 (fixed_size_mode_pod): New typedef.
28957 * builtins.h (target_builtins::x_apply_args_mode)
28958 (target_builtins::x_apply_result_mode): Change type to
28959 fixed_size_mode_pod.
28960 * builtins.c (apply_args_size, apply_result_size, result_vector)
28961 (expand_builtin_apply_args_1, expand_builtin_apply)
28962 (expand_builtin_return): Update accordingly.
28964 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
28966 * cse.c (hash_rtx_cb): Hash only the encoded elements.
28967 * cselib.c (cselib_hash_rtx): Likewise.
28968 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
28969 CONST_VECTOR encoding.
28971 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28972 Jeff Law <law@redhat.com>
28975 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
28976 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
28977 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
28978 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
28981 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
28982 explicitly probe *sp in a noreturn function if there were any callee
28983 register saves or frame pointer is needed.
28985 2018-01-03 Jakub Jelinek <jakub@redhat.com>
28988 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
28989 BLKmode for ternary, binary or unary expressions.
28992 * var-tracking.c (delete_vta_debug_insn): New inline function.
28993 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
28994 insns from get_insns () to NULL instead of each bb separately.
28995 Use delete_vta_debug_insn. No longer static.
28996 (vt_debug_insns_local, variable_tracking_main_1): Adjust
28997 delete_vta_debug_insns callers.
28998 * rtl.h (delete_vta_debug_insns): Declare.
28999 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
29000 instead of variable_tracking_main.
29002 2018-01-03 Martin Sebor <msebor@redhat.com>
29004 PR tree-optimization/83603
29005 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
29006 arguments past the endof the argument list in functions declared
29007 without a prototype.
29008 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
29009 Avoid checking when arguments are null.
29011 2018-01-03 Martin Sebor <msebor@redhat.com>
29014 * doc/extend.texi (attribute const): Fix a typo.
29015 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
29016 issuing -Wsuggest-attribute for void functions.
29018 2018-01-03 Martin Sebor <msebor@redhat.com>
29020 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
29021 offset_int::from instead of wide_int::to_shwi.
29022 (maybe_diag_overlap): Remove assertion.
29023 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
29024 * gimple-ssa-sprintf.c (format_directive): Same.
29025 (parse_directive): Same.
29026 (sprintf_dom_walker::compute_format_length): Same.
29027 (try_substitute_return_value): Same.
29029 2018-01-03 Jeff Law <law@redhat.com>
29031 PR middle-end/83654
29032 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
29033 non-constant residual for zero at runtime and avoid probing in
29034 that case. Reorganize code for trailing problem to mirror handling
29037 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29039 PR tree-optimization/83501
29040 * tree-ssa-strlen.c (get_string_cst): New.
29041 (handle_char_store): Call get_string_cst.
29043 2018-01-03 Martin Liska <mliska@suse.cz>
29045 PR tree-optimization/83593
29046 * tree-ssa-strlen.c: Include tree-cfg.h.
29047 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
29048 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
29049 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
29051 (strlen_dom_walker::before_dom_children): Call
29052 gimple_purge_dead_eh_edges. Dump tranformation with details
29054 (strlen_dom_walker::before_dom_children): Update call by adding
29055 new argument cleanup_eh.
29056 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
29058 2018-01-03 Martin Liska <mliska@suse.cz>
29061 * cif-code.def (VARIADIC_THUNK): New enum value.
29062 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
29065 2018-01-03 Jan Beulich <jbeulich@suse.com>
29067 * sse.md (mov<mode>_internal): Tighten condition for when to use
29068 vmovdqu<ssescalarsize> for TI and OI modes.
29070 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29072 Update copyright years.
29074 2018-01-03 Martin Liska <mliska@suse.cz>
29077 * ipa-visibility.c (function_and_variable_visibility): Skip
29078 functions with noipa attribure.
29080 2018-01-03 Jakub Jelinek <jakub@redhat.com>
29082 * gcc.c (process_command): Update copyright notice dates.
29083 * gcov-dump.c (print_version): Ditto.
29084 * gcov.c (print_version): Ditto.
29085 * gcov-tool.c (print_version): Ditto.
29086 * gengtype.c (create_file): Ditto.
29087 * doc/cpp.texi: Bump @copying's copyright year.
29088 * doc/cppinternals.texi: Ditto.
29089 * doc/gcc.texi: Ditto.
29090 * doc/gccint.texi: Ditto.
29091 * doc/gcov.texi: Ditto.
29092 * doc/install.texi: Ditto.
29093 * doc/invoke.texi: Ditto.
29095 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29097 * vector-builder.h (vector_builder::m_full_nelts): Change from
29098 unsigned int to poly_uint64.
29099 (vector_builder::full_nelts): Update prototype accordingly.
29100 (vector_builder::new_vector): Likewise.
29101 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
29102 (vector_builder::operator ==): Likewise.
29103 (vector_builder::finalize): Likewise.
29104 * int-vector-builder.h (int_vector_builder::int_vector_builder):
29105 Take the number of elements as a poly_uint64 rather than an
29107 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
29108 from unsigned int to poly_uint64.
29109 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
29110 (vec_perm_indices::new_vector): Likewise.
29111 (vec_perm_indices::length): Likewise.
29112 (vec_perm_indices::nelts_per_input): Likewise.
29113 (vec_perm_indices::input_nelts): Likewise.
29114 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
29115 number of elements per input as a poly_uint64 rather than an
29116 unsigned int. Use the original encoding for variable-length
29117 vectors, rather than clamping each individual element.
29118 For the second and subsequent elements in each pattern,
29119 clamp the step and base before clamping their sum.
29120 (vec_perm_indices::series_p): Handle polynomial element counts.
29121 (vec_perm_indices::all_in_range_p): Likewise.
29122 (vec_perm_indices_to_tree): Likewise.
29123 (vec_perm_indices_to_rtx): Likewise.
29124 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
29125 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
29126 (tree_vector_builder::new_binary_operation): Handle polynomial
29127 element counts. Return false if we need to know the number
29128 of elements at compile time.
29129 * fold-const.c (fold_vec_perm): Punt if the number of elements
29130 isn't known at compile time.
29132 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29134 * vec-perm-indices.h (vec_perm_builder): Change element type
29135 from HOST_WIDE_INT to poly_int64.
29136 (vec_perm_indices::element_type): Update accordingly.
29137 (vec_perm_indices::clamp): Handle polynomial element_types.
29138 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
29139 (vec_perm_indices::all_in_range_p): Likewise.
29140 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
29142 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
29143 polynomial vec_perm_indices element types.
29144 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
29145 * fold-const.c (fold_vec_perm): Likewise.
29146 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
29147 * tree-vect-generic.c (lower_vec_perm): Likewise.
29148 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
29149 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
29150 element type to HOST_WIDE_INT.
29152 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29153 Alan Hayward <alan.hayward@arm.com>
29154 David Sherwood <david.sherwood@arm.com>
29156 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
29157 rather than an int. Use plus_constant.
29158 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
29159 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
29161 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29162 Alan Hayward <alan.hayward@arm.com>
29163 David Sherwood <david.sherwood@arm.com>
29165 * calls.c (emit_call_1, expand_call): Change struct_value_size from
29166 a HOST_WIDE_INT to a poly_int64.
29168 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29169 Alan Hayward <alan.hayward@arm.com>
29170 David Sherwood <david.sherwood@arm.com>
29172 * calls.c (load_register_parameters): Cope with polynomial
29173 mode sizes. Require a constant size for BLKmode parameters
29174 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
29175 forces a parameter to be padded at the lsb end in order to
29176 fill a complete number of words, require the parameter size
29177 to be ordered wrt UNITS_PER_WORD.
29179 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29180 Alan Hayward <alan.hayward@arm.com>
29181 David Sherwood <david.sherwood@arm.com>
29183 * reload1.c (spill_stack_slot_width): Change element type
29184 from unsigned int to poly_uint64_pod.
29185 (alter_reg): Treat mode sizes as polynomial.
29187 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29188 Alan Hayward <alan.hayward@arm.com>
29189 David Sherwood <david.sherwood@arm.com>
29191 * reload.c (complex_word_subreg_p): New function.
29192 (reload_inner_reg_of_subreg, push_reload): Use it.
29194 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29195 Alan Hayward <alan.hayward@arm.com>
29196 David Sherwood <david.sherwood@arm.com>
29198 * lra-constraints.c (process_alt_operands): Reject matched
29199 operands whose sizes aren't ordered.
29200 (match_reload): Refer to this check here.
29202 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29203 Alan Hayward <alan.hayward@arm.com>
29204 David Sherwood <david.sherwood@arm.com>
29206 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
29207 that the mode size is in the set {1, 2, 4, 8, 16}.
29209 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29210 Alan Hayward <alan.hayward@arm.com>
29211 David Sherwood <david.sherwood@arm.com>
29213 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
29214 Use plus_constant instead of gen_rtx_PLUS.
29216 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29217 Alan Hayward <alan.hayward@arm.com>
29218 David Sherwood <david.sherwood@arm.com>
29220 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
29221 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
29222 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
29223 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
29224 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
29225 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
29226 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
29227 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
29228 * config/i386/i386.c (ix86_push_rounding): ...this new function.
29229 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
29231 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
29232 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
29233 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
29234 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
29235 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
29236 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
29237 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
29238 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
29239 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
29240 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
29242 * expr.c (emit_move_resolve_push): Treat the input and result
29243 of PUSH_ROUNDING as a poly_int64.
29244 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
29245 (emit_push_insn): Likewise.
29246 * lra-eliminations.c (mark_not_eliminable): Likewise.
29247 * recog.c (push_operand): Likewise.
29248 * reload1.c (elimination_effects): Likewise.
29249 * rtlanal.c (nonzero_bits1): Likewise.
29250 * calls.c (store_one_arg): Likewise. Require the padding to be
29251 known at compile time.
29253 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29254 Alan Hayward <alan.hayward@arm.com>
29255 David Sherwood <david.sherwood@arm.com>
29257 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
29258 Use plus_constant instead of gen_rtx_PLUS.
29260 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29261 Alan Hayward <alan.hayward@arm.com>
29262 David Sherwood <david.sherwood@arm.com>
29264 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
29265 rather than an int.
29267 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29268 Alan Hayward <alan.hayward@arm.com>
29269 David Sherwood <david.sherwood@arm.com>
29271 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
29272 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
29273 via stack temporaries. Treat the mode size as polynomial too.
29275 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29276 Alan Hayward <alan.hayward@arm.com>
29277 David Sherwood <david.sherwood@arm.com>
29279 * expr.c (expand_expr_real_2): When handling conversions involving
29280 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
29281 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
29282 as a poly_uint64 too.
29284 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29285 Alan Hayward <alan.hayward@arm.com>
29286 David Sherwood <david.sherwood@arm.com>
29288 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
29290 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29291 Alan Hayward <alan.hayward@arm.com>
29292 David Sherwood <david.sherwood@arm.com>
29294 * combine.c (can_change_dest_mode): Handle polynomial
29295 REGMODE_NATURAL_SIZE.
29296 * expmed.c (store_bit_field_1): Likewise.
29297 * expr.c (store_constructor): Likewise.
29298 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
29299 and polynomial REGMODE_NATURAL_SIZE.
29300 (gen_lowpart_common): Likewise.
29301 * reginfo.c (record_subregs_of_mode): Likewise.
29302 * rtlanal.c (read_modify_subreg_p): Likewise.
29304 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29305 Alan Hayward <alan.hayward@arm.com>
29306 David Sherwood <david.sherwood@arm.com>
29308 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
29309 numbers of elements.
29311 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29312 Alan Hayward <alan.hayward@arm.com>
29313 David Sherwood <david.sherwood@arm.com>
29315 * match.pd: Cope with polynomial numbers of vector elements.
29317 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29318 Alan Hayward <alan.hayward@arm.com>
29319 David Sherwood <david.sherwood@arm.com>
29321 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
29322 in a POINTER_PLUS_EXPR.
29324 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29325 Alan Hayward <alan.hayward@arm.com>
29326 David Sherwood <david.sherwood@arm.com>
29328 * omp-simd-clone.c (simd_clone_subparts): New function.
29329 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
29330 (ipa_simd_modify_function_body): Likewise.
29332 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29333 Alan Hayward <alan.hayward@arm.com>
29334 David Sherwood <david.sherwood@arm.com>
29336 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
29337 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
29338 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
29339 (expand_vector_condition, vector_element): Likewise.
29340 (subparts_gt): New function.
29341 (get_compute_type): Use subparts_gt.
29342 (count_type_subparts): Delete.
29343 (expand_vector_operations_1): Use subparts_gt instead of
29344 count_type_subparts.
29346 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29347 Alan Hayward <alan.hayward@arm.com>
29348 David Sherwood <david.sherwood@arm.com>
29350 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
29351 (vect_compile_time_alias): ...this new function. Do the calculation
29352 on poly_ints rather than trees.
29353 (vect_prune_runtime_alias_test_list): Update call accordingly.
29355 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29356 Alan Hayward <alan.hayward@arm.com>
29357 David Sherwood <david.sherwood@arm.com>
29359 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
29361 (vect_schedule_slp_instance): Likewise.
29363 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29364 Alan Hayward <alan.hayward@arm.com>
29365 David Sherwood <david.sherwood@arm.com>
29367 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
29368 constant and extern definitions for variable-length vectors.
29369 (vect_get_constant_vectors): Note that the number of units
29370 is known to be constant.
29372 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29373 Alan Hayward <alan.hayward@arm.com>
29374 David Sherwood <david.sherwood@arm.com>
29376 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
29377 of units as polynomial. Choose between WIDE and NARROW based
29380 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29381 Alan Hayward <alan.hayward@arm.com>
29382 David Sherwood <david.sherwood@arm.com>
29384 * tree-vect-stmts.c (simd_clone_subparts): New function.
29385 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
29387 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29388 Alan Hayward <alan.hayward@arm.com>
29389 David Sherwood <david.sherwood@arm.com>
29391 * tree-vect-stmts.c (vectorizable_call): Treat the number of
29392 vectors as polynomial. Use build_index_vector for
29393 IFN_GOMP_SIMD_LANE.
29395 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29396 Alan Hayward <alan.hayward@arm.com>
29397 David Sherwood <david.sherwood@arm.com>
29399 * tree-vect-stmts.c (get_load_store_type): Treat the number of
29400 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
29401 for variable-length vectors.
29402 (vectorizable_mask_load_store): Treat the number of units as
29403 polynomial, asserting that it is constant if the condition has
29404 already been enforced.
29405 (vectorizable_store, vectorizable_load): Likewise.
29407 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29408 Alan Hayward <alan.hayward@arm.com>
29409 David Sherwood <david.sherwood@arm.com>
29411 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
29412 of units as polynomial. Punt if we can't tell at compile time
29413 which vector contains the final result.
29415 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29416 Alan Hayward <alan.hayward@arm.com>
29417 David Sherwood <david.sherwood@arm.com>
29419 * tree-vect-loop.c (vectorizable_induction): Treat the number
29420 of units as polynomial. Punt on SLP inductions. Use an integer
29421 VEC_SERIES_EXPR for variable-length integer reductions. Use a
29422 cast of such a series for variable-length floating-point
29425 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29426 Alan Hayward <alan.hayward@arm.com>
29427 David Sherwood <david.sherwood@arm.com>
29429 * tree.h (build_index_vector): Declare.
29430 * tree.c (build_index_vector): New function.
29431 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
29432 of units as polynomial, forcibly converting it to a constant if
29433 vectorizable_reduction has already enforced the condition.
29434 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
29435 to create a {1,2,3,...} vector.
29436 (vectorizable_reduction): Treat the number of units as polynomial.
29437 Choose vectype_in based on the largest scalar element size rather
29438 than the smallest number of units. Enforce the restrictions
29441 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29442 Alan Hayward <alan.hayward@arm.com>
29443 David Sherwood <david.sherwood@arm.com>
29445 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
29446 number of units as polynomial.
29448 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29449 Alan Hayward <alan.hayward@arm.com>
29450 David Sherwood <david.sherwood@arm.com>
29452 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
29453 * target.def (autovectorize_vector_sizes): Return the vector sizes
29454 by pointer, using vector_sizes rather than a bitmask.
29455 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
29456 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
29457 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
29459 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
29460 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
29461 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
29462 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
29463 * omp-general.c (omp_max_vf): Likewise.
29464 * omp-low.c (omp_clause_aligned_alignment): Likewise.
29465 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
29466 * tree-vect-loop.c (vect_analyze_loop): Likewise.
29467 * tree-vect-slp.c (vect_slp_bb): Likewise.
29468 * doc/tm.texi: Regenerate.
29469 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
29471 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
29472 the vector size as a poly_uint64 rather than an unsigned int.
29473 (current_vector_size): Change from an unsigned int to a poly_uint64.
29474 (get_vectype_for_scalar_type): Update accordingly.
29475 * tree.h (build_truth_vector_type): Take the size and number of
29476 units as a poly_uint64 rather than an unsigned int.
29477 (build_vector_type): Add a temporary overload that takes
29478 the number of units as a poly_uint64 rather than an unsigned int.
29479 * tree.c (make_vector_type): Likewise.
29480 (build_truth_vector_type): Take the number of units as a poly_uint64
29481 rather than an unsigned int.
29483 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29484 Alan Hayward <alan.hayward@arm.com>
29485 David Sherwood <david.sherwood@arm.com>
29487 * target.def (get_mask_mode): Take the number of units and length
29488 as poly_uint64s rather than unsigned ints.
29489 * targhooks.h (default_get_mask_mode): Update accordingly.
29490 * targhooks.c (default_get_mask_mode): Likewise.
29491 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
29492 * doc/tm.texi: Regenerate.
29494 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29495 Alan Hayward <alan.hayward@arm.com>
29496 David Sherwood <david.sherwood@arm.com>
29498 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
29499 * omp-general.c (omp_max_vf): Likewise.
29500 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
29501 (expand_omp_simd): Handle polynomial safelen.
29502 * omp-low.c (omplow_simd_context): Add a default constructor.
29503 (omplow_simd_context::max_vf): Change from int to poly_uint64.
29504 (lower_rec_simd_input_clauses): Update accordingly.
29505 (lower_rec_input_clauses): Likewise.
29507 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29508 Alan Hayward <alan.hayward@arm.com>
29509 David Sherwood <david.sherwood@arm.com>
29511 * tree-vectorizer.h (vect_nunits_for_cost): New function.
29512 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
29513 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
29514 (vect_analyze_slp_cost): Likewise.
29515 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
29516 (vect_model_load_cost): Likewise.
29518 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29519 Alan Hayward <alan.hayward@arm.com>
29520 David Sherwood <david.sherwood@arm.com>
29522 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
29523 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
29524 from an unsigned int * to a poly_uint64_pod *.
29525 (calculate_unrolling_factor): New function.
29526 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
29528 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29529 Alan Hayward <alan.hayward@arm.com>
29530 David Sherwood <david.sherwood@arm.com>
29532 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
29533 from an unsigned int to a poly_uint64.
29534 (_loop_vec_info::slp_unrolling_factor): Likewise.
29535 (_loop_vec_info::vectorization_factor): Change from an int
29537 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
29538 (vect_get_num_vectors): New function.
29539 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
29540 (vect_get_num_copies): Use vect_get_num_vectors.
29541 (vect_analyze_data_ref_dependences): Change max_vf from an int *
29542 to an unsigned int *.
29543 (vect_analyze_data_refs): Change min_vf from an int * to a
29545 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29546 than an unsigned HOST_WIDE_INT.
29547 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
29548 (vect_analyze_data_ref_dependence): Change max_vf from an int *
29549 to an unsigned int *.
29550 (vect_analyze_data_ref_dependences): Likewise.
29551 (vect_compute_data_ref_alignment): Handle polynomial vf.
29552 (vect_enhance_data_refs_alignment): Likewise.
29553 (vect_prune_runtime_alias_test_list): Likewise.
29554 (vect_shift_permute_load_chain): Likewise.
29555 (vect_supportable_dr_alignment): Likewise.
29556 (dependence_distance_ge_vf): Take the vectorization factor as a
29557 poly_uint64 rather than an unsigned HOST_WIDE_INT.
29558 (vect_analyze_data_refs): Change min_vf from an int * to a
29560 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
29561 vfm1 as a poly_uint64 rather than an int. Make the same change
29562 for the returned bound_scalar.
29563 (vect_gen_vector_loop_niters): Handle polynomial vf.
29564 (vect_do_peeling): Likewise. Update call to
29565 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
29566 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
29568 * tree-vect-loop.c (vect_determine_vectorization_factor)
29569 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
29570 (vect_get_known_peeling_cost): Likewise.
29571 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
29572 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
29573 (vect_transform_loop): Likewise. Use the lowest possible VF when
29574 updating the upper bounds of the loop.
29575 (vect_min_worthwhile_factor): Make static. Return an unsigned int
29576 rather than an int.
29577 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
29578 polynomial unroll factors.
29579 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
29580 (vect_make_slp_decision): Likewise.
29581 (vect_supported_load_permutation_p): Likewise, and polynomial
29583 (vect_analyze_slp_cost): Handle polynomial vf.
29584 (vect_slp_analyze_node_operations): Likewise.
29585 (vect_slp_analyze_bb_1): Likewise.
29586 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
29587 than an unsigned HOST_WIDE_INT.
29588 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
29589 (vectorizable_load): Handle polynomial vf.
29590 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
29592 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
29594 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29595 Alan Hayward <alan.hayward@arm.com>
29596 David Sherwood <david.sherwood@arm.com>
29598 * match.pd: Handle bit operations involving three constants
29599 and try to fold one pair.
29601 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29603 * tree-vect-loop-manip.c: Include gimple-fold.h.
29604 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
29605 niters_maybe_zero parameters. Handle other cases besides a step of 1.
29606 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
29607 Add a path that uses a step of VF instead of 1, but disable it
29609 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
29610 and niters_no_overflow parameters. Update calls to
29611 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
29612 Create a new SSA name if the latter choses to use a ste other
29613 than zero, and return it via niters_vector_mult_vf_var.
29614 * tree-vect-loop.c (vect_transform_loop): Update calls to
29615 vect_do_peeling, vect_gen_vector_loop_niters and
29616 slpeel_make_loop_iterate_ntimes.
29617 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
29618 (vect_gen_vector_loop_niters): Update declarations after above changes.
29620 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
29622 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
29623 128-bit round to integer instructions.
29624 (ceil<mode>2): Likewise.
29625 (btrunc<mode>2): Likewise.
29626 (round<mode>2): Likewise.
29628 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29630 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
29631 unaligned VSX load/store on P8/P9.
29632 (expand_block_clear): Allow the use of unaligned VSX
29633 load/store on P8/P9.
29635 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29637 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
29639 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
29640 swap associated with both a load and a store.
29642 2018-01-02 Andrew Waterman <andrew@sifive.com>
29644 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
29645 * config/riscv/riscv.md (clear_cache): Use it.
29647 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
29649 * web.c: Remove out-of-date comment.
29651 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29653 * expr.c (fixup_args_size_notes): Check that any existing
29654 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
29655 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
29656 (emit_single_push_insn): ...here.
29658 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29660 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
29661 (const_vector_encoded_nelts): New function.
29662 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
29663 (const_vector_int_elt, const_vector_elt): Declare.
29664 * emit-rtl.c (const_vector_int_elt_1): New function.
29665 (const_vector_elt): Likewise.
29666 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
29667 of CONST_VECTOR_ELT.
29669 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29671 * expr.c: Include rtx-vector-builder.h.
29672 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
29673 directly on the tree encoding.
29674 (const_vector_from_tree): Likewise.
29675 * optabs.c: Include rtx-vector-builder.h.
29676 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
29677 sequence of "u" values.
29678 * vec-perm-indices.c: Include rtx-vector-builder.h.
29679 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
29680 directly on the vec_perm_indices encoding.
29682 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29684 * doc/rtl.texi (const_vector): Describe new encoding scheme.
29685 * Makefile.in (OBJS): Add rtx-vector-builder.o.
29686 * rtx-vector-builder.h: New file.
29687 * rtx-vector-builder.c: Likewise.
29688 * rtl.h (rtx_def::u2): Add a const_vector field.
29689 (CONST_VECTOR_NPATTERNS): New macro.
29690 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
29691 (CONST_VECTOR_DUPLICATE_P): Likewise.
29692 (CONST_VECTOR_STEPPED_P): Likewise.
29693 (CONST_VECTOR_ENCODED_ELT): Likewise.
29694 (const_vec_duplicate_p): Check for a duplicated vector encoding.
29695 (unwrap_const_vec_duplicate): Likewise.
29696 (const_vec_series_p): Check for a non-duplicated vector encoding.
29697 Say that the function only returns true for integer vectors.
29698 * emit-rtl.c: Include rtx-vector-builder.h.
29699 (gen_const_vec_duplicate_1): Delete.
29700 (gen_const_vector): Call gen_const_vec_duplicate instead of
29701 gen_const_vec_duplicate_1.
29702 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
29703 (gen_const_vec_duplicate): Use rtx_vector_builder.
29704 (gen_const_vec_series): Likewise.
29705 (gen_rtx_CONST_VECTOR): Likewise.
29706 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
29707 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29708 Build a new vector rather than modifying a CONST_VECTOR in-place.
29709 (handle_special_swappables): Update call accordingly.
29710 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
29711 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
29712 Build a new vector rather than modifying a CONST_VECTOR in-place.
29713 (handle_special_swappables): Update call accordingly.
29715 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29717 * simplify-rtx.c (simplify_const_binary_operation): Use
29718 CONST_VECTOR_ELT instead of XVECEXP.
29720 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29722 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
29723 the selector elements to be different from the data elements
29724 if the selector is a VECTOR_CST.
29725 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
29726 ssizetype for the selector.
29728 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29730 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
29731 before testing each element individually.
29732 * tree-vect-generic.c (lower_vec_perm): Likewise.
29734 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29736 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
29737 * selftest-run-tests.c (selftest::run_tests): Call it.
29738 * vector-builder.h (vector_builder::operator ==): New function.
29739 (vector_builder::operator !=): Likewise.
29740 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
29741 (vec_perm_indices::all_from_input_p): New function.
29742 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
29743 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
29744 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
29745 instead of reading the VECTOR_CST directly. Detect whether both
29746 vector inputs are the same before constructing the vec_perm_indices,
29747 and update the number of inputs argument accordingly. Use the
29748 utility functions added above. Only construct sel2 if we need to.
29750 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29752 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
29753 the broadcast of the low byte.
29754 (expand_mult_highpart): Use an explicit encoding for the permutes.
29755 * optabs-query.c (can_mult_highpart_p): Likewise.
29756 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
29757 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29758 (vectorizable_bswap): Likewise.
29759 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
29760 explicit encoding for the power-of-2 permutes.
29761 (vect_permute_store_chain): Likewise.
29762 (vect_grouped_load_supported): Likewise.
29763 (vect_permute_load_chain): Likewise.
29765 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29767 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
29768 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
29769 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
29770 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
29771 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
29772 (vect_gen_perm_mask_any): Likewise.
29774 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29776 * int-vector-builder.h: New file.
29777 * vec-perm-indices.h: Include int-vector-builder.h.
29778 (vec_perm_indices): Redefine as an int_vector_builder.
29779 (auto_vec_perm_indices): Delete.
29780 (vec_perm_builder): Redefine as a stand-alone class.
29781 (vec_perm_indices::vec_perm_indices): New function.
29782 (vec_perm_indices::clamp): Likewise.
29783 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
29784 (vec_perm_indices::new_vector): New function.
29785 (vec_perm_indices::new_expanded_vector): Update for new
29786 vec_perm_indices class.
29787 (vec_perm_indices::rotate_inputs): New function.
29788 (vec_perm_indices::all_in_range_p): Operate directly on the
29789 encoded form, without computing elided elements.
29790 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
29791 encoding. Update for new vec_perm_indices class.
29792 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
29793 the given vec_perm_builder.
29794 (expand_vec_perm_var): Update vec_perm_builder constructor.
29795 (expand_mult_highpart): Use vec_perm_builder instead of
29796 auto_vec_perm_indices.
29797 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
29798 vec_perm_indices instead of auto_vec_perm_indices. Use a single
29799 or double series encoding as appropriate.
29800 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
29801 vec_perm_indices instead of auto_vec_perm_indices.
29802 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29803 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
29804 (vect_permute_store_chain): Likewise.
29805 (vect_grouped_load_supported): Likewise.
29806 (vect_permute_load_chain): Likewise.
29807 (vect_shift_permute_load_chain): Likewise.
29808 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
29809 (vect_transform_slp_perm_load): Likewise.
29810 (vect_schedule_slp_instance): Likewise.
29811 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
29812 (vectorizable_mask_load_store): Likewise.
29813 (vectorizable_bswap): Likewise.
29814 (vectorizable_store): Likewise.
29815 (vectorizable_load): Likewise.
29816 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
29817 vec_perm_indices instead of auto_vec_perm_indices. Use
29818 tree_to_vec_perm_builder to read the vector from a tree.
29819 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
29820 vec_perm_builder instead of a vec_perm_indices.
29821 (have_whole_vector_shift): Use vec_perm_builder and
29822 vec_perm_indices instead of auto_vec_perm_indices. Leave the
29823 truncation to calc_vec_perm_mask_for_shift.
29824 (vect_create_epilog_for_reduction): Likewise.
29825 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
29826 from auto_vec_perm_indices to vec_perm_indices.
29827 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29828 instead of changing individual elements.
29829 (aarch64_vectorize_vec_perm_const): Use new_vector to install
29830 the vector in d.perm.
29831 * config/arm/arm.c (expand_vec_perm_d::perm): Change
29832 from auto_vec_perm_indices to vec_perm_indices.
29833 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
29834 instead of changing individual elements.
29835 (arm_vectorize_vec_perm_const): Use new_vector to install
29836 the vector in d.perm.
29837 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
29838 Update vec_perm_builder constructor.
29839 (rs6000_expand_interleave): Likewise.
29840 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
29841 (rs6000_expand_interleave): Likewise.
29843 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29845 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
29846 to qimode could truncate the indices.
29847 * optabs.c (expand_vec_perm_var): Likewise.
29849 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29851 * Makefile.in (OBJS): Add vec-perm-indices.o.
29852 * vec-perm-indices.h: New file.
29853 * vec-perm-indices.c: Likewise.
29854 * target.h (vec_perm_indices): Replace with a forward class
29856 (auto_vec_perm_indices): Move to vec-perm-indices.h.
29857 * optabs.h: Include vec-perm-indices.h.
29858 (expand_vec_perm): Delete.
29859 (selector_fits_mode_p, expand_vec_perm_var): Declare.
29860 (expand_vec_perm_const): Declare.
29861 * target.def (vec_perm_const_ok): Replace with...
29862 (vec_perm_const): ...this new hook.
29863 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
29864 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
29865 * doc/tm.texi: Regenerate.
29866 * optabs.def (vec_perm_const): Delete.
29867 * doc/md.texi (vec_perm_const): Likewise.
29868 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
29869 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
29870 expand_vec_perm for constant permutation vectors. Assert that
29871 the mode of variable permutation vectors is the integer equivalent
29872 of the mode that is being permuted.
29873 * optabs-query.h (selector_fits_mode_p): Declare.
29874 * optabs-query.c: Include vec-perm-indices.h.
29875 (selector_fits_mode_p): New function.
29876 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
29877 is defined, instead of checking whether the vec_perm_const_optab
29878 exists. Use targetm.vectorize.vec_perm_const instead of
29879 targetm.vectorize.vec_perm_const_ok. Check whether the indices
29880 fit in the vector mode before using a variable permute.
29881 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
29882 vec_perm_indices instead of an rtx.
29883 (expand_vec_perm): Replace with...
29884 (expand_vec_perm_const): ...this new function. Take the selector
29885 as a vec_perm_indices rather than an rtx. Also take the mode of
29886 the selector. Update call to shift_amt_for_vec_perm_mask.
29887 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
29888 Use vec_perm_indices::new_expanded_vector to expand the original
29889 selector into bytes. Check whether the indices fit in the vector
29890 mode before using a variable permute.
29891 (expand_vec_perm_var): Make global.
29892 (expand_mult_highpart): Use expand_vec_perm_const.
29893 * fold-const.c: Includes vec-perm-indices.h.
29894 * tree-ssa-forwprop.c: Likewise.
29895 * tree-vect-data-refs.c: Likewise.
29896 * tree-vect-generic.c: Likewise.
29897 * tree-vect-loop.c: Likewise.
29898 * tree-vect-slp.c: Likewise.
29899 * tree-vect-stmts.c: Likewise.
29900 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
29902 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
29903 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
29904 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
29905 (aarch64_vectorize_vec_perm_const): ...this new function.
29906 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29907 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29908 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
29909 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
29910 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29911 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29912 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
29914 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
29915 check for NEON modes.
29916 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
29917 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
29918 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
29919 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
29921 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
29922 the old VEC_PERM_CONST conditions.
29923 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
29924 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
29925 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
29926 (ia64_vectorize_vec_perm_const_ok): Merge into...
29927 (ia64_vectorize_vec_perm_const): ...this new function.
29928 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
29929 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
29930 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
29931 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
29932 * config/mips/mips.c (mips_expand_vec_perm_const)
29933 (mips_vectorize_vec_perm_const_ok): Merge into...
29934 (mips_vectorize_vec_perm_const): ...this new function.
29935 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
29936 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
29937 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
29938 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
29939 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
29940 (rs6000_expand_vec_perm_const): Delete.
29941 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
29943 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29944 (altivec_expand_vec_perm_const_le): Take each operand individually.
29945 Operate on constant selectors rather than rtxes.
29946 (altivec_expand_vec_perm_const): Likewise. Update call to
29947 altivec_expand_vec_perm_const_le.
29948 (rs6000_expand_vec_perm_const): Delete.
29949 (rs6000_vectorize_vec_perm_const_ok): Delete.
29950 (rs6000_vectorize_vec_perm_const): New function.
29951 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
29952 an element count and rtx array.
29953 (rs6000_expand_extract_even): Update call accordingly.
29954 (rs6000_expand_interleave): Likewise.
29955 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
29956 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
29957 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
29958 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
29959 (rs6000_expand_vec_perm_const): Delete.
29960 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
29961 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29962 (altivec_expand_vec_perm_const_le): Take each operand individually.
29963 Operate on constant selectors rather than rtxes.
29964 (altivec_expand_vec_perm_const): Likewise. Update call to
29965 altivec_expand_vec_perm_const_le.
29966 (rs6000_expand_vec_perm_const): Delete.
29967 (rs6000_vectorize_vec_perm_const_ok): Delete.
29968 (rs6000_vectorize_vec_perm_const): New function. Remove stray
29969 reference to the SPE evmerge intructions.
29970 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
29971 an element count and rtx array.
29972 (rs6000_expand_extract_even): Update call accordingly.
29973 (rs6000_expand_interleave): Likewise.
29974 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
29975 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
29977 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
29979 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29981 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
29982 vector mode and that that mode matches the mode of the data
29984 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
29985 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
29986 directly using expand_vec_perm_1 when forcing selectors into
29988 (expand_vec_perm_var): New function, split out from expand_vec_perm.
29990 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
29992 * optabs-query.h (can_vec_perm_p): Delete.
29993 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
29994 * optabs-query.c (can_vec_perm_p): Split into...
29995 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
29996 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
29997 particular selector is valid.
29998 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
29999 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
30000 (vect_grouped_load_supported): Likewise.
30001 (vect_shift_permute_load_chain): Likewise.
30002 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
30003 (vect_transform_slp_perm_load): Likewise.
30004 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
30005 (vectorizable_bswap): Likewise.
30006 (vect_gen_perm_mask_checked): Likewise.
30007 * fold-const.c (fold_ternary_loc): Likewise. Don't take
30008 implementations of variable permutation vectors into account
30009 when deciding which selector to use.
30010 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
30011 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
30012 with a false third argument.
30013 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
30014 to test whether the constant selector is valid and can_vec_perm_var_p
30015 to test whether a variable selector is valid.
30017 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30019 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
30020 * optabs-query.c (can_vec_perm_p): Likewise.
30021 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
30022 instead of vec_perm_indices.
30023 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
30024 (vect_gen_perm_mask_checked): Likewise,
30025 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
30026 (vect_gen_perm_mask_checked): Likewise,
30028 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
30030 * optabs-query.h (qimode_for_vec_perm): Declare.
30031 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
30032 (qimode_for_vec_perm): ...this new function.
30033 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
30035 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30037 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
30038 does not have a conditional at the top.
30040 2018-01-02 Richard Biener <rguenther@suse.de>
30042 * ipa-inline.c (big_speedup_p): Fix expression.
30044 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
30047 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
30050 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
30054 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
30055 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
30056 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
30057 cond_taken_branch_cost 3->4.
30059 2018-01-01 Jakub Jelinek <jakub@redhat.com>
30061 PR tree-optimization/83581
30062 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
30063 TODO_cleanup_cfg if any changes have been made.
30065 PR middle-end/83608
30066 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
30067 convert_modes if target mode has the right side, but different mode
30070 PR middle-end/83609
30071 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
30072 last argument when extracting from CONCAT. If either from_real or
30073 from_imag is NULL, use expansion through memory. If result is not
30074 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
30075 the parts directly to inner mode, if even that fails, use expansion
30078 PR middle-end/83623
30079 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
30080 check for bswap in mode rather than HImode and use that in expand_unop
30083 Copyright (C) 2018 Free Software Foundation, Inc.
30085 Copying and distribution of this file, with or without modification,
30086 are permitted in any medium without royalty provided the copyright
30087 notice and this notice are preserved.